Data ONTAP 8.2 High Availability and MetroCluster

Data ONTAP 8.2 High Availability and MetroCluster
Data ONTAP® 8.2
High Availability and MetroCluster Configuration Guide
For 7-Mode
NetApp, Inc.
495 East Java Drive
Sunnyvale, CA 94089
U.S.
Telephone: +1(408) 822-6000
Fax: +1(408) 822-4501
Support telephone: +1 (888) 463-8277
Web: www.netapp.com
Feedback: doccomments@netapp.com
Part number: 215-07985_A0
May 2013
Table of Contents | 3
Contents
Understanding HA pairs ............................................................................ 11
What an HA pair is .................................................................................................... 11
How HA pairs support nondisruptive operations and fault tolerance ....................... 11
Where to find procedures for nondisruptive operations with HA pairs ........ 12
How the HA pair improves fault tolerance ................................................... 13
Standard HA pair diagram ......................................................................................... 16
Comparison of HA pair types .................................................................................... 17
Understanding mirrored HA pairs ........................................................... 19
Advantages of mirrored HA pairs ............................................................................. 19
Asymmetrically mirrored HA pairs ........................................................................... 19
Understanding takeover and giveback ..................................................... 20
When takeovers occur ............................................................................................... 20
Failover event cause-and-effect table ............................................................ 21
How hardware-assisted takeover speeds up takeover ............................................... 28
What happens during takeover .................................................................................. 29
What happens during giveback ................................................................................. 29
Background disk firmware update and takeover and giveback ................................. 30
Planning your HA pair configuration ...................................................... 31
Best practices for HA pairs ....................................................................................... 31
Setup requirements and restrictions for standard HA pairs ....................................... 32
Setup requirements and restrictions for mirrored HA pairs ...................................... 33
Multipath HA requirement ........................................................................................ 34
What multipath HA for HA pairs is .............................................................. 34
How the connection types are used ............................................................... 35
Advantages of multipath HA ......................................................................... 35
Requirements for hardware-assisted takeover ........................................................... 36
Possible storage configurations in the HA pairs ....................................................... 36
HA pairs and storage system model types ................................................................ 36
Single-chassis and dual-chassis HA pairs ..................................................... 37
Interconnect cabling for systems with variable HA configurations .............. 37
HA configuration and the HA state PROM value ......................................... 38
Table of storage system models and HA configuration differences ............. 38
4 | High Availability and MetroCluster Configuration Guide
Installing and cabling an HA pair ............................................................. 41
System cabinet or equipment rack installation .......................................................... 41
HA pairs in an equipment rack ...................................................................... 41
HA pairs in a system cabinet ......................................................................... 41
Required documentation ........................................................................................... 42
Required tools ........................................................................................................... 43
Required equipment .................................................................................................. 43
Preparing your equipment ......................................................................................... 45
Installing the nodes in equipment racks ........................................................ 45
Installing the nodes in a system cabinet ........................................................ 45
Cabling a standard HA pair ....................................................................................... 46
Determining which Fibre Channel ports to use for Fibre Channel disk
shelf connections ..................................................................................... 46
Cabling Node A to DS14mk2 or DS14mk4 disk shelves ............................. 47
Cabling Node B to DS14mk2 or DS14mk4 disk shelves .............................. 49
Cabling the HA interconnect (all systems except 32xx) ............................... 51
Cabling the HA interconnect (32xx systems in separate chassis) ................. 52
Cabling a mirrored HA pair ...................................................................................... 52
Determining which Fibre Channel ports to use for Fibre Channel disk
shelf connections ..................................................................................... 53
Creating your port list for mirrored HA pairs ............................................... 54
Cabling the Channel A DS14mk2 disk shelf loops ....................................... 55
Cabling the Channel B DS14mk2 or DS14mk4 disk shelf loops ................. 57
Cabling the redundant multipath HA connection for each loop .................... 59
Cabling the HA interconnect (all systems except 32xx) ............................... 61
Cabling the HA interconnect (32xx systems in separate chassis) ................. 61
Required connections for using uninterruptible power supplies with standard or
mirrored HA pairs ................................................................................................ 62
Reconfiguring an HA pair into two stand-alone systems ....................... 63
Ensuring uniform disk ownership within disk shelves and loops in the system ....... 63
Disabling controller failover ..................................................................................... 64
Configuring a node for non-HA (stand-alone) use .................................................... 65
Reconfiguring nodes using disk shelves for stand-alone operation .......................... 67
Requirements when changing a node using array LUNs to stand-alone ................... 69
Reconfiguring nodes using array LUNs for stand-alone operation ........................... 70
Configuring an HA pair ............................................................................. 72
Table of Contents | 5
Bringing up the HA pair ............................................................................................ 72
Considerations for HA pair setup .................................................................. 72
Configuring shared interfaces with setup ...................................................... 73
Configuring dedicated interfaces with setup ................................................. 74
Configuring standby interfaces with setup .................................................... 74
Enabling HA mode capability and controller failover .............................................. 75
Setting options and parameters ................................................................................. 75
Option types for HA pairs ............................................................................. 76
Setting matching node options ...................................................................... 76
Parameters that must be the same on each node ........................................... 77
Best practices for cf options .......................................................................... 77
Disabling the change_fsid option in MetroCluster configurations ............... 79
Verifying and setting the HA state on controller modules and chassis ......... 80
Configuring hardware-assisted takeover ....................................................... 81
Configuring network interfaces for HA pairs ........................................................... 83
Understanding interfaces in an HA pair ........................................................ 83
Making nondisruptive changes to the interface groups ................................. 86
Configuring network interfaces for the HA pair ........................................... 87
Configuring a partner interface in an HA pair .............................................. 88
Configuring partner addresses on different subnets (MetroCluster
configurations only) ................................................................................. 89
Verifying the HA pair cabling and configuration ..................................................... 93
Testing takeover and giveback .................................................................................. 93
Managing takeover and giveback ............................................................. 95
Monitoring an HA pair in normal mode .................................................................... 95
Monitoring HA pair status ............................................................................. 95
Description of HA pair status messages ........................................................ 95
Monitoring the hardware-assisted takeover feature ...................................... 97
Displaying the partner's name ....................................................................... 99
Displaying disk and array LUN information on an HA pair ......................... 99
What takeover and giveback are ............................................................................. 100
When takeovers occur ................................................................................. 100
What happens during takeover .................................................................... 101
What happens after takeover ....................................................................... 101
What happens during giveback ................................................................... 101
Configuring automatic takeover .............................................................................. 102
6 | High Availability and MetroCluster Configuration Guide
Reasons for automatic takeover .................................................................. 102
Commands for performing a manual takeover ............................................ 104
Halting a node without takeover ................................................................. 105
Rebooting a node without takeover ............................................................. 106
Enabling and disabling takeover ................................................................. 106
Enabling and disabling takeover on reboot ................................................. 106
Enabling and disabling automatic takeover of a panicked partner .............. 107
Specifying the time period before takeover ................................................ 107
Enabling or disabling negotiated failover for a network interface .............. 108
Takeover of vFiler units and the vFiler limit .............................................. 109
Managing an HA pair in takeover mode ................................................................. 109
Determining why takeover occurred ........................................................... 109
Statistics in takeover mode .......................................................................... 109
Managing emulated nodes ....................................................................................... 110
Management exceptions for emulated nodes .............................................. 110
Accessing the emulated node from the takeover node ................................ 110
Using Remote Shell ..................................................................................... 112
Emulated node command exceptions .......................................................... 112
Performing dumps and restores for a failed node ................................................... 114
Giveback operations ................................................................................................ 115
Performing a manual giveback .................................................................... 115
Configuring giveback .................................................................................. 118
Configuring automatic giveback ................................................................. 118
Troubleshooting HA issues ..................................................................................... 120
Managing DS14mk2 or DS14mk4 disk shelves in an HA pair ............. 122
Adding DS14mk2 or DS14mk4 disk shelves to a multipath HA loop .................... 122
Upgrading or replacing modules in an HA pair ...................................................... 123
About the disk shelf modules .................................................................................. 124
Restrictions for changing module types .................................................................. 124
Best practices for changing module types ............................................................... 124
Testing the modules ................................................................................................ 125
Determining path status for your HA pair ............................................................... 125
Hot-swapping a module .......................................................................................... 127
Performing nondisruptive shelf replacement in a MetroCluster configuration ....... 128
Preparing for nondisruptive shelf replacement ........................................... 129
Replacing the disk shelf nondisruptively .................................................... 130
Table of Contents | 7
Verifying the disks after the shelf replacement ........................................... 131
Where to find procedures for nondisruptive operations with HA
pairs ...................................................................................................... 133
Overview of a MetroCluster configuration ............................................ 134
Types of MetroCluster configurations .................................................................... 134
How mirroring works in a MetroCluster configuration .......................................... 134
How Data ONTAP works with MetroCluster configurations ................................. 135
Advantages of stretch MetroCluster configurations ................................................ 135
Configuration variations for stretch MetroCluster configurations .......................... 136
Advantages of fabric-attached MetroCluster configurations .................................. 136
Configuration variations for fabric-attached MetroCluster configurations ............. 137
Implementing stretch MetroCluster configurations with disks ........... 138
Planning a stretch MetroCluster configuration using disks .................................... 138
Setup requirements and restrictions for stretch MetroCluster
configurations with disks ....................................................................... 138
Hardware requirements for using disks in a MetroCluster configuration
with V-Series systems ........................................................................... 139
Required connections for using uninterruptible power supplies with
MetroCluster configurations .................................................................. 140
Stretch MetroCluster configuration with disks ........................................... 141
Required documentation, tools, and equipment .......................................... 142
Cabling a stretch MetroCluster configuration with disks ........................................ 146
Stretch MetroCluster configuration on single-enclosure HA pairs ............. 147
Assigning disk pools in a stretch MetroCluster configuration .................... 147
Managing the default configuration speed of a stretch MetroCluster
configuration ...................................................................................................... 149
Changing the default configuration speed of a stretch MetroCluster
configuration .......................................................................................... 149
Resetting a stretch MetroCluster configuration to the default speed .......... 151
Implementing fabric-attached MetroCluster configurations with
disks ...................................................................................................... 153
Planning a fabric-attached MetroCluster configuration using disks ....................... 153
Setup requirements and restrictions for fabric-attached MetroCluster
configurations with disks ....................................................................... 153
Configuration limitations for fabric-attached MetroCluster
configurations with disks ....................................................................... 155
8 | High Availability and MetroCluster Configuration Guide
Hardware requirements for using disks in a MetroCluster configuration
with V-Series systems ........................................................................... 155
Required connections for using uninterruptible power supplies with
MetroCluster configurations .................................................................. 156
Requirements for a shared-switches configuration ..................................... 157
Fabric-attached MetroCluster configuration with disks .............................. 157
How fabric-attached MetroCluster configurations use Brocade and
Cisco Fibre Channel switches ............................................................... 158
Planning the fabric-attached MetroCluster installation ............................... 159
Required documentation, tools, and equipment .......................................... 160
Converting an HA pair to a fabric-attached MetroCluster configuration ............... 164
Cabling a fabric-attached MetroCluster configuration with disks .......................... 166
Configuring the switches ............................................................................. 168
Cabling Node A ........................................................................................... 168
Cabling Node B ........................................................................................... 172
Assigning disk pools ................................................................................... 176
Verifying disk paths .................................................................................... 177
Setting up a shared-switches configuration ................................................. 178
Fabric-attached MetroCluster configuration on single-enclosure HA
pairs ....................................................................................................... 180
Configuration differences for fabric-attached MetroCluster
configurations on single-enclosure HA pairs ........................................ 181
Setting preferred primary port in a MetroCluster configuration ............................. 182
Removing the preferred primary port in a fabric-attached MetroCluster
configuration ...................................................................................................... 183
Implementing a stretch or fabric-attached MetroCluster
configuration with array LUNs .......................................................... 184
Planning for a MetroCluster configuration with array LUNs ................................. 184
Implementation overview for a MetroCluster configuration with array
LUNs ..................................................................................................... 185
Requirements for a MetroCluster configuration with array LUNs ............. 186
Requirements for a shared-switches MetroCluster configuration with
array LUNs ............................................................................................ 188
Recommended fabric-attached MetroCluster configuration with array
LUNs ..................................................................................................... 188
Recommended stretch MetroCluster configuration with array LUNs ........ 192
Table of Contents | 9
Cabling guidelines for a MetroCluster configuration with array LUNs ..... 195
Planning zoning for a MetroCluster configuration with array LUNs ......... 196
Connecting devices in a MetroCluster configuration with array LUNs .................. 198
Connecting the local V-Series systems in a MetroCluster configuration . . . 199
Connecting the remote V-Series systems in a MetroCluster
configuration .......................................................................................... 203
Connecting the switch fabric in a MetroCluster configuration with array
LUNs ..................................................................................................... 205
Connecting the fabric and storage array in a MetroCluster configuration
with array LUNs .................................................................................... 207
Setting up a shared-switches configuration ................................................. 209
Setting preferred primary port in a MetroCluster configuration ................. 212
Removing the preferred primary port in a fabric-attached MetroCluster
configuration .......................................................................................... 213
Configuring zoning in a MetroCluster configuration with array LUNs ...... 213
Changing the configuration speed of a stretch MetroCluster
configuration .......................................................................................... 214
Setting up Data ONTAP after connecting devices in a MetroCluster
configuration with array LUNs ......................................................................... 215
Testing a MetroCluster configuration with array LUNs ......................................... 215
Testing zoning of FC-VI ports in a MetroCluster configuration with
array LUNs ............................................................................................ 215
Verifying proper setup at MetroCluster sites with storage arrays ............... 216
Simulating a disaster recovery in a MetroCluster configuration with
array LUNs ............................................................................................ 217
Implementing a MetroCluster configuration with both disks and
array LUNs ........................................................................................... 219
Planning a MetroCluster configuration with disks and array LUNs ....................... 219
Guidelines for implementing a MetroCluster configuration with disks
and array LUNs ..................................................................................... 219
Supported fabric-attached MetroCluster configuration with disks and
array LUNs ............................................................................................ 221
Supported stretch MetroCluster configuration with disks and array
LUNs ..................................................................................................... 223
Testing a MetroCluster configuration with disks and array LUNs ......................... 225
10 | High Availability and MetroCluster Configuration Guide
Performing nondisruptive shelf replacement in a MetroCluster
configuration ........................................................................................ 227
Preparing for nondisruptive shelf replacement ....................................................... 227
Replacing the disk shelf nondisruptively ................................................................ 228
Verifying the disks after the shelf replacement ....................................................... 230
Recovering from a disaster by using MetroCluster configurations ..... 232
Conditions that constitute a disaster ........................................................................ 232
Ways to determine whether a disaster occurred .......................................... 232
Failures that do not require disaster recovery ............................................. 232
Recovering from a disaster ...................................................................................... 233
Restricting access to the disaster site node .................................................. 234
Forcing a node into takeover mode ............................................................. 235
Remounting volumes of the failed node ..................................................... 235
Recovering LUNs of the failed node ........................................................... 236
Fixing failures caused by the disaster ......................................................... 237
Reestablishing the MetroCluster configuration ........................................... 238
Feature update record .............................................................................. 243
Copyright information ............................................................................. 249
Trademark information ........................................................................... 250
How to send your comments .................................................................... 251
Index ........................................................................................................... 252
11
Understanding HA pairs
HA pairs provide hardware redundancy that is required for nondisruptive operations and fault
tolerance and give each node in the pair the software functionality to take over its partner's storage
and subsequently give back the storage.
What an HA pair is
An HA pair is two storage systems (nodes) whose controllers are connected to each other either
directly or, in the case of a fabric-attached MetroCluster, through switches and FC-VI interconnect
adapters. In this configuration, one node can take over its partner's storage to provide continued data
service if the partner goes down.
You can configure the HA pair so that each node in the pair shares access to a common set of
storage, subnets, and tape drives, or each node can own its own distinct set of storage.
The controllers are connected to each other through an HA interconnect. This allows one node to
serve data that resides on the disks of its failed partner node. Each node continually monitors its
partner, mirroring the data for each other’s nonvolatile memory (NVRAM or NVMEM). The
interconnect is internal and requires no external cabling if both controllers are in the same chassis.
Takeover is the process in which a node takes over the storage of its partner. Giveback is the process
in which that storage is returned to the partner. Both processes can be initiated manually or
configured for automatic initiation.
How HA pairs support nondisruptive operations and fault
tolerance
HA pairs provide fault tolerance and let you perform nondisruptive operations, including hardware
and software upgrades, and hardware maintenance.
•
•
Fault tolerance
When one node fails or becomes impaired and a takeover occurs, the partner node continues to
serve the failed node’s data.
Nondisruptive software upgrades or hardware maintenance
During hardware maintenance or upgrades, when you halt one node and a takeover occurs
(automatically, unless you specify otherwise), the partner node continues to serve data for the
halted node while you upgrade or perform maintenance on the node you halted.
During nondisruptive upgrades of Data ONTAP, the user manually enters the storage
failover takeover command to take over the partner node to allow the software upgrade to
occur. The takeover node continues to serve data for both nodes during this operation.
12 | High Availability and MetroCluster Configuration Guide
For more information about nondisruptive software upgrades, see the Data ONTAP Upgrade and
Revert/Downgrade Guide for 7-Mode.
The HA pair supplies nondisruptive operation and fault tolerance due to the following aspects of its
configuration:
•
•
The controllers in the HA pair are connected to each other either through an HA interconnect
consisting of adapters and cables, or, in systems with two controllers in the same chassis, through
an internal interconnect.
The nodes use the interconnect to perform the following tasks:
• Continually check whether the other node is functioning
• Mirror log data for each other’s NVRAM or NVMEM
They use two or more disk shelf loops, or storage arrays, in which the following conditions apply:
•
•
Each node manages its own disks or array LUNs.
In case of takeover, the surviving node provides read/write access to the partner's disks or
array LUNs until the failed node becomes available again.
Note: Disk ownership is established by Data ONTAP or the administrator rather than by which
disk shelf the disk is attached to.
For more information about disk ownership, see the Data ONTAP Storage Management Guide
for 7-Mode.
•
•
They own their spare disks, spare array LUNs, or both, and do not share them with the other
node.
They each have mailbox disks or array LUNs on the root volume that perform the following
tasks:
•
•
•
Maintain consistency between the pair
Continually check whether the other node is running or whether it has performed a takeover
Store configuration information
Related concepts
Where to find procedures for nondisruptive operations with HA pairs on page 12
Managing DS14mk2 or DS14mk4 disk shelves in an HA pair on page 122
Where to find procedures for nondisruptive operations with HA pairs
By taking advantage of an HA pair's takeover and giveback operations, you can change hardware
components and perform software upgrades in your configuration without disrupting access to the
system's storage. You can refer to the specific documents for the required procedures.
You can perform nondisruptive operations on a system by having its partner take over the system's
storage, performing maintenance, and then giving back the storage. The following table lists where
you can find information on specific procedures:
Understanding HA pairs | 13
If you want to perform this task
nondisruptively...
See the...
Upgrade Data ONTAP
Data ONTAP Upgrade and Revert/Downgrade
Guide for 7-Mode
Replace a hardware FRU component
FRU procedures for your platform
How the HA pair improves fault tolerance
A storage system has a variety of single points of failure, such as certain cables or hardware
components. An HA pair greatly reduces the number of single points of failure because if a failure
occurs, the partner can take over and continue serving data for the affected system until the failure is
fixed.
Single point of failure definition
A single point of failure represents the failure of a single hardware component that can lead to loss of
data access or potential loss of data.
Single point of failure does not include multiple/rolling hardware errors, such as triple disk failure,
dual disk shelf module failure, and so on.
All hardware components included with your storage system have demonstrated very good reliability
with low failure rates. If a hardware component such as a controller or adapter fails, you can use the
controller failover function to provide continuous data availability and preserve data integrity for
client applications and users.
Single point of failure analysis for HA pairs
Different individual hardware components and cables in the storage system are single points of
failure, but an HA configuration can eliminate these points to improve data availability.
Hardware
components
Single point of failure
Controller
Yes
No
If a controller fails, the node automatically fails
over to its partner node. The partner (takeover)
node serves data for both of the nodes.
NVRAM
Yes
No
If an NVRAM adapter fails, the node
automatically fails over to its partner node. The
partner (takeover) node serves data for both of
the nodes.
Stand-alone HA pair
How storage failover eliminates single point
of failure
14 | High Availability and MetroCluster Configuration Guide
Hardware
components
Single point of failure
CPU fan
Yes
Multiple NICs with
interface groups
(virtual interfaces)
Maybe, if all No
NICs fail
If one of the networking links within an
interface group fails, the networking traffic is
automatically sent over the remaining
networking links on the same node. No failover
is needed in this situation.
If all the NICs in an interface group fail, the
node automatically fails over to its partner
node if failover is enabled for the interface
group.
Single NIC
Yes
No
If a NIC fails, the node automatically fails over
to its partner node if failover is enabled for the
NIC.
FC-AL adapter or SAS Yes
HBA
No
If an FC-AL adapter for the primary loop fails
for a configuration without multipath HA, the
partner node attempts a takeover at the time of
failure. With multipath HA, no takeover is
required.
If the FC-AL adapter for the secondary loop
fails for a configuration without multipath HA,
the failover capability is disabled, but both
nodes continue to serve data to their respective
applications and users, with no impact or delay.
With multipath HA, failover capability is not
affected.
No, if dual- No
path cabling
is used
If an FC-AL loop or SAS stack breaks in a
configuration that does not have multipath HA,
the break could lead to a failover, depending on
the shelf type. The partnered nodes invoke the
negotiated failover feature to determine which
node is best for serving data, based on the disk
shelf count. When multipath HA is used, no
failover is required.
FC-AL or SAS cable
(controller-to-shelf,
shelf-to-shelf )
Stand-alone HA pair
No
How storage failover eliminates single point
of failure
If the CPU fan fails, the node automatically
fails over to its partner node. The partner
(takeover) node serves data for both of the
nodes.
Understanding HA pairs | 15
Hardware
components
Single point of failure
Disk shelf module
No, if dual- No
path cabling
is used
If a disk shelf module fails in a configuration
that does not have multipath HA, the failure
could lead to a failover. The partnered nodes
invoke the negotiated failover feature to
determine which node is best for serving data,
based on the disk shelf count. When multipath
HA is used, there is no impact.
Disk drive
No
No
If a disk fails, the node can reconstruct data
from the RAID4 parity disk. No failover is
needed in this situation.
Power supply
Maybe, if
both power
supplies fail
No
Both the controller and disk shelf have dual
power supplies. If one power supply fails, the
second power supply automatically kicks in.
No failover is needed in this situation. If both
power supplies fail, the node automatically
fails over to its partner node, which serves data
for both nodes.
Fan (controller or disk
shelf)
Maybe, if
both fans
fail
No
Both the controller and disk shelf have multiple
fans. If one fan fails, the second fan
automatically provides cooling. No failover is
needed in this situation. If both fans fail, the
node automatically fails over to its partner
node, which serves data for both nodes.
HA interconnect
adapter
Not
applicable
No
If an HA interconnect adapter fails, the failover
capability is disabled but both nodes continue
to serve data to their respective applications
and users.
HA interconnect cable
Not
applicable
No
The HA interconnect adapter supports dual HA
interconnect cables. If one cable fails, the
heartbeat and NVRAM data are automatically
sent over the second cable with no delay or
interruption.
If both cables fail, the failover capability is
disabled but both nodes continue to serve data
to their respective applications and users.
Stand-alone HA pair
How storage failover eliminates single point
of failure
16 | High Availability and MetroCluster Configuration Guide
Standard HA pair diagram
Each node in a standard HA pair requires a network connection, an HA interconnect between the
controllers, and connections both to its own disk shelves as well as its partner node's shelves.
Standard HA pair
Network
HA Interconnect
Node1
Node2
Node1
Storage
Node2
Storage
Primary connection
Redundant primary connection
Standby connection
Redundant standby connection
This diagram shows a standard HA pair with native disk shelves and multipath HA.
This diagram shows DS4243 disk shelves. For more information about cabling SAS disk
shelves, see the Universal SAS and ACP Cabling Guide on the NetApp Support Site.
Understanding HA pairs | 17
Comparison of HA pair types
The different types of HA pairs support different capabilities for data duplication, distance between
nodes, and failover.
HA pair type Data
Distance
duplication? between nodes
Failover
Notes
possible after
loss of entire
node
(including
storage)?
Standard HA
pair
No
Use this configuration to provide
higher availability by protecting
against many hardware singlepoints-of-failure.
No
Use this configuration to add
increased data protection to the
benefits of a standard HA pair.
Yes
Use this configuration to provide
data and hardware duplication to
protect against a local disaster
(for example, a power outage to
one node).
No
Up to 500
meters
Note: SAS
configurations
are limited to
5 meters
between
nodes
Mirrored HA
pair
Yes
Up to 500
meters
Note: SAS
configurations
are limited to
5 meters
between
nodes
Stretch
MetroCluster
Yes
Up to 500
meters (270
meters if
operating at 4
Gbps)
Note: SAS
configurations
are limited to
20 meters
between
nodes
18 | High Availability and MetroCluster Configuration Guide
HA pair type Data
Distance
duplication? between nodes
Failover
Notes
possible after
loss of entire
node
(including
storage)?
Fabricattached
MetroCluster
Yes
Yes
Up to 200
kilometers,
depending on
switch
configuration.
See the
Interoperability
Matrix at
support.netapp.c
om/NOW/
products/
interoperability.
Use this configuration to provide
data and hardware duplication to
protect against a larger-scale
disaster, such as the loss of an
entire site.
19
Understanding mirrored HA pairs
Mirrored HA pairs provide high availability through failover, just as standard HA pairs do.
Additionally, mirrored HA pairs maintain two complete copies of all mirrored data. These copies are
called plexes and are continually and synchronously updated every time Data ONTAP writes to a
mirrored aggregate. The plexes can be physically separated to protect against the loss of one set of
disks or array LUNs.
Note: Mirrored HA pairs do not provide the capability to fail over to the partner node if one node
is completely lost. For example, if power is lost to one entire node, including its storage, you
cannot fail over to the partner node. For this capability, use a MetroCluster.
Mirrored HA pairs use SyncMirror. For more information about SyncMirror, see the Data ONTAP 7Mode Data Protection Online Backup and Recovery Guide.
Advantages of mirrored HA pairs
Data mirroring provides additional data protection in the event of disk failures and reduces the need
for failover in the event of other component failures.
Mirroring your data protects it from the following problems that would cause data loss without
mirroring:
•
•
•
•
The failure or loss of two or more disks in a RAID4 aggregate
The failure or loss of three or more disks in a RAID-DP (RAID double-parity) aggregate
The failure of an array LUN; for example, because of a double disk failure on the storage array
The failure of a storage array
The failure of an FC-AL adapter, SAS HBA, disk shelf loop or stack, or disk shelf module does not
require a failover in a mirrored HA pair.
Similar to standard HA pairs, if either node in a mirrored HA pair becomes impaired or cannot access
its data, the other node can automatically serve the impaired node’s data until the problem is
corrected.
Asymmetrically mirrored HA pairs
You can selectively mirror your storage. For example, you could mirror all the storage on one node,
but none of the storage on the other node. Takeover will function normally. However, any
unmirrored data is lost if the storage that contains it is damaged or destroyed.
Note: You must connect the unmirrored storage to both nodes, just as for mirrored storage. You
cannot have storage that is connected to only one node in an HA pair.
20 | High Availability and MetroCluster Configuration Guide
Understanding takeover and giveback
Takeover and giveback are the operations that let you take advantage of the HA configuration to
perform nondisruptive operations and avoid service interruptions. Takeover is the process in which a
node takes over the storage of its partner. Giveback is the process in which the storage is returned to
the partner. You can initiate the processes in different ways.
When takeovers occur
Takeovers can be initiated manually or occur automatically when a failover event happens,
depending on how you configure the HA pair. In some cases, takeovers occur automatically
regardless of configuration.
Takeovers can occur under the following conditions:
•
•
•
A takeover is manually initiated.
A node is in an HA pair with the default configuration for immediate takeover on panic, and that
node undergoes a software or system failure that leads to a panic.
By default, the node automatically performs a giveback to return the partner to normal operation
after the partner has recovered from the panic and booted up.
A node that is in an HA pair undergoes a system failure (for example, a loss of power) and cannot
reboot.
Note: If the storage for a node also loses power at the same time, a standard takeover is not
possible. For MetroCluster configurations, you can initiate a forced takeover in this situation.
•
•
•
•
•
One or more network interfaces that are configured to support failover become unavailable.
A node does not receive heartbeat messages from its partner.
This could happen if the partner experienced a hardware or software failure that did not result in a
panic but still prevented it from functioning correctly.
You halt one of the nodes without using the -f parameter.
You reboot one of the nodes without using the -f parameter.
Hardware-assisted takeover is enabled and triggers a takeover when the remote management
device (RLM or Service Processor) detects failure of the partner node.
Understanding takeover and giveback | 21
Failover event cause-and-effect table
Failover events cause a controller failover in HA pairs. The storage system responds differently
depending on the event and the type of HA pair.
Cause-and-effect table for standard or mirrored HA pairs
Event
Does the event
Does the event
trigger failover? prevent a future
failover from
occurring, or a
failover from
occurring
successfully?
Is data still available on the affected
volume after the event?
Single storage
system
Standard or
mirrored HA pair
Single disk
failure
No
No
Yes
Yes
Double disk
failure (2
disks fail in
same RAID
group)
Yes, unless you
are using
SyncMirror or
RAID-DP, then
no.
Maybe. If root
No, unless you are
No, unless you are
volume has
using RAID-DP or
using RAID-DP or
double disk
SyncMirror, then yes. SyncMirror, then
failure, or if the
yes.
mailbox disks are
affected, no
failover is
possible.
Triple disk
failure (3
disks fail in
same RAID
group)
Maybe. If
SyncMirror is
being used, no
takeover occurs;
otherwise, yes.
Maybe. If root
No
volume has triple
disk failure, no
failover is
possible.
No
Single HBA
(initiator)
failure, Loop
A
Maybe. If
SyncMirror or
multipath HA is
in use, then no;
otherwise, yes.
Maybe. If root
volume has
double disk
failure, no
failover is
possible.
Yes, if multipath HA
or SyncMirror is
being used.
Yes, if multipath HA
or SyncMirror is
being used, or if
failover succeeds.
Single HBA
(initiator)
failure, Loop
B
No
Yes, unless you Yes, if multipath HA
are using
or SyncMirror is
SyncMirror or
being used.
multipath HA
and the mailbox
disks are not
affected, then no.
Yes, if multipath HA
or SyncMirror is
being used, or if
failover succeeds.
22 | High Availability and MetroCluster Configuration Guide
Event
Does the event
Does the event
trigger failover? prevent a future
failover from
occurring, or a
failover from
occurring
successfully?
Is data still available on the affected
volume after the event?
Single storage
system
Standard or
mirrored HA pair
Single HBA
initiator
failure (both
loops at the
same time)
Yes, unless the
data is mirrored
on a different
(up) loop or
multipath HA is
in use, then no
takeover needed.
Maybe. If the
data is mirrored
or multipath HA
is being used and
the mailbox disks
are not affected,
then no;
otherwise, yes.
No, unless the data is
mirrored or multipath
HA is in use, then
yes.
No failover needed if
data is mirrored or
multipath HA is in
use.
AT-FCX
failure (Loop
A)
Only if multidisk
volume failure or
open loop
condition occurs,
and neither
SyncMirror nor
multipath HA is
in use.
Maybe. If root
volume has
double disk
failure, no
failover is
possible.
No
Yes, if failover
succeeds.
AT-FCX
failure (Loop
B)
No
Maybe. If
SyncMirror or
multipath HA is
in use, then no;
otherwise, yes.
Yes, if multipath HA
or SyncMirror is in
use.
Yes
IOM failure
(Loop A)
Only if multidisk
volume failure or
open loop
condition occurs,
and neither
SyncMirror nor
multipath HA is
in use.
Maybe. If root
volume has
double disk
failure, no
failover is
possible.
No
Yes, if failover
succeeds.
IOM failure
(Loop B)
No
Maybe. If
SyncMirror or
multipath HA is
in use, then no;
otherwise, yes.
Yes, if multipath HA
or SyncMirror is in
use.
Yes
Understanding takeover and giveback | 23
Event
Does the event
Does the event
trigger failover? prevent a future
failover from
occurring, or a
failover from
occurring
successfully?
Is data still available on the affected
volume after the event?
Single storage
system
Standard or
mirrored HA pair
Shelf
(backplane)
failure
Only if multidisk
volume failure or
open loop
condition occurs,
and data isn’t
mirrored.
Maybe. If root
volume has
double disk
failure or if the
mailboxes are
affected, no
failover is
possible.
Maybe. If data is
mirrored, then yes;
otherwise, no.
Maybe. If data is
mirrored, then yes;
otherwise, no.
Shelf, single
power failure
No
No
Yes
Yes
Shelf, dual
power failure
Only if multidisk
volume failure or
open loop
condition occurs
and data is not
mirrored.
Maybe. If root
Maybe. If data is
volume has
mirrored, then yes;
double disk
otherwise, no.
failure, or if the
mailbox disks are
affected, no
failover is
possible.
Maybe. If data is
mirrored, then yes;
otherwise, no.
Controller,
single power
failure
No
No
Yes
Yes
Controller,
dual power
failure
Yes
Yes, until power
is restored.
No
Yes, if failover
succeeds.
HA
interconnect
failure (1
port)
No
No
Not applicable
Yes
HA
interconnect
failure (both
ports)
No
Yes
Not applicable
Yes
24 | High Availability and MetroCluster Configuration Guide
Event
Does the event
Does the event
trigger failover? prevent a future
failover from
occurring, or a
failover from
occurring
successfully?
Is data still available on the affected
volume after the event?
Single storage
system
Standard or
mirrored HA pair
Ethernet
interface
failure
(primary, no
interface
group)
Yes, if set up to
do so.
No
Yes
Yes
Ethernet
interface
failure
(primary,
interface
group)
Yes, if set up to
do so.
No
Yes
Yes
Ethernet
interface
failure
(secondary,
interface
group)
Yes, if set up to
do so.
No
Yes
Yes
Ethernet
interface
failure
(interface
group, all
ports)
Yes, if set up to
do so.
No
Yes
Yes
Tape
interface
failure
No
No
Yes
Yes
Heat exceeds
permissible
amount
Yes
No
No
No
Fan failures
(disk shelves
or controller)
No
No
Yes
Yes
Understanding takeover and giveback | 25
Event
Does the event
Does the event
trigger failover? prevent a future
failover from
occurring, or a
failover from
occurring
successfully?
Is data still available on the affected
volume after the event?
Single storage
system
Standard or
mirrored HA pair
Reboot
Yes
No
No
Yes, if failover
occurs.
Panic
Yes
No
No
Yes, if failover
occurs.
Cause-and-effect table for stretch and fabric-attached MetroClusters
Event
Does the
Does the
Is data still available on the affected volume
event trigger event
after the event?
failover?
prevent a
Stretch MetroCluster Fabric Attached
future
MetroCluster
failover from
occurring, or
a failover
from
occurring
successfully?
Single disk
failure
No
No
Yes
Yes
Double disk
failure (2 disks
fail in same
RAID group)
No
No
Yes
Yes, with no failover
necessary.
Triple disk
failure (3 disks
fail in same
RAID group)
No
No
Yes, with no failover
necessary.
Yes, with no failover
necessary.
Single HBA
No
(initiator)
failure, Loop A
No
Yes
Yes, with no failover
necessary.
Single HBA
No
(initiator)
failure, Loop B
No
Yes
Yes, with no failover
necessary.
26 | High Availability and MetroCluster Configuration Guide
Event
Does the
Does the
Is data still available on the affected volume
event trigger event
after the event?
failover?
prevent a
Stretch MetroCluster Fabric Attached
future
MetroCluster
failover from
occurring, or
a failover
from
occurring
successfully?
Single HBA
No
initiator failure,
(both loops at
the same time)
No
Yes, with no failover
necessary.
Yes, with no failover
necessary.
AT-FCX
failure (Loop
A)
No
No
Yes
Not applicable
AT-FCX
failure (Loop
B)
No
No
Yes
Not applicable
ESH4 or IOM
Failure (loop
A)
No
No
Yes
Yes
ESH4 or IOM
Failure (loop
B)
No
No
Yes
Yes
Shelf
(backplane)
failure
No
No
Yes
Yes, with no failover
necessary.
Shelf, single
power failure
No
No
Yes
Yes
Shelf, dual
power failure
No
No
Yes
Yes, with no failover
necessary.
Controller,
single power
failure
No
No
Yes
Yes
Controller,
dual power
failure
Yes
Yes, until
power is
restored.
Yes, if failover
succeeds.
Yes, if failover
succeeds.
Understanding takeover and giveback | 27
Event
Does the
Does the
Is data still available on the affected volume
event trigger event
after the event?
failover?
prevent a
Stretch MetroCluster Fabric Attached
future
MetroCluster
failover from
occurring, or
a failover
from
occurring
successfully?
HA
interconnect
failure (1 port)
No
No
Yes
Yes
HA
interconnect
failure (both
ports)
No
No; failover
is possible.
Yes
Yes
Ethernet
interface
failure
(primary, no
interface
group)
Yes, if set up
to do so.
No
Yes
Yes
Ethernet
interface
failure
(primary,
interface
group)
Yes, if set up
to do so.
No
Yes
Yes
Ethernet
interface
failure
(secondary,
interface
group)
Yes, if set up
to do so.
No
Yes
Yes
Ethernet
interface
failure
(interface
group, all
ports)
Yes, if set up
to do so.
No
Yes
Yes
28 | High Availability and MetroCluster Configuration Guide
Event
Does the
Does the
Is data still available on the affected volume
event trigger event
after the event?
failover?
prevent a
Stretch MetroCluster Fabric Attached
future
MetroCluster
failover from
occurring, or
a failover
from
occurring
successfully?
Tape interface
failure
No
No
Yes
Yes
Heat exceeds
permissible
amount
Yes
No
Yes, if failover occurs.
Yes, if failover occurs.
Fan failures
No
(disk shelves or
controller)
No
Yes
Yes
Reboot
Yes
No
Maybe. Depends on
cause of reboot.
Maybe. Depends on
cause of reboot.
Panic
Yes
No
Maybe. Depends on
cause of panic.
Maybe. Depends on
cause of panic.
How hardware-assisted takeover speeds up takeover
Hardware-assisted takeover speeds up the takeover process by using a node's remote management
device (SP or RLM) to detect failures and quickly initiate the takeover rather than waiting for Data
ONTAP to recognize that the partner's heartbeat has stopped.
Without hardware-assisted takeover, if a failure occurs, the partner waits until it notices that the node
is no longer giving a heartbeat, confirms the loss of heartbeat, and then initiates the takeover.
The hardware-assisted takeover feature uses the following process to take advantage of the remote
management device and avoid that wait:
1. The remote management device monitors the local system for certain types of failures.
2. If a failure is detected, the remote management device immediately sends an alert to the partner
node.
3. Upon receiving the alert, the partner initiates takeover.
The hardware-assisted takeover option (cf.hw_assist.enable) is enabled by default.
Understanding takeover and giveback | 29
What happens during takeover
When a takeover occurs, the unimpaired partner node takes over the functions and disk drives of the
failed node by creating an emulated storage system.
The emulated system performs the following tasks:
•
•
Assumes the identity of the failed node
Accesses the failed node’s disks, array LUNs, or both, and serves its data to clients
The partner node maintains its own identity and its own primary functions, but also handles the
added functionality of the failed node through the emulated node.
Note: When a takeover occurs, existing CIFS sessions are terminated. A graceful shutdown of the
CIFS sessions is not possible, and some data loss could occur for CIFS users.
If the node doing the takeover panics
If the node that is performing the takeover panics within 60 seconds of initiating takeover, the
following events occur:
•
•
•
The node that panicked reboots.
After it reboots, the node performs self-recovery operations and is no longer in takeover mode.
Failover is disabled.
What happens during giveback
The local node returns ownership of the aggregates and volumes to the partner node after any issues
on the partner node are resolved or maintenance is complete. In addition, the local node returns
ownership when the partner node has booted up and giveback is initiated either manually or
automatically.
When the failed node is functioning again, the following events can occur:
•
•
•
You issue a cf giveback command that terminates the emulated node on the partner.
The failed node resumes normal operation, serving its own data.
The HA pair resumes normal operation, with each node ready to take over for its partner if the
partner fails.
30 | High Availability and MetroCluster Configuration Guide
Background disk firmware update and takeover and
giveback
Background disk firmware updates affect HA pair takeover and giveback operations differently,
depending on how those operations are initiated.
How background disk firmware update affects takeover and giveback:
•
•
If a background disk firmware update is occurring on a disk on the source (or takeover) node,
takeover operations are delayed for less than one minute until the disk firmware update completes
on that disk.
If a background disk firmware update is occurring on a disk on a node and it panics, takeover of
the panicked node begins immediately.
31
Planning your HA pair configuration
As you plan your HA pair, you must consider recommended best practices, the requirements, and the
possible variations.
Best practices for HA pairs
To ensure that your HA pair is robust and operational, you need to be familiar with configuration
best practices.
•
•
•
•
•
•
•
•
•
•
•
•
Make sure that each power supply unit in the storage system is on a different power grid so that a
single power outage does not affect all power supply units.
Use interface groups (virtual interfaces) to provide redundancy and improve availability of
network communication.
Follow the documented procedures in the Data ONTAP Upgrade and Revert/Downgrade Guide
for 7-Mode when upgrading your HA pair.
Maintain consistent configuration between the two nodes.
An inconsistent configuration is often the cause of failover problems.
Test the failover capability routinely (for example, during planned maintenance) to ensure proper
configuration.
Make sure that each node has sufficient resources to adequately support the workload of both
nodes during takeover mode.
Use the Config Advisor tool to help ensure that failovers are successful.
If your system supports remote management (through an RLM or Service Processor), make sure
that you configure it properly, as described in the Data ONTAP System Administration Guide for
7-Mode.
Follow recommended limits for FlexVol volumes, dense volumes, Snapshot copies, and LUNs to
reduce the takeover or giveback time.
When adding traditional or FlexVol volumes to an HA pair, consider testing the takeover and
giveback times to ensure that they fall within your requirements.
For systems using disks, check for and remove any failed disks, as described in the Data ONTAP
Storage Management Guide for 7-Mode.
Multipath HA is required on all HA pairs except for some FAS22xx system configurations, which
use single-path HA and lack the redundant standby connections.
To ensure that you receive prompt notification if takeover becomes disabled, configure your
system for automatic e-mail notification for the takeover impossible EMS messages:
•
•
•
•
ha.takeoverImpVersion
ha.takeoverImpLowMem
ha.takeoverImpDegraded
ha.takeoverImpUnsync
32 | High Availability and MetroCluster Configuration Guide
•
• ha.takeoverImpIC
• ha.takeoverImpHotShelf
• ha.takeoverImpNotDef
Set the cf.giveback.auto.cancel.on_network_failure option to true if networkrelated failovers are enabled.
Related tasks
Verifying the HA pair cabling and configuration on page 93
Setup requirements and restrictions for standard HA pairs
You must follow certain requirements and restrictions when setting up a new standard HA pair.
These requirements help you ensure the data availability benefits of the HA pair design.
The following list specifies the requirements and restrictions you should be aware of when setting up
a new standard HA pair:
•
•
•
Architecture compatibility
Both nodes must have the same system model and be running the same Data ONTAP software
and system firmware versions. See the Data ONTAP Release Notes for 7-Mode for the list of
supported systems.
Nonvolatile memory (NVRAM or NVMEM) size and version compatibility
The size and version of the system's nonvolatile memory must be identical on both nodes in an
HA pair.
Storage capacity
The number of disks or array LUNs must not exceed the maximum configuration capacity. If
your system uses both native disks and array LUNs, the combined total of disks and array LUNs
cannot exceed the maximum configuration capacity. In addition, the total storage attached to each
node must not exceed the capacity for a single node.
To determine the maximum capacity for a system using disks, array LUNs, or both, see the
Hardware Universe (formerly the System Configuration Guide) at support.netapp.com/
knowledge/docs/hardware/NetApp/syscfg/index.shtml.
Note: After a failover, the takeover node temporarily serves data from all the storage in the HA
pair.
•
Disks and disk shelf compatibility
•
•
FC, SATA, and SAS storage are supported in standard HA pairs.
FC disks cannot be mixed on the same loop as SATA or SAS disks.
• One node can have only one type of storage and the partner node can have a different type, if
needed.
• Multipath HA is required on all HA pairs except for some FAS22xx system configurations,
which use single-path HA and lack the redundant standby connections.
Mailbox disks or array LUNs on the root volume
Planning your HA pair configuration | 33
•
•
Two disks are required if the root volume is on a disk shelf.
One array LUN is required if the root volume is on a storage array.
The mailbox disks and LUNs are used for the following tasks:
•
•
•
• Maintain consistency between the nodes in the HA pair
• Continually check whether the other node is running or whether it has performed a takeover
• Store configuration information that is not specific to any particular node
HA interconnect adapters and cables must be installed unless the system has two controllers in
the chassis and an internal interconnect.
Nodes must be attached to the same network and the Network Interface Cards (NICs) must be
configured correctly.
The same system software, such as Common Internet File System (CIFS) or Network File System
(NFS), must be licensed and enabled on both nodes.
Note: If a takeover occurs, the takeover node can provide only the functionality for the licenses
installed on it. If the takeover node does not have a license that was being used by the partner
node to serve data, your HA pair loses functionality after a takeover.
•
For an HA pair using array LUNs, both nodes in the pair must be able to detect the same array
LUNs.
However, only the node that is the configured owner of a LUN has read-and-write access to the
LUN. During takeover operations, the emulated storage system maintains read-and-write access
to the LUN.
Setup requirements and restrictions for mirrored HA pairs
The restrictions and requirements for mirrored HA pairs include those for a standard HA pair with
these additional requirements for disk pool assignments and cabling.
•
You must ensure that your pools are configured correctly:
•
•
•
•
•
Disks or array LUNs in the same plex must be from the same pool, with those in the opposite
plex from the opposite pool.
There must be sufficient spares in each pool to account for a disk or array LUN failure.
Bboth plexes of a mirror should not reside on the same disk shelf, as it would result in a single
point of failure.
See the Data ONTAP 7-Mode Data Protection Online Backup and Recovery Guide for more
information about requirements for setting up SyncMirror with array LUNs.
The cf.mode option must be set to ha.
If you are using array LUNs, paths to an array LUN must be redundant.
Related concepts
Setup requirements and restrictions for standard HA pairs on page 32
34 | High Availability and MetroCluster Configuration Guide
Multipath HA requirement
Multipath HA is required on all HA pairs except for some FAS22xx system configurations, which
use single-path HA and lack the redundant standby connections. Multipath HA was previously
referred to as Multipath Storage.
What multipath HA for HA pairs is
Multipath HA provides redundancy for the path from each controller to every disk shelf in the
configuration. It is the preferred method for cabling a storage system. An HA pair without multipath
HA has only one path from each controller to every disk, but an HA pair with multipath HA has two
paths from each controller to each disk, regardless of which node owns the disk.
The following diagram shows the connections between the controllers and the disk shelves for an
example HA pair using multipath HA. The redundant primary connections and the redundant standby
connections are the additional connections required for multipath HA for HA pairs.
Network
HA Interconnect
Node1
Node2
Node1
Storage
Node2
Storage
Primary connection
Redundant primary connection
Standby connection
Redundant standby connection
Planning your HA pair configuration | 35
How the connection types are used
A multipath HA configuration uses primary, redundant and standby connections to ensure continued
service in the event of the failure of an individual connection.
The following table outlines the connection types used for multipath HA for HA pairs, and how the
connections are used:
Connection type
How the connection is used
Primary connection
For normal operation, used to serve data (loadbalanced with redundant primary connection).
Redundant primary connection
For normal operation, used to serve data (loadbalanced with primary connection).
Standby connection
For normal operation, used for heartbeat
information only. After a takeover, assumes role
of primary connection.
Redundant standby connection
Not used for normal operation. After a takeover,
assumes role of redundant primary connection.
If the standby connection is unavailable at
takeover time, assumes role of primary
connection.
Advantages of multipath HA
Multipath connections in an HA pair reduce single-points-of-failure.
By providing two paths from each controller to every disk shelf, multipath HA provides the
following advantages:
•
•
•
The loss of a disk shelf module, connection, or host bus adapter (HBA) does not require a
failover.
The same storage system can continue to access the data using the redundant path.
The loss of a single disk shelf module, connection, or HBA does not prevent a successful failover.
The takeover node can access its partner’s disks using the redundant path.
You can replace modules without having to initiate a failover.
Note: While multipath HA adds value to a stretch MetroCluster environment, it is not necessary in
a fabric MetroCluster configuration since multiple paths already exist.
36 | High Availability and MetroCluster Configuration Guide
Requirements for hardware-assisted takeover
The hardware-assisted takeover feature is available only on systems with an RLM or SP module
configured for remote management. Remote management provides remote platform management
capabilities, including remote access, monitoring, troubleshooting, logging, and alerting features.
Although a system with remote management on both nodes provides hardware-assisted takeover for
both, hardware-assisted takeover is also supported on HA pairs in which only one of the two systems
has remote management configured. Remote management does not have to be configured on both
nodes in the HA pair. Remote management can detect failures on the system in which it is installed
and provide faster takeover times if a failure occurs on the system with remote management.
See the Data ONTAP 7-Mode System Administration Guide for information about setting up remote
management.
Possible storage configurations in the HA pairs
HA pairs can be configured symmetrically, asymmetrically, as an active/passive pair, or with shared
disk shelf stacks.
Symmetrical
configurations
In a symmetrical HA pair, each node has the same amount of storage.
Asymmetrical
configurations
In an asymmetrical standard HA pair, one node has more storage than the
other. This is supported as long as neither node exceeds the maximum
capacity limit for the node.
Active/passive
configurations
In this configuration, the passive node has only a root volume, and the
active node has all the remaining storage and services all data requests
during normal operation. The passive node responds to data requests only if
it has taken over the active node.
Shared loops or
stacks
You can share a loop or stack between the two nodes. This is particularly
useful for active/passive configurations, as described in the preceding
bullet.
HA pairs and storage system model types
Different model storage systems support some different HA configurations. This includes the
physical configuration of the HA pair and the manner in which the system recognizes that it is in an
HA pair.
Note: The physical configuration of the HA pair does not affect the cluster cabling of the nodes in
the HA pair.
Planning your HA pair configuration | 37
Single-chassis and dual-chassis HA pairs
Depending on the model of the storage system, an HA pair can consist of two controllers in a single
chassis, or two controllers in two separate chassis. Some models can be configured either way, while
other models can be configured only as a single-chassis HA pair or dual-chassis HA pair.
The following example shows a single-chassis HA pair:
In a single-chassis HA pair, both controllers are in the same chassis. The HA interconnect is provided
by the internal backplane. No external HA interconnect cabling is required.
The following example shows a dual-chassis HA pair and the HA interconnect cables:
In a dual-chassis HA pair, the controllers are in separate chassis. The HA interconnect is provided by
external cabling.
Interconnect cabling for systems with variable HA configurations
In systems that can be configured either as a single-chassis or dual-chassis HA pair, the interconnect
cabling is different depending on the configuration.
The following table describes the interconnect cabling for 32xx and 62xx systems:
If the controller modules in
the HA pair are...
The HA interconnect cabling is...
Both in the same chassis
Not required. An internal interconnect is used.
38 | High Availability and MetroCluster Configuration Guide
If the controller modules in
the HA pair are...
The HA interconnect cabling is...
Each in a separate chassis
Required.
HA configuration and the HA state PROM value
Some controller modules and chassis automatically record in a PROM whether they are in an HA
pair or stand-alone. This record is the HA state and must be the same on all components within the
stand-alone system or HA pair. The HA state can be manually configured if necessary.
Related tasks
Verifying and setting the HA state on controller modules and chassis on page 80
Table of storage system models and HA configuration differences
The supported storage systems have key differences in their HA configuration, depending on the
model.
The following table lists the supported storage systems and their HA configuration differences:
Storage system model HA configuration
(single-chassis, dualchassis, or either)
Interconnect type
(internal InfiniBand,
external InfiniBand,
or external 10-Gb
Ethernet)
Uses HA state PROM
value?
6290
Single-chassis or dualchassis
Dual-chassis: External Yes
InfiniBand using
NVRAM adapter
Single-chassis: Internal
InfiniBand
6280
Single-chassis or dualchassis
Dual-chassis: External Yes
InfiniBand using
NVRAM adapter
Single-chassis: Internal
InfiniBand
6250
Single-chassis or dualchassis
Dual-chassis: External Yes
InfiniBand using
NVRAM adapter
Single-chassis: Internal
InfiniBand
Planning your HA pair configuration | 39
Storage system model HA configuration
(single-chassis, dualchassis, or either)
Interconnect type
(internal InfiniBand,
external InfiniBand,
or external 10-Gb
Ethernet)
Uses HA state PROM
value?
6240
Single-chassis or dualchassis
Dual-chassis: External Yes
InfiniBand using
NVRAM adapter
Single-chassis: Internal
InfiniBand
6220
Single-chassis
Internal InfiniBand
Yes
6210
Single-chassis
Internal InfiniBand
Yes
60xx
Dual-chassis
External InfiniBand
No
using NVRAM adapter
3270
Single-chassis or dualchassis
Dual-chassis: External Yes
10-Gb Ethernet using
onboard ports c0a and
c0b
These ports are
dedicated HA
interconnect ports.
Regardless of the
system configuration,
these ports cannot be
used for data or other
purposes.
Single-chassis: Internal
InfiniBand
3250
Dual-chassis
External 10-Gb
Yes
Ethernet using onboard
ports c0a and c0b
These ports are
dedicated HA
interconnect ports.
Regardless of the
system configuration,
these ports cannot be
used for data or other
purposes.
40 | High Availability and MetroCluster Configuration Guide
Storage system model HA configuration
(single-chassis, dualchassis, or either)
Interconnect type
(internal InfiniBand,
external InfiniBand,
or external 10-Gb
Ethernet)
Uses HA state PROM
value?
3240
Single-chassis or dualchassis
Dual-chassis: External Yes
10-Gb Ethernet using
onboard ports c0a and
c0b
These ports are
dedicated HA
interconnect ports.
Regardless of the
system configuration,
these ports cannot be
used for data or other
purposes.
Single-chassis: Internal
InfiniBand
3220
Single-chassis or dualchassis
Dual-chassis: External Yes
10-Gb Ethernet using
onboard ports c0a and
c0b
These ports are
dedicated HA
interconnect ports.
Regardless of the
system configuration,
these ports cannot be
used for data or other
purposes.
Single-chassis: Internal
InfiniBand
3210
Single-chassis
Internal Infiniband
Yes
31xx
Single-chassis
Internal InfiniBand
No
FAS22xx
Single-chassis
Internal InfiniBand
Yes
41
Installing and cabling an HA pair
To install and cable a new standard or mirrored HA pair, you must have the correct tools and
equipment and you must connect the controllers to the disk shelves (for FAS systems or V-Series
systems using native disk shelves). If it is a dual-chassis HA pair, you must also cable the HA
interconnect between the nodes. HA pairs can be installed in either NetApp system cabinets or in
equipment racks.
The specific procedure you use depends on the following aspects of your configuration:
•
•
Whether you have a standard or mirrored HA pair
Whether you are using FC or SAS disk shelves
Note: If your configuration includes SAS disk shelves, see the Universal SAS and ACP
Cabling Guide on the NetApp Support Site at support.netapp.com for information about
cabling. For cabling the HA interconnect between the nodes, use the procedures in this guide.
Multipath HA is required on all HA pairs except for some FAS22xx system configurations, which
use single-path HA and lack the redundant standby connections.
System cabinet or equipment rack installation
You need to install your HA pair in one or more NetApp system cabinets or in standard telco
equipment racks. Each of these options has different requirements.
HA pairs in an equipment rack
Depending on the amount of storage you ordered, you need to install the equipment in one or more
telco-style equipment racks.
The equipment racks can hold one or two nodes on the bottom and eight or more disk shelves. For
information about how to install the disk shelves and nodes into the equipment racks, see the
appropriate documentation that came with your equipment.
HA pairs in a system cabinet
Depending on the number of disk shelves, the HA pair you ordered arrives in a single system cabinet
or multiple system cabinets.
The number of system cabinets you receive depends on how much storage you ordered. All internal
adapters, such as networking adapters, Fibre Channel adapters, and other adapters, arrive preinstalled
in the nodes.
If it comes in a single system cabinet, both the Channel A and Channel B disk shelves are cabled, and
the HA adapters are also pre-cabled.
42 | High Availability and MetroCluster Configuration Guide
If the HA pair you ordered has more than one cabinet, you must complete the cabling by cabling the
local node to the partner node’s disk shelves and the partner node to the local node’s disk shelves.
You must also cable the nodes together by cabling the NVRAM HA interconnects. If the HA pair
uses switches, you must install the switches, as described in the accompanying switch
documentation. The system cabinets might also need to be connected to each other. See your System
Cabinet Guide for information about connecting your system cabinets together.
Required documentation
Installation of an HA pair requires the correct documentation.
The following table lists and briefly describes the documentation you might need to refer to when
preparing a new HA pair, or converting two stand-alone systems into an HA pair:
Manual name
Description
Hardware Universe
This guide describes the physical requirements
that your site must meet to install NetApp
equipment.
The appropriate system cabinet guide
This guide describes how to install NetApp
equipment into a system cabinet.
The appropriate disk shelf guide
These guides describe how to cable a disk shelf
to a storage system.
The appropriate hardware documentation for
your storage system model
These guides describe how to install the storage
system, connect it to a network, and bring it up
for the first time.
Diagnostics Guide
This guide describes the diagnostics tests that
you can run on the storage system.
Data ONTAP Network Management Guide for
7-Mode
This guide describes how to perform network
configuration for the storage system.
Data ONTAP Upgrade and Revert/Downgrade
Guide for 7-Mode
This guide describes how to upgrade storage
system and disk firmware, and how to upgrade
storage system software.
Data ONTAP Data Protection Online Backup
and Recovery Guide for 7-Mode
This guide describes, among other topics,
SyncMirror technology, which is used for
mirrored HA pairs.
Data ONTAP System Administration Guide for
7-Mode
This guide describes general storage system
administration.
Installing and cabling an HA pair | 43
Manual name
Description
Data ONTAP Software Setup Guide for 7-Mode This guide describes how to configure the
software of a new storage system for the first
time.
Note: If you are installing a V-Series HA pair with third-party storage, see the V-Series
Installation Requirements and Reference Guide for information about cabling V-Series systems to
storage arrays, and see the V-Series Implementation Guide for Third-Party Storage for information
about configuring storage arrays to work with V-Series systems.
Related information
Data ONTAP Information Library
Required tools
Installation of an HA pair requires the correct tools.
The following list specifies the tools you need to install the HA pair:
•
•
•
#1 and #2 Phillips screwdrivers
Hand level
Marker
Required equipment
When you receive your HA pair, you should receive the equipment listed in the following table. See
the Hardware Universe (formerly the System Configuration Guide) at support.netapp.com/
knowledge/docs/hardware/NetApp/syscfg/index.shtml to confirm your storage-system type, storage
capacity, and so on.
Required equipment
Details
Storage system
Two of the same type of storage systems
Storage
See the Hardware Universe (formerly the
System Configuration Guide) at
support.netapp.com/knowledge/docs/hardware/
NetApp/syscfg/index.shtml
44 | High Availability and MetroCluster Configuration Guide
Required equipment
Details
HA interconnect adapter (for controller modules InfiniBand (IB) HA adapter
that do not share a chassis)
(The NVRAM adapter functions as the HA
interconnect adapter on FAS900 series and later
Note: When 32xx systems are in a dualstorage systems, except the 32xx systems)
chassis HA pair, the c0a and c0b 10-GbE
ports are the HA interconnect ports. They do
not require an HA interconnect adapter.
Regardless of configuration, the 32xx
system's c0a and c0b ports cannot be used for
data. They are only for the HA interconnect.
For DS14mk2 disk shelves: FC-AL or FC HBA
(FC HBA for Disk) adapters
For SAS disk shelves: SAS HBAs, if applicable
Minimum of two FC-AL adapters or two SAS
HBAs
Fibre Channel switches
N/A
SFP (Small Form Pluggable) modules
N/A
NVRAM HA adapter media converter
Only if using fiber cabling
Cables (provided with shipment unless
otherwise noted)
•
•
•
One optical controller-to-disk shelf cable per
loop
Multiple disk shelf-to-disk shelf cables
Two 4xIB copper cables, or two 4xIB optical
cables
Note: You must purchase longer optical
cables separately for cabling distances
greater than 30 meters.
•
•
Two optical cables with media converters for
systems using the IB HA adapter
The 32xx systems, when in a dual-chassis
HA pair, require 10 GbE cables (Twinax or
SR) for the HA interconnect.
Installing and cabling an HA pair | 45
Preparing your equipment
You must install your nodes in your system cabinets or equipment racks, depending on your
installation type.
Installing the nodes in equipment racks
Before you cable your nodes together, you install the nodes and disk shelves in the equipment rack,
label the disk shelves, and connect the nodes to the network.
Steps
1. Install the nodes in the equipment rack, as described in the guide for your disk shelf, hardware
documentation, or Quick Start guide that came with your equipment.
2. Install the disk shelves in the equipment rack, as described in the appropriate disk shelf guide.
3. Label the interfaces, where appropriate.
4. Connect the nodes to the network, as described in the setup instructions for your system.
Result
The nodes are now in place and connected to the network and power is available.
After you finish
Proceed to cable the HA pair.
Installing the nodes in a system cabinet
Before you cable your nodes together, you must install the system cabinet, nodes, and any disk
shelves, and connect the nodes to the network. If you have two cabinets, the cabinets must be
connected together.
Steps
1. Install the system cabinets, nodes, and disk shelves as described in the System Cabinet Guide.
If you have multiple system cabinets, remove the front and rear doors and any side panels that
need to be removed, and connect the system cabinets together.
2. Connect the nodes to the network, as described in the Installation and Setup Instructions for your
system.
3. Connect the system cabinets to an appropriate power source and apply power to the cabinets.
Result
The nodes are now in place and connected to the network, and power is available.
46 | High Availability and MetroCluster Configuration Guide
After you finish
Proceed to cable the HA pair.
Cabling a standard HA pair
To cable a standard HA pair, you identify the ports you need to use on each node, then you cable the
ports, and then you cable the HA interconnect.
About this task
This procedure explains how to cable a configuration using DS14mk2 or DS14mk4 disk shelves.
For cabling SAS disk shelves in an HA pair, see the Universal SAS and ACP Cabling Guide.
Note: If you are installing an HA pair between V-Series systems using array LUNs, see the VSeries Installation Requirements and Reference Guide for information about cabling V-Series
systems to storage arrays. See the V-Series Implementation Guide for Third-Party Storage for
information about configuring storage arrays to work with Data ONTAP.
The sections for cabling the HA interconnect apply to all systems regardless of disk shelf type.
Steps
1. Determining which Fibre Channel ports to use for Fibre Channel disk shelf connections on page
46
2. Cabling Node A to DS14mk2 or DS14mk4 disk shelves on page 47
3. Cabling Node B to DS14mk2 or DS14mk4 disk shelves on page 49
4. Cabling the HA interconnect (all systems except 32xx) on page 51
5. Cabling the HA interconnect (32xx systems in separate chassis) on page 52
Determining which Fibre Channel ports to use for Fibre Channel disk shelf
connections
Before cabling your HA pair, you need to identify which Fibre Channel ports to use to connect your
disk shelves to each storage system, and in what order to connect them.
Keep the following guidelines in mind when identifying ports to use:
•
•
•
•
Every disk shelf loop in the HA pair requires two ports on the node, one for the primary
connection and one for the redundant multipath HA connection.
A standard HA pair with one loop for each node uses four ports on each node.
Onboard Fibre Channel ports should be used before using ports on expansion adapters.
Always use the expansion slots in the order shown in the Hardware Universe (formerly the
System Configuration Guide) at support.netapp.com/knowledge/docs/hardware/NetApp/syscfg/
index.shtml for your platform for an HA pair.
If using Fibre Channel HBAs, insert the adapters in the same slots on both systems.
Installing and cabling an HA pair | 47
See the Hardware Universe (formerly the System Configuration Guide) at support.netapp.com/
knowledge/docs/hardware/NetApp/syscfg/index.shtml to obtain all slot assignment information for
the various adapters you use to cable your HA pair.
After identifying the ports, you should have a numbered list of Fibre Channel ports for both nodes,
starting with Port 1.
Cabling guidelines for a quad-port Fibre Channel HBA
If using ports on the quad-port, 4-Gb Fibre Channel HBAs, use the procedures in the following
sections, with the following additional guidelines:
•
•
•
•
•
•
Disk shelf loops using ESH4 modules must be cabled to the quad-port HBA first.
Disk shelf loops using AT-FCX modules must be cabled to dual-port HBA ports or onboard ports
before using ports on the quad-port HBA.
Port A of the HBA must be cabled to the In port of Channel A of the first disk shelf in the loop.
Port A of the partner node's HBA must be cabled to the In port of Channel B of the first disk shelf
in the loop. This ensures that disk names are the same for both nodes.
Additional disk shelf loops must be cabled sequentially with the HBA’s ports.
Port A is used for the first loop, port B for the second loop, and so on.
If available, ports C or D must be used for the redundant multipath HA connection after cabling
all remaining disk shelf loops.
All other cabling rules described in the documentation for the HBA and the Hardware Universe
must be observed.
Cabling Node A to DS14mk2 or DS14mk4 disk shelves
To cable Node A, you must use the Fibre Channel ports you previously identified and cable the disk
shelf loops owned by the node to these ports.
About this task
•
•
This procedure uses multipath HA, which is required on all systems.
This procedure does not apply to SAS disk shelves.
For cabling SAS disk shelves in an HA pair, see the Universal SAS and ACP Cabling Guide.
Note: You can find additional cabling diagrams in your system's Installation and Setup
Instructions on the NetApp Support Site.
Steps
1. Review the cabling diagram before proceeding to the cabling steps.
•
•
The circled numbers in the diagram correspond to the step numbers in the procedure.
The location of the Input and Output ports on the disk shelves vary depending on the disk
shelf models.
Make sure that you refer to the labeling on the disk shelf rather than to the location of the port
shown in the diagram.
48 | High Availability and MetroCluster Configuration Guide
•
•
•
The location of the Fibre Channel ports on the controllers is not representative of any
particular storage system model; determine the locations of the ports you are using in your
configuration by inspection or by using the Installation and Setup Instructions for your model.
The port numbers refer to the list of Fibre Channel ports you created.
The diagram only shows one loop per node and one disk shelf per loop.
Your installation might have more loops, more disk shelves, or different numbers of disk
shelves between nodes.
2. Cable Fibre Channel port A1 of Node A to the Channel A Input port of the first disk shelf of
Node A loop 1.
3. Cable the Node A disk shelf Channel A Output port to the Channel A Input port of the next disk
shelf in loop 1.
4. Repeat step 3 for any remaining disk shelves in loop 1.
5. Cable the Channel A Output port of the last disk shelf in the loop to Fibre Channel port B4 of
Node B.
This provides the redundant multipath HA connection for Channel A.
6. Cable Fibre Channel port A2 of Node A to the Channel B Input port of the first disk shelf of
Node B loop 1.
7. Cable the Node B disk shelf Channel B Output port to the Channel B Input port of the next disk
shelf in loop 1.
8. Repeat step 7 for any remaining disk shelves in loop 1.
9. Cable the Channel B Output port of the last disk shelf in the loop to Fibre Channel port B3 of
Node B.
This provides the redundant multipath HA connection for Channel B.
Installing and cabling an HA pair | 49
10. Repeat steps 2 to 9 for each pair of loops in the HA pair, using ports 3 and 4 for the next loop,
ports 5 and 6 for the next one, and so on.
Result
Node A is completely cabled.
After you finish
Proceed to cabling Node B.
Cabling Node B to DS14mk2 or DS14mk4 disk shelves
To cable Node B, you must use the Fibre Channel ports you previously identified and cable the disk
shelf loops owned by the node to these ports.
About this task
•
•
This procedure uses multipath HA, required on all systems.
This procedure does not apply to SAS disk shelves.
For cabling SAS disk shelves in an HA pair, see the Universal SAS and ACP Cabling Guide.
Note: You can find additional cabling diagrams in your system's Installation and Setup
Instructions on the NetApp Support Site at support.netapp.com.
Steps
1. Review the cabling diagram before proceeding to the cabling steps.
•
•
•
•
•
The circled numbers in the diagram correspond to the step numbers in the procedure.
The location of the Input and Output ports on the disk shelves vary depending on the disk
shelf models.
Make sure that you refer to the labeling on the disk shelf rather than to the location of the port
shown in the diagram.
The location of the Fibre Channel ports on the controllers is not representative of any
particular storage system model; determine the locations of the ports you are using in your
configuration by inspection or by using the Installation and Setup Instructions for your model.
The port numbers refer to the list of Fibre Channel ports you created.
The diagram only shows one loop per node and one disk shelf per loop.
Your installation might have more loops, more disk shelves, or different numbers of disk
shelves between nodes.
50 | High Availability and MetroCluster Configuration Guide
2. Cable Port B1 of Node B to the Channel B Input port of the first disk shelf of Node A loop 1.
Both channels of this disk shelf are connected to the same port on each node. This is not required,
but it makes your HA pair easier to administer because the disks have the same ID on each node.
This is true for Step 5 also.
3. Cable the disk shelf Channel B Output port to the Channel B Input port of the next disk shelf in
loop 1.
4. Repeat step 3 for any remaining disk shelves in loop 1.
5. Cable the Channel B Output port of the last disk shelf in the loop to Fibre Channel port A4 of
Node A.
This provides the redundant multipath HA connection for Channel B.
6. Cable Fibre Channel port B2 of Node B to the Channel A Input port of the first disk shelf of Node
B loop 1.
7. Cable the disk shelf Channel A Output port to the Channel A Input port of the next disk shelf in
loop 1.
8. Repeat step 7 for any remaining disk shelves in loop 1.
9. Cable the Channel A Output port of the last disk shelf in the loop to Fibre Channel port A3 of
Node A.
This provides the redundant multipath HA connection for Channel A.
10. Repeat steps 2 to 9 for each pair of loops in the HA pair, using ports 3 and 4 for the next loop,
ports 5 and 6 for the next one, and so on.
Installing and cabling an HA pair | 51
Result
Node B is completely cabled.
After you finish
Proceed to cable the HA interconnect.
Cabling the HA interconnect (all systems except 32xx)
To cable the HA interconnect between the HA pair nodes, you must make sure that your interconnect
adapter is in the correct slot and connect the adapters on each node with the optical cable.
About this task
This procedure applies to all dual-chassis HA pairs (HA pairs in which the two controller modules
reside in separate chassis) except 32xx systems, regardless of disk shelf type.
Steps
1. See the Hardware Universe (formerly the System Configuration Guide) at support.netapp.com/
knowledge/docs/hardware/NetApp/syscfg/index.shtml to ensure that your interconnect adapter is
in the correct slot for your system in an HA pair.
For systems that use an NVRAM adapter, the NVRAM adapter functions as the HA interconnect
adapter.
2. Plug one end of the optical cable into one of the local node's HA adapter ports, then plug the
other end into the partner node's corresponding adapter port.
You must not cross-cable the HA interconnect adapter. Cable the local node ports only to the
identical ports on the partner node.
If the system detects a cross-cabled HA interconnect, the following message appears:
HA interconnect port <port> of this appliance seems to be connected to
port <port> on the partner appliance.
3. Repeat Step 2 for the two remaining ports on the HA adapters.
Result
The nodes are connected to each other.
After you finish
Proceed to configure the system.
52 | High Availability and MetroCluster Configuration Guide
Cabling the HA interconnect (32xx systems in separate chassis)
To enable the HA interconnect between 32xx controller modules that reside in separate chassis, you
must cable the onboard 10-GbE ports on one controller module to the onboard GbE ports on the
partner.
About this task
This procedure applies to 32xx systems regardless of the type of attached disk shelves.
Steps
1. Plug one end of the 10-GbE cable to the c0a port on one controller module.
2. Plug the other end of the 10-GbE cable to the c0a port on the partner controller module.
3. Repeat the preceding steps to connect the c0b ports.
Do not cross-cable the HA interconnect adapter; cable the local node ports only to the identical
ports on the partner node.
Result
The nodes are connected to each other.
After you finish
Proceed to configure the system.
Cabling a mirrored HA pair
To cable a mirrored HA pair, you identify the ports you need to use on each node, then you cable the
ports, and then you cable the HA interconnect.
About this task
This procedure explains how to cable a configuration using DS14mk2 disk shelves.
For cabling SAS disk shelves in an HA pair, see the Universal SAS and ACP Cabling Guide.
Note: If you are installing an HA pair between V-Series systems using array LUNs, see the V-
Series Installation Requirements and Reference Guide for information about cabling V-Series
systems to storage arrays. See the V-Series Implementation Guide for Third-Party Storage for
information about configuring storage arrays to work with Data ONTAP.
The sections for cabling the HA interconnect apply to all systems regardless of disk shelf type.
Installing and cabling an HA pair | 53
Steps
1. Determining which Fibre Channel ports to use for Fibre Channel disk shelf connections on page
53
2. Creating your port list for mirrored HA pairs on page 54
3. Cabling the Channel A DS14mk2 disk shelf loops on page 55
4. Cabling the Channel B DS14mk2 or DS14mk4 disk shelf loops on page 57
5. Cabling the redundant multipath HA connection for each loop on page 59
6. Cabling the HA interconnect (all systems except 32xx) on page 61
7. Cabling the HA interconnect (32xx systems in separate chassis) on page 61
Determining which Fibre Channel ports to use for Fibre Channel disk shelf
connections
Before cabling your HA pair, you need to identify which Fibre Channel ports to use to connect your
disk shelves to each storage system, and in what order to connect them.
Keep the following guidelines in mind when identifying ports to use:
•
•
•
•
Every disk shelf loop in the HA pair requires two ports on the node, one for the primary
connection and one for the redundant multipath HA connection.
A standard HA pair with one loop for each node uses four ports on each node.
Onboard Fibre Channel ports should be used before using ports on expansion adapters.
Always use the expansion slots in the order shown in the Hardware Universe (formerly the
System Configuration Guide) at support.netapp.com/knowledge/docs/hardware/NetApp/syscfg/
index.shtml for your platform for an HA pair.
If using Fibre Channel HBAs, insert the adapters in the same slots on both systems.
See the Hardware Universe (formerly the System Configuration Guide) at support.netapp.com/
knowledge/docs/hardware/NetApp/syscfg/index.shtml to obtain all slot assignment information for
the various adapters you use to cable your HA pair.
After identifying the ports, you should have a numbered list of Fibre Channel ports for both nodes,
starting with Port 1.
Cabling guidelines for a quad-port Fibre Channel HBA
If using ports on the quad-port, 4-Gb Fibre Channel HBAs, use the procedures in the following
sections, with the following additional guidelines:
•
•
•
•
Disk shelf loops using ESH4 modules must be cabled to the quad-port HBA first.
Disk shelf loops using AT-FCX modules must be cabled to dual-port HBA ports or onboard ports
before using ports on the quad-port HBA.
Port A of the HBA must be cabled to the In port of Channel A of the first disk shelf in the loop.
Port A of the partner node's HBA must be cabled to the In port of Channel B of the first disk shelf
in the loop. This ensures that disk names are the same for both nodes.
Additional disk shelf loops must be cabled sequentially with the HBA’s ports.
54 | High Availability and MetroCluster Configuration Guide
•
•
Port A is used for the first loop, port B for the second loop, and so on.
If available, ports C or D must be used for the redundant multipath HA connection after cabling
all remaining disk shelf loops.
All other cabling rules described in the documentation for the HBA and the Hardware Universe
must be observed.
Creating your port list for mirrored HA pairs
After you determine which Fibre Channel ports to use, you can create a table identifying which ports
belong to which port pool.
About this task
Mirrored HA pairs, regardless of disk shelf type, use SyncMirror to separate each aggregate into two
plexes that mirror each other. One plex uses disks in pool 0 and the other plex uses disks in pool 1.
You must assign disks to the pools appropriately.
Follow the guidelines for software-based disk ownership in the Data ONTAP Storage Management
Guide for 7-Mode.
For more information about SyncMirror, see the Data ONTAP Data Protection Online Backup and
Recovery Guide for 7-Mode.
Step
1. Create a table specifying the port usage; the cabling diagrams in this document use the notation
“P1-3” (the third port for pool 1).
For a 32xx HA pair that has two mirrored loops, the port list might look like the following
example:
Pool 0
Pool 1
P0-1: onboard port 0a
P1-1: onboard port 0c
P0-2: onboard port 0b
P1-2: onboard port 0d
P0-3: slot 2 port A
P1-3: slot 4 port A
P0-4: slot 2 port B
P1-4: slot 4 port B
After you finish
Proceed to cable the Channel A loops.
Installing and cabling an HA pair | 55
Cabling the Channel A DS14mk2 disk shelf loops
To begin cabling the disk shelves, you must connect the appropriate pool ports on the node to the
Channel A modules of the disk shelf stack for the pool.
About this task
•
•
This procedure uses multipath HA, which is required on all systems.
This procedure does not apply to SAS disk shelves.
For cabling SAS disk shelves in an HA pair, see the Universal SAS and ACP Cabling Guide.
Steps
1. Complete your port list.
2. Review the cabling diagram before proceeding to the cabling steps.
•
•
•
•
•
The circled numbers in the diagram correspond to the step numbers in the procedure.
The location of the Input and Output ports on the disk shelves vary depending on the disk
shelf models.
Make sure that you refer to the labeling on the disk shelf rather than to the location of the port
shown in the diagram.
The location of the Fibre Channel ports on the controllers is not representative of any
particular storage system model; determine the locations of the ports you are using in your
configuration by inspection or by using the Installation and Setup Instructions for your model.
The port numbers refer to the list of Fibre Channel ports you created.
The diagram only shows one loop per node and one disk shelf per loop.
Your installation might have more loops, more disk shelves, or different numbers of disk
shelves between nodes.
56 | High Availability and MetroCluster Configuration Guide
3. Cable Channel A for Node A.
a) Cable the first port for pool 0 (P0-1) of Node A to the first Node A disk shelf Channel A Input
port of disk shelf pool 0.
b) Cable the first port for pool 1 (P1-1) of Node A to the first Node A disk shelf Channel A Input
port of disk shelf pool 1.
c) Cable the disk shelf Channel A Output port to the next disk shelf Channel A Input port in the
loop for both disk pools.
Note: The illustration shows only one disk shelf per disk pool. The number of disk shelves
per pool might be different for your configuration.
Installing and cabling an HA pair | 57
d) Repeat substep 3c, connecting the next Channel A output to the next disk shelf Channel A
Input port for any remaining disk shelves in this loop for each disk pool.
e) Repeat substep 3a through substep 3d for any additional loops for Channel A, Node A, using
the odd-numbered port numbers (P0-3 and P1-3, P0-5, and P1-5, and so on).
4. Cable Channel A for Node B.
a) Cable the second port for pool 0 (P0-2) of Node B to the first Node B disk shelf Channel A
Input port of disk shelf pool 0.
b) Cable the second port for pool 1 (P1-2) of Node B to the first Node B disk shelf Channel A
Input port of disk shelf pool 1.
c) Cable the disk shelf Channel A Output port to the next disk shelf Channel A Input port in the
loop for both disk pools.
d) Repeat substep 4c, connecting Channel A output to input, for any remaining disk shelves in
each disk pool.
e) Repeat substep 4a through substep 4d for any additional loops on Channel A, Node B, using
the even-numbered port numbers (P0-4 and P1-4, P0-6, and P1-6, and so on).
After you finish
Proceed to cable the Channel B loops.
Cabling the Channel B DS14mk2 or DS14mk4 disk shelf loops
To provide the mirrored storage, you cable the mirrored pool ports on the node to the Channel B
modules of the appropriate disk shelf stack.
Steps
1. Review the cabling diagram before proceeding to the cabling steps.
•
•
•
•
•
The circled numbers in the diagram correspond to the step numbers in the procedure.
The location of the Input and Output ports on the disk shelves vary depending on the disk
shelf models.
Make sure that you refer to the labeling on the disk shelf rather than to the location of the port
shown in the diagram.
The location of the Fibre Channel ports on the controllers is not representative of any
particular storage system model; determine the locations of the ports you are using in your
configuration by inspection or by using the Installation and Setup Instructions for your model.
The port numbers refer to the list of Fibre Channel ports you created.
The diagram only shows one loop per node and one disk shelf per loop.
Your installation might have more loops, more disk shelves, or different numbers of disk
shelves between nodes.
58 | High Availability and MetroCluster Configuration Guide
2. Cable Channel B for Node A.
a) Cable the second port for pool 0 (P0-2) of Node A to the first Node B disk shelf Channel B
Input port of disk shelf pool 0.
Note: Both channels of this disk shelf are connected to the same port on each node. This is
not required, but it makes your HA pair easier to administer because the disks have the
same ID on each node.
b) Cable the second port for pool 1 (P1-2) of Node A to the first Node B disk shelf Channel B
Input port of disk shelf pool 1.
Installing and cabling an HA pair | 59
c) Cable the disk shelf Channel B Output port to the next disk shelf Channel B Input port in the
loop for both disk pools.
Note: The illustration shows only one disk shelf per disk pool. The number of disk shelves
per pool might be different for your configuration.
d) Repeat Substep 2c, connecting Channel B output to input, for any remaining disk shelves in
each disk pool.
e) Repeat Substep 2a through Substep 2d for any additional loops on Channel B, Node A, using
the even numbered port numbers (P0-4 and P1-4, P0-6, and P1-6, and so on).
3. Cable Channel B for Node B.
a) Cable the first port for pool 0 (P0-1) of Node B to the first Node A disk shelf Channel B Input
port of disk shelf pool 0.
b) Cable the first port for pool 1 (P1-1) of Node B to the first Node A disk shelf Channel B Input
port of disk shelf pool 1.
c) Cable the disk shelf Channel B Output port to the next disk shelf Channel B Input port in the
loop for both disk pools.
d) Repeat Substep 3c, connecting Channel B output to input, for any remaining disk shelves in
each disk pool.
e) Repeat Substep 3a through Substep 3d for any additional loops for Channel B, Node B, using
the odd numbered port numbers (P0-3 and P1-3, P0-5, and P1-5, and so on).
After you finish
Proceed to cable the HA interconnect.
Cabling the redundant multipath HA connection for each loop
To complete the multipath HA cabling for the disk shelves, you must add the final connection for
each channel on the final disk shelf in each loop.
Steps
1. Review the cabling diagram before proceeding to the cabling steps.
•
•
•
•
•
The circled numbers in the diagram correspond to the step numbers in the procedure.
The location of the Input and Output ports on the disk shelves vary depending on the disk
shelf models.
Make sure that you refer to the labeling on the disk shelf rather than to the location of the port
shown in the diagram.
The location of the Fibre Channel ports on the controllers is not representative of any
particular storage system model; determine the locations of the ports you are using in your
configuration by inspection or by using the Installation and Setup Instructions for your model.
The port numbers refer to the list of Fibre Channel ports you created.
The diagram only shows one loop per node and one disk shelf per loop.
Your installation might have more loops, more disk shelves, or different numbers of disk
shelves between nodes.
60 | High Availability and MetroCluster Configuration Guide
2. Connect the Channel A output port on the last disk shelf for each loop belonging to Node A to an
available port on Node B in the same pool.
3. Connect the Channel B output port on the last disk shelf for each loop belonging to Node A to an
available port on Node B in the same pool.
4. Connect the Channel A output port on the last disk shelf for each loop belonging to Node B to an
available port on Node B in the same pool.
5. Connect the Channel B output port on the last disk shelf for each loop belonging to Node B to an
available port on Node B in the same pool.
Installing and cabling an HA pair | 61
Cabling the HA interconnect (all systems except 32xx)
To cable the HA interconnect between the HA pair nodes, you must make sure that your interconnect
adapter is in the correct slot and connect the adapters on each node with the optical cable.
About this task
This procedure applies to all dual-chassis HA pairs (HA pairs in which the two controller modules
reside in separate chassis) except 32xx systems, regardless of disk shelf type.
Steps
1. See the Hardware Universe (formerly the System Configuration Guide) at support.netapp.com/
knowledge/docs/hardware/NetApp/syscfg/index.shtml to ensure that your interconnect adapter is
in the correct slot for your system in an HA pair.
For systems that use an NVRAM adapter, the NVRAM adapter functions as the HA interconnect
adapter.
2. Plug one end of the optical cable into one of the local node's HA adapter ports, then plug the
other end into the partner node's corresponding adapter port.
You must not cross-cable the HA interconnect adapter. Cable the local node ports only to the
identical ports on the partner node.
If the system detects a cross-cabled HA interconnect, the following message appears:
HA interconnect port <port> of this appliance seems to be connected to
port <port> on the partner appliance.
3. Repeat Step 2 for the two remaining ports on the HA adapters.
Result
The nodes are connected to each other.
After you finish
Proceed to configure the system.
Cabling the HA interconnect (32xx systems in separate chassis)
To enable the HA interconnect between 32xx controller modules that reside in separate chassis, you
must cable the onboard 10-GbE ports on one controller module to the onboard GbE ports on the
partner.
About this task
This procedure applies to 32xx systems regardless of the type of attached disk shelves.
62 | High Availability and MetroCluster Configuration Guide
Steps
1. Plug one end of the 10-GbE cable to the c0a port on one controller module.
2. Plug the other end of the 10-GbE cable to the c0a port on the partner controller module.
3. Repeat the preceding steps to connect the c0b ports.
Do not cross-cable the HA interconnect adapter; cable the local node ports only to the identical
ports on the partner node.
Result
The nodes are connected to each other.
After you finish
Proceed to configure the system.
Required connections for using uninterruptible power
supplies with standard or mirrored HA pairs
You can use a UPS (uninterruptible power supply) with your HA pair. The UPS enables the system
to fail over gracefully if power fails for one of the nodes, or to shut down gracefully if power fails for
both nodes. You must ensure that the correct equipment is connected to the UPS.
To gain the full benefit of the UPS, you must ensure that all the required equipment is connected to
the UPS. The equipment that needs to be connected depends on whether your configuration is a
standard or a mirrored HA pair.
For a standard HA pair, you must connect the controller, disks, and any FC switches in use.
For a mirrored HA pair, you must connect the controller and any FC switches to the UPS, as for a
standard HA pair. However, if the two sets of disk shelves have separate power sources, you do not
have to connect the disks to the UPS. If power is interrupted to the local controller and disks, the
controller can access the remote disks until it shuts down gracefully or the power supply is restored.
In this case, if power is interrupted to both sets of disks at the same time, the HA pair cannot shut
down gracefully.
63
Reconfiguring an HA pair into two stand-alone
systems
To divide an HA pair so that the nodes become stand-alone systems without redundancy, you must
disable the HA software features and then remove the hardware connections.
About this task
This procedure applies to all HA pairs regardless of disk shelf type.
Steps
1.
2.
3.
4.
5.
6.
Ensuring uniform disk ownership within disk shelves and loops in the system on page 63
Disabling controller failover on page 64
Configuring a node for non-HA (stand-alone) use on page 65
Reconfiguring nodes using disk shelves for stand-alone operation on page 67
Requirements when changing a node using array LUNs to stand-alone on page 69
Reconfiguring nodes using array LUNs for stand-alone operation on page 70
Ensuring uniform disk ownership within disk shelves and
loops in the system
If a disk shelf or loop contains a mix of disks owned by Node A and Node B, you must use this
procedure to move the data and make disk ownership uniform within the disk shelf or loop.
Before you begin
You must ensure the following:
•
•
Disk ownership is uniform within all disk shelves and loops in the system
All the disks within a disk shelf or loop belong to a single node and pool
About this task
Note: It is a best practice to always assign all disks on the same loop to the same node and pool.
Steps
1. Use the following command to identify any disk shelves or loops that contain both disks
belonging to Node A and disks belonging to Node B:
disk show -v
64 | High Availability and MetroCluster Configuration Guide
2. Determine which node the disk shelf or loop with mixed ownership will be attached to when the
HA feature is unconfigured and record this information.
For example, if the majority of the disks in the loop belong to Node A, you probably want the
entire loop to belong to stand-alone Node A.
After you finish
Proceed to disable the HA software.
Disabling controller failover
You need to disable the controller failover functionality in the software before reconfiguring the
hardware to create two separate, stand-alone storage systems.
Before you begin
Before performing this procedure, you must ensure that all loops and disk shelves in the system
contain disks that each belong to only one of the nodes. The disk shelves and loops cannot contain a
mix of disks belonging to Node A and Node B. In any disk shelves or loops containing such a mix of
disks, you must move data.
Steps
1. Enter the following command on either node console:
cf disable
2. Disable HA mode by entering the following command:
options cf.mode non_ha
3. Open the /etc/rc file with a text editor and remove references to the partner node in the
ifconfig entries, as shown in the following example:
Example
Original entry:
ifconfig e0 199.9.204.254 partner 199.9.204.255
Edited entry:
ifconfig e0 199.9.204.254
4. Enter the following command on the node console to reboot the node:
reboot
5. Repeat Step 2 through Step 4 on the partner node.
Reconfiguring an HA pair into two stand-alone systems | 65
After you finish
Reconfigure the hardware.
Related concepts
Requirements when changing a node using array LUNs to stand-alone on page 69
Related tasks
Reconfiguring nodes using array LUNs for stand-alone operation on page 70
Configuring a node for non-HA (stand-alone) use
By default, storage controllers are configured for use in HA mode. To use a controller in stand-alone
mode, you must change the node to non-HA mode.
Before you begin
The HA mode state of the storage controller can vary. You can use the cf status command to
determine the current configuration.
About this task
When a storage controller is shipped from the factory or when Data ONTAP is reinstalled using
option four of the Data ONTAP boot menu (Clean configuration and initialize all
disks), HA mode is enabled by default, and the system's nonvolatile memory (NVRAM or
NVMEM) is split. If you plan to use the controller in standalone mode, you must configure the node
as non-HA. Reconfiguring as non-HA mode enables full use of the system nonvolatile memory.
Note: Configuring the node as standalone removes the availability benefits of the HA
configuration and creates a single point of failure.
For information on managing the storage system by using the boot menu, see the Data ONTAP
System Administration Guide for 7-Mode.
Choices
•
If the cf status output displays Non-HA mode, then the node is configured for non-HA mode
and you are finished:
Example
node > cf status
Non-HA mode.
•
If the cf status output directs you to reboot, you must reboot the node to enable full use of the
system's nonvolatile memory:
66 | High Availability and MetroCluster Configuration Guide
Example
node> cf status
Non-HA mode. Reboot to use full NVRAM.
a) Reboot the node using the following command:
node> reboot
•
After the node reboots, you are finished.
If the cf status output displays Controller Failover enabled, you must disable both
controller failover and HA mode and then reboot the node to enable full use of the system's
nonvolatile memory:
Example
node> cf status
Controller Failover enabled
a) Disable controller failover using the following command:
node> cf disable
b) Set the mode to non-HA using the following command:
node> options cf.mode non_ha
c) Reboot the node using the following command:
node> reboot
•
After the node reboots, you are finished.
If the cf status output displays Controller Failover disabled, then HA mode is still
enabled, so you must set the HA mode to non-HA and reboot the node to enable full use of the
system's nonvolatile memory:
Example
node> cf status
Controller Failover disabled
a) Set the mode to non-HA using the following command:
node> options cf.mode non_ha
b) Reboot the node using the following command:
node> reboot
After the node reboots, you are finished.
Reconfiguring an HA pair into two stand-alone systems | 67
Reconfiguring nodes using disk shelves for stand-alone
operation
You can reconfigure the hardware to return to a single-controller configuration.
Before you begin
The HA pair software must be disabled.
Steps
1. Halt both nodes by entering the following command on each console:
halt
2. Using the information you recorded earlier in the disk shelves or loops with mixed storage,
physically move the disks to a disk shelf in a loop that belongs to the node that owns the disks.
For example, if the disk is owned by Node B, move it to a disk shelf in a loop that is owned by
Node B.
Note: Alternatively, you can move the data on the disks using a product such as Snapshot
software rather than physically moving the disks. See the Data ONTAP Data Protection Online
Backup and Recovery Guide for 7-Mode.
After moving the data from the disk, you can zero the disk and use the disk
remove_ownership command to erase the disk's ownership information. See the Data
ONTAP Storage Management Guide for 7-Mode.
3. If you are completely removing one node so that all the disk shelves will belong to a single standalone node, complete the following substeps:
a) Boot the node that you are removing into Maintenance mode as described in the Data ONTAP
System Administration Guide for 7-Mode.
b) Enter the following command to reassign all disk shelves so that they all belong to the
remaining node:
disk reassign [-o old_name | -s old_sysid] [-n new_name] -d new_sysid
c) Halt the node by entering the following command:
halt
4. Turn off the power to each node, then turn off the power to the disk shelves and unplug them
from the power source.
5. Ground yourself, then remove the HA interconnect cables from both nodes.
See the hardware documentation for your system for details.
6. Move or remove the adapter used for the HA interconnect:
68 | High Availability and MetroCluster Configuration Guide
If your system uses...
Then...
An HA interconnect adapter or an FC-VI
adapter
Remove the adapter from the system.
An NVRAM5 or NVRAM6 adapter
You might need to change the slot position of the adapter.
See the Hardware Universe for details about expansion slot
usage for the adapter.
7. Recable the system, depending on the type of system:
If you are converting a...
Then...
System with nonmirrored disks
a. Disconnect all cabling from the Channel B loop on the local node.
b. Repeat for the partner node.
System with mirrored disks or a
redundant Channel B loop
a. Connect the local node to the open Channel B loop in its local
disk shelves, as described in the appropriate disk shelf guide.
b. Repeat for the partner node.
8. Power on the disk shelves, then the individual nodes, monitoring the system console for error
messages during the boot process.
9. Run all system diagnostics at the boot prompt by entering the following command on the system
console:
boot_diags
10. Unset the partner system ID by entering the following command at the prompt:
unsetenv partner-sysid
11. Boot the node by entering the following command:
boot_ontap
12. Check HA pair status by entering the following command:
cf status
If the HA pair is disabled, you see the following message:
Non-HA mode.
13. Repeat Step 1 through Step 12 for the partner node.
Related concepts
Requirements when changing a node using array LUNs to stand-alone on page 69
Related tasks
Reconfiguring nodes using array LUNs for stand-alone operation on page 70
Reconfiguring an HA pair into two stand-alone systems | 69
Requirements when changing a node using array LUNs to
stand-alone
After uncoupling V-Series systems in an HA pair, you might need to perform additional
reconfiguration related to Data ONTAP ownership of array LUNs.
The following table summarizes the requirements when uncoupling a storage system using array
LUNs from an HA pair.
If you want to...
Requirements for
Requirements for array LUN
uncoupling systems are... assignments to systems are...
Make both systems in the
pair stand-alone systems
Remove the HA
configuration software
and interconnect cabling
No Data ONTAP reconfiguration of array
LUNs is necessary. Each system can
continue to own the array LUNs assigned
to it.
Remove one system in the
pair from service
Remove the HA
configuration software
and interconnect cabling
After uncoupling the pair, you must do
one of the following:
•
•
If you want to continue to use the
array LUNs for Data ONTAP,
reassign the array LUNs owned by the
system you are removing to another
storage system.
Prepare the array LUNs assigned to
the system you are removing for use
by systems that do not run Data
ONTAP.
Related tasks
Disabling controller failover on page 64
Reconfiguring nodes using array LUNs for stand-alone operation on page 70
70 | High Availability and MetroCluster Configuration Guide
Reconfiguring nodes using array LUNs for stand-alone
operation
After uncoupling the nodes in an HA pair, each node can continue to own its assigned array LUNs,
you can reassign its array LUNs to another V-Series system, or you can release the persistent
reservations on the array LUNs so the LUNs can be used by a non Data ONTAP host.
Before you begin
The HA pair software must be disabled.
About this task
If you want both nodes in the HA pair to remain in service and operate as stand-alone systems, each
system can continue to own the array LUNs that were assigned to it. Each system, as a stand-alone,
continues to see the array LUNs owned by the other system because both systems are still part of the
same V-Series neighborhood. However, only the system that is the owner of the array LUNs can read
from or write to the array LUN, and the systems can no longer fail over to each other.
Steps
1. On each node, halt the node by entering the following command at the console:
halt
2. Turn off the power to each node.
3. Ground yourself, then remove the HA interconnect cables from both nodes. See the hardware
documentation for your system for details.
4. Move or remove the adapter used for the HA interconnect.
If your system uses...
Then...
An HA interconnect adapter or an FC-VI
adapter
Remove the adapter from the system.
An NVRAM5 or NVRAM6 adapter
You might need to change the slot position of the adapter.
See the Hardware Universe for details about expansion slot
usage for the adapter.
5. On each node, perform the following steps:
a) Power on the node, monitoring the system console for error messages during the boot process.
b) Unset the partner system ID by entering the following command at the prompt:
unsetenv partner-sysid
6. Perform the appropriate step in the following table for what you intend to do with your system
and its storage:
Reconfiguring an HA pair into two stand-alone systems | 71
If you want to...
Then...
Keep both systems in service as Boot both systems by entering the following command on each system:
stand-alone systems and continue
boot_ontap
with both systems owning the
array LUNs that were already
assigned to them
Remove one of the systems from
service but still use the storage
that was assigned to that system
for Data ONTAP
a. Boot the node being removed into Maintenance mode, as described
in the Data ONTAP System Administration Guide for 7-Mode.
b. Use the disk reassign command to reassign all the array
LUNs so that they all belong to the node that remains.
The disk reassign command has the following syntax:
disk reassign [-o <old_name> | -s
<old_sysid>] [-n <new_name>] -d <new_sysid>
c. Remove the node from service.
d. Boot the node you are keeping in service by entering the following
command:
boot_ontap
Remove one of the systems from
service and use the array LUNs
that are currently assigned to it
for a host that does not run Data
ONTAP
Release the persistent reservations that Data ONTAP placed on those
array LUNs so that the storage administrator can use those LUNs for
other hosts.
See the Data ONTAP Storage Management Guide for 7-Mode for
information about what you need to do to prepare for taking a system
using array LUNs out of service.
Related tasks
Disabling controller failover on page 64
72 | High Availability and MetroCluster Configuration Guide
Configuring an HA pair
Bringing up and configuring a standard or mirrored HA pair for the first time can require enabling
HA mode capability and failover, setting options, configuring network connections, and testing the
configuration.
These tasks apply to all HA pairs regardless of disk shelf type.
Steps
1.
2.
3.
4.
5.
6.
Bringing up the HA pair on page 72
Enabling HA mode capability and controller failover on page 75
Setting options and parameters on page 75
Configuring network interfaces for HA pairs on page 83
Verifying the HA pair cabling and configuration on page 93
Testing takeover and giveback on page 93
Bringing up the HA pair
The first time you bring up the HA pair, you must ensure that the nodes are correctly connected and
powered up, and then use the setup program to configure the systems.
Considerations for HA pair setup
When the setup program runs on a storage system in an HA pair, it prompts you to answer some
questions specific for HA pairs.
The following list outlines some of the questions about your installation that you should think about
before proceeding through the setup program:
•
Do you want to configure interface groups for your network interfaces?
For information about interface groups, see the Data ONTAP Network Management Guide for 7Mode.
Note: You are advised to use interface groups with HA pairs to reduce SPOFs (single-points-
of-failure).
•
How do you want to configure your interfaces for takeover?
Note: If you do not want to configure your network for use in an HA pair when you run setup for
the first time, you can configure it later. You can do so either by running setup again, or by using
the ifconfig command and editing the /etc/rc file manually. However, you must provide at
least one local IP address to exit setup.
Configuring an HA pair | 73
Related tasks
Configuring shared interfaces with setup on page 73
Configuring dedicated interfaces with setup on page 74
Configuring standby interfaces with setup on page 74
Configuring shared interfaces with setup
During setup of the storage system, you can assign an IP address to a network interface and assign a
partner IP address that the interface takes over if a failover occurs.
Steps
1. Enter the IP address for the interface that you are configuring:
Example
Please enter the IP address for Network Interface e0 []:nnn.nn.nn.nnn
nnn.nn.nn.nnn is the local address for the node you are configuring.
2. Enter the netmask for the interface you are configuring, or press Return if the default value is
correct:
Example
Please enter the netmask for Network Interface e1 [255.255.0.0]:
3. Specify that this interface is to take over a partner IP address:
Example
Should interface e1 take over a partner IP address during failover?
[n]: y
4. Enter the IP address or interface name of the partner:
Example
Please enter the IP address or interface name to be taken over by e1
[]: :nnn.nn.nn.nnn
Note: If the partner is a interface group, you must use the interface name.
74 | High Availability and MetroCluster Configuration Guide
Configuring dedicated interfaces with setup
You can assign a dedicated IP address to a network interface, so that the interface does not have a
partner IP address.
About this task
This procedure is performed during setup of the storage system.
Steps
1. Enter the IP address for the interface you are configuring:
Example
Please enter the IP address for Network Interface e0 []::nnn.nn.nn.nnn
nnn.nn.nn.nnn is the local address for the node you are configuring.
2. Enter the netmask for the interface you are configuring, or press Enter if the default value is
correct:
Example
Please enter the netmask for Network Interface e1 [255.255.0.0]:
3. Specify that this interface does not take over a partner IP address:
Example
Should interface e1 take over a partner IP address during failover?
[n]: n
Configuring standby interfaces with setup
You can assign a standby IP address to a network interface, so that the interface does not have a
partner IP address.
About this task
This procedure is performed during setup of the storage system.
Steps
1. Do not enter an IP address for a standby interface; press Return.
Configuring an HA pair | 75
For example:
Please enter the IP address for Network Interface e0 []:
2. Enter the netmask for the interface you are configuring, or press Return if the default value is
correct.
For example:
Please enter the netmask for Network Interface e1 [255.255.0.0]:
3. Specify that this interface is to take over a partner IP address.
For example:
Should interface e1 take over a partner IP address during failover? [n]: y
Enabling HA mode capability and controller failover
The HA license is no longer required in Data ONTAP 8.2. You must manually configure each node
to enable or disable HA pair (high-availability) mode capability and controller failover.
Steps
1. Enter the following command on each of the node consoles to enable HA mode capability:
options cf.mode ha
2. Enter the following command on each of the node consoles to reboot the nodes:
reboot
3. Enter the following command on either of the node consoles to enable controller failover:
cf enable
4. Verify that controller failover is enabled by entering the following command on each node
console:
cf status
The system displays the following output if controller failover is enabled:
Controller Failover enabled, node2 is up.
Setting options and parameters
Options help you maintain various functions of your node, such as security, file access, and network
communication. During takeover, the value of an option might be changed by the node doing the
76 | High Availability and MetroCluster Configuration Guide
takeover. This can cause unexpected behavior during a takeover. To avoid unexpected behavior,
specific option values must be the same on both the local and partner node.
Option types for HA pairs
Some options must be the same on both nodes in the HA pair, while some can be different, and some
are affected by failover events.
In an HA pair, options are one of the following types:
•
•
•
•
Options that must be the same on both nodes for the HA pair to function correctly
Options that might be overwritten on the node that is failing over
These options must be the same on both nodes to avoid losing system state after a failover.
Options that should be the same on both nodes so that system behavior does not change during
failover
Options that can be different on each node
Note: You can find out whether an option must be the same on both nodes of an HA pair from the
comments that accompany the option value when you enter the option command. If there are no
comments, the option can be different on each node.
Setting matching node options
Because some Data ONTAP options need to be the same on both the local and partner node, you
need to check these options with the options command on each node and change them as necessary.
Steps
1. View and note the values of the options on the local and partner nodes by entering the following
command on each console:
options
The current option settings for the node are displayed on the console. Output similar to the
following is displayed:
autosupport.doit DONT
autosupport.enable on
2. Verify that the options with the following comments in parentheses are set to the same value for
both nodes:
Value might be overwritten in takeover
Same value required in local+partner
Same value in local+partner recommended
3. Correct any mismatched options by entering the options option_name option_value
command.
Note: See the na_options man page for more information about the options.
Configuring an HA pair | 77
Parameters that must be the same on each node
The parameters must be the same on each node so that takeover is smooth and data is transferred
correctly between the nodes.
The parameters listed in the following table must be the same so that takeover is smooth and data is
transferred between the nodes correctly.
Parameter...
Setting for...
Date
date, rdate
NDMP (on or off)
ndmp (on or off)
Route table published
route
Routed enabled
routed (on or off)
Time zone
timezone
Best practices for cf options
You should generally use the default options, since they represent the best choice for most
configurations.
The following table lists best practices for the cf options available in Data ONTAP. These settings
are changed with the options command. You use the man options command to display the man
page with detailed descriptions of the options.
Option
Recommended value and notes
cf.giveback.auto.cancel.on_
network_failure
ON
Leave this option on to avoid automatic giveback without
assurance that the partner's network interfaces are fully
operational.
cf.giveback.auto.cifs.termi
nate.minutes
5 minutes (default value)
cf.giveback.auto.delay.seco
nds
600 seconds (default value)
cf.giveback.auto.enable
OFF (default value)
Leave this option off so that in cases other than takeover due
to reboot or panic, you can verify that the failure that caused
the takeover is fixed before performing a giveback.
Each node in an HA pair can have a different setting for this
option
78 | High Availability and MetroCluster Configuration Guide
Option
Recommended value and notes
cf.giveback.auto.after.pani
c.takeover
ON (default value)
cf.giveback.auto.override.v
etoes
OFF (default value)
Each node in an HA pair can have a different setting for this
option.
cf.giveback.check.partner
ON (default value)
Leave on to reduce downtime caused by a unsuccessful
giveback.
Each node in an HA pair can have a different setting for this
option.
cf.hw_assist.enable
ON (default value)
The node must support remote management through a
Remote LAN Management card or Service Processor to
enable the hardware-assisted takeover.
cf.hw_assist.partner.addres
s
When using hardware-assisted takeover, the value of this
option should be equal to partner's node management IP
address.
cf.hw_assist.partner.port
4444 (default value)
cf.takeover.change_fsid
ON (default value)
Each node in an HA pair can have a different setting for this
option.
cf.takeover.detection.secon
ds
15 seconds (default value)
If sk.process.timeout.override has been manually set, it is
strongly advised that this option is set to a value larger than
or equal to the value of sk.process.timeout.override+5.
cf.takeover.on_failure
ON (default value)
Changing the value on one node automatically changes the
value on the partner.
cf.takeover.on_network_inte
rface_failure
OFF (default value)
Changing the value on one node automatically changes the
value on the partner.
cf.takeover.on_network_inte
rface_failure.policy
Use all_nics to avoid spurious takeovers due to any one
network interface failure.
Configuring an HA pair | 79
Option
Recommended value and notes
cf.takeover.on_panic
ON (default value)
Use caution when manually changing the option value. In
normal circumstances, leave this option on to avoid outages
due to panics.
Changing the value on one node automatically changes the
value on the partner.
cf.takeover.on_reboot
For systems not licensed for FC or iSCSI, the takeover on
reboot outage and subsequent giveback outage are shorter
than the longer outage incurred by the reboot. You should
use the default value of ON for these systems.
For systems licensed for FC or iSCSI, you should use the
default value of OFF. In these systems, the reboot outage is
shorter than the combined takeover and giveback outage.
Changing the value on one node automatically changes the
value on the partner.
cf.takeover.on_short_uptime
ON (default value)
Leave this option on to avoid outages due to failures
occurring early in the node's boot process.
Changing the value on one node automatically changes the
value on the partner.
cf.takeover.use_mcrc_file
OFF (default value)
Use only in MetroCluster environments.
Disabling the change_fsid option in MetroCluster configurations
In a MetroCluster configuration, you can take advantage of the change_fsid option in Data
ONTAP to simplify site takeover when the cf forcetakeover -d command is used.
About this task
In a MetroCluster configuration, if a site takeover initiated by the cf forcetakeover -d
command occurs, the following happens:
•
•
•
Data ONTAP changes the file system IDs (FSIDs) of volumes and aggregates because ownership
changes.
Because of the FSID change, clients must remount their volumes if a takeover occurs.
If using Logical Units (LUNs), the LUNs must also be brought back online after the takeover.
To avoid the FSID change in the case of a site takeover, you can set the change_fsid option to off
(the default is on). Setting this option to off has the following results if a site takeover is initiated by
the cf forcetakeover -d command:
80 | High Availability and MetroCluster Configuration Guide
•
•
•
Data ONTAP refrains from changing the FSIDs of volumes and aggregates.
Users can continue to access their volumes after site takeover without remounting.
LUNs remain online.
Attention: If the option is set to off, any data written to the failed node that did not get written to
the surviving node's NVRAM is lost. Disable the change_fsid option with great care.
Step
1. Enter the following command to disable the change_fsid option:
options cf.takeover.change_fsid off
By default, the change_fsid option is enabled (set to on ).
Related concepts
Recovering from a disaster by using MetroCluster configurations on page 232
When data loss can occur if the change_fsid option is enabled
Ensure that you have a good understanding of when data loss can occur before you disable the
change_fsid option. Disabling this option can create a seamless takeover for clients in the event of
a disaster, but there is potential for data loss.
If both the ISLs between the two sites in a fabric MetroCluster go offline, then both the systems
remain operational. However, in that scenario, client data is written only to the local plex and the
plexes become unsynchronized.
If, subsequently, a disaster occurs at one site, and the cf forcetakeover -d command is issued,
the remote plex that survived the disaster is not current. With the change_fsid option set to off,
clients switch to the stale remote plex without interruption.
If the change_fsid option is set to on, the system changes the fsids when the cf forcetakeover
-d command is issued so clients are forced to remount their volumes and check the integrity of the
data before proceeding.
Verifying and setting the HA state on controller modules and chassis
Some storage system models recognize that they are in an HA pair based on HA state information in
the controller module and chassis PROMs. If that state information is incorrect (possibly after a
chassis or controller module replacement), you can verify the state, and, if necessary, update the
state.
About this task
•
•
The ha-config show and ha-config modify commands are Maintenance mode commands.
The ha-config command only applies to the local controller module and, in the case of a dualchassis HA pair, the local chassis.
Configuring an HA pair | 81
To ensure consistent HA state information throughout the configuration, you must also run these
commands on the partner controller module and chassis, if necessary.
Steps
1. Boot into Maintenance mode.
2. Enter the following command to display the HA state of the local controller module and chassis:
ha-config show
3. If necessary, enter the following command to set the HA state of the controller module:
ha-config controller ha
The HA state of the controller module is changed.
4. If necessary, enter the following command to set the HA state of the chassis:
ha-config chassis ha
The HA state of the chassis is changed.
5. Repeat the preceding steps on the partner controller module and chassis, if necessary.
Configuring hardware-assisted takeover
You can configure hardware-assisted takeover to speed up takeover times. Hardware-assisted
takeover uses the remote management device to quickly communicate local status changes to the
partner node.
How hardware-assisted takeover speeds up takeover
Hardware-assisted takeover speeds up the takeover process by using a node's remote management
device (SP or RLM) to detect failures and quickly initiate the takeover rather than waiting for Data
ONTAP to recognize that the partner's heartbeat has stopped.
Without hardware-assisted takeover, if a failure occurs, the partner waits until it notices that the node
is no longer giving a heartbeat, confirms the loss of heartbeat, and then initiates the takeover.
The hardware-assisted takeover feature uses the following process to take advantage of the remote
management device and avoid that wait:
1. The remote management device monitors the local system for certain types of failures.
2. If a failure is detected, the remote management device immediately sends an alert to the partner
node.
3. Upon receiving the alert, the partner initiates takeover.
The hardware-assisted takeover option (cf.hw_assist.enable) is enabled by default.
82 | High Availability and MetroCluster Configuration Guide
Disabling and enabling the hardware-assisted takeover option
Hardware-assisted takeover is enabled by default on systems that use remote management.
Hardware-assisted takeover speeds the takeover process by using the RLM or SP to quickly detect
potential takeover events and alerting the partner node.
Step
1. Enter the following command to disable or enable the cf.hw_assist option:
options cf.hw_assist.enable off
options cf.hw_assist.enable on
Setting the partner address for hardware-assisted takeover
By default, on systems with an e0M port, the hardware-assisted takeover feature uses the IP address
of the partner's e0M port to communicate with the partner. On systems without an e0M port, the
system automatically selects another configured IP address. You can use the
cf.hw_assist.partner.address option to select a different IP address.
Step
1. Enter the following command to set the IP address or host name to which the hardware failure
notification is sent:
options cf.hw_assist.partner.address address
Setting the partner port for hardware-assisted takeover
When hardware-assisted takeover is enabled, the remote management (either RLM or SP) sends
hardware failure notifications to the partner. The cf.hw_assist.partner.port option enables
you to change the partner port. The default is 4444.
Step
1. Enter the following command to set the partner port to which the hardware failure notification is
sent:
options cf.hw_assist.partner.port port_number
Configuring an HA pair | 83
Configuring network interfaces for HA pairs
Configuring network interfaces requires that you understand the available configurations for takeover
and that you configure different types of interfaces (shared, dedicated, and standby) depending on
your needs.
Understanding interfaces in an HA pair
You can configure three types of interfaces on nodes in an HA pair.
What the networking interfaces do
When a node in an HA pair fails, the surviving node must be able to assume the identity of the failed
node on the network. Networking interfaces allow individual nodes in the HA pair to maintain
communication with the network if the partner fails.
See the Data ONTAP 7-Mode Network Management Guide for a description of available options and
the function each performs.
Note: You should always use multiple NICs with interface groups to improve networking
availability for both stand-alone storage systems and systems in an HA pair.
Shared, dedicated, and standby interfaces
These different types of interfaces have different roles in normal and takeover mode.
The following table lists the three types of interface configurations that you can enable in an HA pair:
Interface type
Description
Shared
This type of interface supports both the local and
partner nodes. It contains both the local node
and partner node IP addresses. During takeover,
it supports the identity of both nodes.
Dedicated
This type of interface only supports the node in
which it is installed. It contains the local node IP
address only and does not participate in network
communication beyond local node support
during takeover. It is paired with a standby
interface.
Standby
This type of interface is on the local node, but
only contains the IP address of the partner node.
It is paired with a dedicated interface.
84 | High Availability and MetroCluster Configuration Guide
Note: Most HA pair interfaces are configured as shared interfaces because they do not require an
extra NIC.
Interface roles in normal and takeover modes
You can configure shared, dedicated, and standby interfaces in an HA pair. Each type has a different
role in normal and takeover mode.
The following table shows the role of each interface type in normal and takeover mode.
Interface type
Normal mode
Takeover mode
Shared
Supports the identity of the
local node
Supports the identity of both
the local node and the failed
node
Dedicated
Supports the identity of the
local node
Supports the identity of the
local node
Standby
Idle
Supports the identity of the
failed node
Takeover configuration with shared interfaces
You can configure two NICs, one on each node, to provide two shared interfaces to each node.
In the following configuration illustration, you use two NICs to provide the two interfaces:
If Node 1 fails, interface e0 on Node 1 stops functioning, but the secondary address on e0 on Node 2
handles the Node 1 network connection with the 230 network.
If Node 2 fails, e0 on Node 2 stops functioning, but e0 on Node 1 substitutes for the failed interface
and handles the Node 2 network connection with the 230 network.
Configuring an HA pair | 85
Takeover configuration with dedicated and standby interfaces
With two NICs on each node, one can provide a dedicated interface and the other can act as a
standby interface.
In the following configuration illustration, you use two NICs for each interface, one on each storage
system. One NIC acts as a dedicated interface and the other acts as a standby interface.
If Node 1 fails, interface e0 on Node 1 stops functioning, but e0 on Node 2 substitutes for the failed
interface and handles the Node 1 network connection with the 230 network.
If Node 2 fails, e1 on Node 2 stops functioning, but e1 on Node 1 substitutes for the failed interface
and handles the Node 2 network connection with the 230 network.
Interface types and configurations
This table lists the configurations supported by each type of interface in an HA pair.
Interface
Shared
Dedicated
Standby
Partner
parameter
Ethernet
Supported
Supported
Supported
IP address or
interface name
Gigabit Ethernet
Supported
Supported
Supported
IP address or
interface name
Logical interface
(LIF)
Supported
Supported
Supported
LIF name
VLAN interface
Supported
Supported
Supported
IP address or
interface name
86 | High Availability and MetroCluster Configuration Guide
Note: Some storage systems, such as the 31xx systems, include an e0M interface that is dedicated
to management traffic. This port can be partnered in an HA pair in the same way as a regular
Ethernet interface.
IPv6 considerations in an HA pair
When enabled, IPv6 provides features such as address autoconfiguration. Using these IPv6 features
requires an understanding of how these features work with the HA pair functionality.
For additional information about IPv6, see the Data ONTAP Network Management Guide for 7Mode.
Configuration requirements for using IPv6
To use IPv6 in an HA pair, IPv6 must be enabled on both nodes. If a node that does not have IPv6
enabled attempts to take over a node using IPv6, the IPv6 addresses configured on the partner’s
interfaces are lost because the takeover node does not recognize them.
Using the ifconfig command
When using the ifconfig command with IPv4, the partner's interface can be mapped to a local
interface or the partner's IP address. When using IPv6, you must specify the partner interface, not an
IP address.
Generation of addresses during takeover
For manually configured IPv6 addresses, during takeover, the mechanism used to configure partner's
IP address remains same as in the case of IPv4.
For link-local auto-configured IPv6 addresses, during takeover, the address is auto-generated based
on the partner's MAC address.
Prefix-based auto-configured addresses are also generated during takeover, based on the prefixes in
router advertisements (RAs) received on the local link and on the partner's MAC address.
Duplicate Address Detection (DAD) is performed on all IPv6 partner addresses during takeover. This
can potentially keep the addresses in tentative state for some amount of time.
Making nondisruptive changes to the interface groups
You can use the cf takeover and cf giveback commands to make changes to interface groups
in the HA pair in a nondisruptive manner.
About this task
Changes to the /etc/rc file require a reboot to make the changes effective. You can use the cf
takeover and cf giveback commands to take over one node in the HA pair, causing it to reboot
while its storage is taken over by the partner.
Configuring an HA pair | 87
Steps
1. Edit the /etc/rc file on the desired node to modify the interface groups.
See the Data ONTAP Network Management Guide for 7-Mode for more information about
configuring interface groups.
2. From the partner node (the partner of the node on which you performed step 1), enter the
following command:
cf takeover
3. Enter the following command:
cf giveback
The node on which the changes were made reboots and its /etc/rc file is reread. The rc file is
responsible for creating the interface groups.
4. Repeat these steps, making any required changes to the /etc/rc file on the partner node.
Configuring network interfaces for the HA pair
You must configure network interfaces so that if takeover occurs, interfaces on the operating node
takes over interfaces on the failed-over node and hosts can still reach data over the network.
Before you begin
Both nodes in the HA pair must have interfaces that access the same collection of networks and
subnetworks.
You must gather the following information before configuring the interfaces:
•
•
•
The IP address for both the local node and partner node.
The netmask for both the local node and partner node.
The MTU size for both the local node and partner node.
The MTU size must be the same on both the local and partner interface.
Note: You should always use multiple NICs with interface groups to improve networking
availability for both stand-alone storage systems and systems in an HA pair.
About this task
If you configured your interfaces using setup when you first applied power to your storage systems,
you do not need to configure them again.
Note: For information about configuring an HA pair to use FC, see the Data ONTAP SAN
Administration Guide for 7-Mode.
Steps
1. Determine whether to use shared interfaces or dedicated and standby interfaces.
88 | High Availability and MetroCluster Configuration Guide
2. Configure your interfaces on one node.
3. Repeat Step 2 on the other node.
4. If desired, configure automatic takeover for the interfaces or interface groups.
Configuring a partner interface in an HA pair
To prepare for a successful takeover in an HA configuration, you can map a network interface to an
IP address or to another network interface on the partner node. During a takeover, the network
interface on the surviving node assumes the identity of the partner interface.
Before you begin
When specifying the partner IP address, both the local network interface and the partner’s network
interface must be attached to the same network segment or network switch.
About this task
•
•
•
•
•
If the network interface is an interface group, the partner interface must be denoted by an
interface name and not an IP address.
The partner interface can be an interface group or a physical network interface.
You cannot specify the underlying physical ports of an interface group in a partner configuration.
If IPv6 addresses are to be taken over, you must specify the partner interface, and not an IP
address.
Address to address mapping is not supported for IPv6 addresses.
For the partner configuration to be persistent across reboots, you must include the ifconfig
command in the /etc/rc file.
For a successful takeover in both directions, you must repeat the partner configuration in
the /etc/rc files of each node.
When specifying the partner interface name, you can configure the interfaces symmetrically, for
example map interface e1 on one node to interface e1 on the partner node.
Though symmetrical configuration is not mandatory, it simplifies administration and
troubleshooting tasks.
Step
1. Depending on the partner configuration that you want to specify, enter the following command:
If you want specify a...
Enter the following command...
Partner IP address
ifconfig interface_name partner address
interface_name is the name of the network interface.
address is the partner IP address.
Configuring an HA pair | 89
If you want specify a...
Enter the following command...
Partner interface name
ifconfig interface_name partner partner_interface
partner_interface is the name of the partner network interface.
Example: Specifying a partner IP address and partner interface name
Consider node1 and node2 are two storage systems in an HA configuration.
If the IP address of the interface e8 on node2 is 198.9.200.38, the following command allows
the interface e1 of node1 to take over the IP address of node2 for the duration of the takeover:
node1> ifconfig e1 partner 198.9.200.38
Instead of specifying the IP address, you can also specify the partner interface name. The
following command allows the interface e1 of node1 to assume the identity of e8 of node2 for
the duration of the takeover:
node1> ifconfig e1 partner e8
Configuring partner addresses on different subnets (MetroCluster
configurations only)
On MetroCluster configurations, you can configure partner addresses on different subnets. To do
this, you must create a separate /etc/mcrc file and enable the cf.takeover.use_mcrc_file
option. When taking over its partner, the node uses the partner's /etc/mcrc file to configure partner
addresses locally. These addresses will reside on the local subnetwork.
The /etc/mcrc file
The /etc/mcrc file, in conjunction with the cf.takeover.use_mcrc_file option, should be
used on MetroCluster configurations in which the partner nodes reside on separate subnetworks.
Normally, when a node (for example, nodeA) takes over its partner (nodeB), nodeA runs
nodeB's /etc/rc file to configure interfaces on nodeA to handle incoming traffic for the taken-over
partner, nodeB. This requires that the local and partner addresses are on the same subnetwork.
When the cf.takeover.use_mcrc_file option is enabled on nodeA, nodeA will use
nodeB's /etc/mcrc file upon takeover, instead of nodeB's /etc/rc file. The ifconfig commands in
the /etc/mcrc file can configure IP addresses on nodeA's subnetwork. With the correct ifconfig,
virtual IP (VIP), and routing commands in the /etc/mcrc file, the resulting configuration allows
hosts connecting to nodeB to connect to node A.
Note: The /etc/mcrc file must be maintained manually and kept in sync with the /etc/rc file.
90 | High Availability and MetroCluster Configuration Guide
Examples of /etc/rc and /etc/mcrc files
NodeA's /etc/rc file, which configures its local addresses and a partner address (which
matches the address configured in NodeB's /etc/mcrc file):
hostname nodeA
ifconfig e0a 10.1.1.1 netmask 255.255.255.0
ifconfig e0a partner 10.1.1.100
ifconfig vip add 5.5.5.5
route add default 10.1.1.50 1
routed on
options dns.domainname mycompany.com
options dns.enable on
options nis.enable off
savecore
NodeA's /etc/mcrc file, which configures a partner address on NodeB's subnetwork:
hostname nodeA
ifconfig e0a 20.1.1.200 netmask 255.255.255.0
ifconfig vip add 5.5.5.5
route add default 20.1.1.50 1
routed on
options dns.domainname mycompany.com
options dns.enable on
options nis.enable off
savecore
NodeB's /etc/rc file, which configures its local addresses and a partner address (which
matches the address configured in NodeA's /etc/mcrc file):
hostname nodeB
ifconfig e0a 20.1.1.1 netmask 255.255.255.0
ifconfig e0a partner 20.1.1.200
ifconfig vip add 7.7.7.7
route add default 20.1.1.50 1
routed on
options dns.domainname mycompany.com
options dns.enable on
options nis.enable off
savecore
NodeB's /etc/mcrc file, which configures a partner address on NodeA's subnetwork:
hostname nodeB
ifconfig e0a 10.1.1.100 netmask 255.255.255.0
ifconfig vip add 7.7.7.7
Configuring an HA pair | 91
route add default 10.1.1.50 1
routed on
options dns.domainname mycompany.com
options dns.enable on
options nis.enable off
savecore
Creating an /etc/mcrc file
You should create an /etc/mcrc file on each node of your MetroCluster configuration if the nodes
are on separate subnetworks.
Steps
1. Create an /etc/mcrc file on one node (nodeA) and place it in the /etc directory.
You might want to create the /etc/mcrc file by copying the /etc/rc file.
Note: The /etc/mcrc file must be configured manually. It is not updated automatically. It
must include all commands necessary to implement the network configuration on the partner
node in the event the node is taken over by the partner.
2. Enter the following commands in nodeA's /etc/mcrc file:
hostname nodeA
ifconfig interface MetroCluster-partner-address netmask netmask
ifconfig vip add virtual-IP-address
route add default route-for-MetroCluster-partner-address 1
routed on
other-required-options
interface is the interface on which the corresponding MetroCluster-partner-address
will reside.
MetroCluster-partner-address is the partner address of nodeB. It corresponds to the
partner address configured by an ifconfig command in nodeB's /etc/rc file.
virtual-IP-address is the virtual address of the partner (nodeB).
other-required-options denotes whatever other options are needed to correctly configure
the interface in your network environment.
Example
Example of nodeA's /etc/mcrc file:
hostname nodeA
ifconfig e0a 20.1.1.200 netmask 255.255.255.0
92 | High Availability and MetroCluster Configuration Guide
ifconfig vip add 5.5.5.5
route add default 20.1.1.50 1
routed on
options dns.domainname mycompany.com
options dns.enable on
options nis.enable off
savecore
3. Create an /etc/mcrc file on the other node (nodeB) and place it in the /etc directory.
The /etc/mcrc file must include an ifconfig command that configures the address that
corresponds to the address specified in the partner parameter in the partner node's /etc/rc.
You might want to create the /etc/mcrc file by copying the /etc/rc file.
Note: The /etc/mcrc file must be configured manually. It is not updated automatically. It
must include all commands necessary to configure the interfaces.
4. Enter the following commands in nodeB's /etc/mcrc file:
hostname nodeB
ifconfig interface MetroCluster-partner-address netmask netmask
ifconfig vip add virtual-IP-address
route add default route-for-MetroCluster-partner-address 1
routed on
other-required-options
interface is the interface on which the corresponding MetroCluster-partner-address
will reside.
MetroCluster-partner-address is the partner address of nodeA. It corresponds to the
partner address configured by an ifconfig command in nodeA's /etc/rc file.
virtual-IP-address is the virtual address of the partner (nodeA).
other-required-options denotes whatever other options are needed to correctly configure
the interface in your network environment.
Example
Example of nodeB's /etc/mcrc file:
hostname nodeB
ifconfig e0a 10.1.1.100 netmask 255.255.255.0
ifconfig vip add 7.7.7.7
route add default 10.1.1.50 1
routed on
options dns.domainname mycompany.com
options dns.enable on
options nis.enable off
savecore
Configuring an HA pair | 93
Setting the system to use the partner's /etc/mcrc file at takeover
You must enable the cf.takeover.use_mcrc_file option to cause the system to use the
partner's /etc/mcrc in the event that the local system takes over the partner. This allows the partner
IP addresses to reside on separate subnetworks. This option should be set on both nodes in a
MetroCluster configuration.
Step
1. Enter the following command on both nodes:
options cf.takeover.use_mcrc_file on
The default is off.
Verifying the HA pair cabling and configuration
You can go to the NetApp Support Site and download the Config Advisor tool to check for common
configuration errors.
About this task
Config Advisor is a configuration validation and health check tool for NetApp systems. It can be
deployed at both secure sites and non-secure sites for data collection and analysis.
Note: Support for Config Advisor is limited, and available only online.
Steps
1. Log in to the NetApp Support Site and go to Downloads > Utility ToolChest.
2. Click Config Advisor (WireGauge renamed).
3. Follow the directions on the web page for downloading and running the utility.
Testing takeover and giveback
After you configure all aspects of your HA pair, you need to verify that it operates as expected in
maintaining uninterrupted access to both nodes' storage during takeover and giveback operations.
Throughout the takeover process, the local (or takeover) node should continue serving the data
normally provided by the partner node. During giveback, control and delivery of the partner's storage
should return transparently to the partner node.
Steps
1. Check the cabling on the HA interconnect cables to make sure that they are secure.
94 | High Availability and MetroCluster Configuration Guide
2. Verify that you can create and retrieve files on both nodes for each licensed protocol.
3. Enter the following command from the local node console:
cf takeover
See the man page for command details.
The local node takes over the partner node and gives the following output:
Failover monitor: takeover completed
4. Use the sysconfig -r command to ensure that the local (takeover) node can access its partner's
disks.
5. Give back the partner node's data service after it displays the Waiting for giveback message
by entering the following command:
cf giveback
The local node releases the partner node, which reboots and resumes normal operation. The
following message is displayed on the console when the process is complete:
giveback completed
6. Proceed depending on whether you saw the message that giveback was completed successfully:
If takeover and giveback...
Then...
Is completed successfully
Repeat Step 2 through Step 5 on the partner node.
Fails
Correct the takeover or giveback failure and then repeat this procedure.
95
Managing takeover and giveback
An HA pair allows one partner to take over the storage of the other, and return the storage using the
giveback operation. Management of the nodes in the HA pair differs depending on whether one
partner has taken over the other, and the takeover and giveback operations themselves have different
options.
This information applies to all HA pairs regardless of disk shelf type.
Monitoring an HA pair in normal mode
You can display information about the status and configuration of HA pair in normal mode (when
neither node has taken over the other).
Monitoring HA pair status
You can use commands on the local node to determine whether the controller failover feature is
enabled and whether the other node in the HA pair is up.
Step
1. Enter the following command:
cf status
The following example shows that the HA pair is enabled and the interconnect is up and working
correctly:
node1>cf status
Controller Failover enabled, node2 is up.
RDMA Interconnect is up (Link 0 up).
If the output shows that one link is down, the HA pair is degraded and you must configure the
link so that it is up while the other link is still active.
Note: Depending on the storage system model, the output might display either RDMA
interconnect or VIA interconnect in the last line.
Note: Data ONTAP can disable controller failover if a software or hardware problem exists
that prevents a successful takeover. In this case, the message returned from the cf status
command describes the reason failover is disabled.
Description of HA pair status messages
The cf status command displays information about the status of the HA pair.
The following table shows messages that the cf status command can display.
96 | High Availability and MetroCluster Configuration Guide
Message
Meaning
HA enabled, partner_name is up.
The HA pair is operating normally.
local_node has taken over
partner_node.
One node has taken over the other node.
Interconnect not present.
The system does not recognize the existence of a
HA interconnect adapter.
Interconnect is down.
The HA interconnect adapter cannot access the
partner. This might be due to cabling problems
or the partner might be down.
Interconnect is up.
The HA interconnect adapter is active and can
transmit data to the partner.
partner_name_1 has detected a
mailbox disk error, takeover of
partner_name_2 disabled.
One node cannot access multiple mailbox disks.
Check access to both the local and partner root
volumes and mirrors, if they exist. Also check
for disk or FC-AL problems or offline storage
adapters.
partner_name_2 may be down, takeover
disabled because of reason (partner
halted in notakeover mode)
One node might be down.
partner_name_1 has disabled takeover
by partner_name_2 (interconnect
error)
interconnect_type Interconnect is
down (Link 0 down).
Version mismatch.
The partner node has an incompatible version of
Data ONTAP.
partner_name_1 is attempting
takeover of partner_name_2, takeover
is in module n of N modules.
A takeover is being attempted (includes
information about how far the takeover has
progressed).
partner_name_1 has taken over
partner_name_2, giveback in
progress, giveback is in module n of
N modules.
A giveback is being attempted (includes
information about how far the giveback has
progressed).
partner_name_1 has taken over
partner_name_2, partner_name_2 is
ready for giveback.
The takeover node has received information that
the failed node is ready for giveback.
Managing takeover and giveback | 97
Message
Meaning
partner_name_1 has taken over
partner_name_2, partner_name_2 is
ready for giveback.
The takeover node received information that the
failed node is ready for giveback, but giveback
cannot take place because the number of retries
exceeded the limit.
Automatic giveback is disabled due
to exceeding retry count.
Monitoring the hardware-assisted takeover feature
You can check and test the hardware-assisted takeover configuration using the hw_assist
command. You can also use the command to review statistics relating to hardware-assisted takeover.
Checking the hardware-assisted takeover status of the local and partner node
You can check the status of the hardware-assisted takeover configuration with the cf hw_assist
status command. It shows the current status for the local and partner nodes.
Step
1. Enter the following command to display the hardware-assisted takeover status:
cf hw_assist status
Example of hardware-assisted takeover status
The following example shows output from the cf hw_assist status command:
Local Node Status - ha1
Active: Monitoring alerts from partner(ha2)
port 4004 IP address 172.27.1.14
Partner Node Status - ha2
Active: Monitoring alerts from partner(ha1)
port 4005 IP address 172.27.1.15
98 | High Availability and MetroCluster Configuration Guide
Testing the hardware-assisted takeover configuration
You can test the hardware-assisted takeover configuration with the cf hw_assist test command.
About this task
The cf hw_assist test command sends a test alert to the partner. If the alert is received the
partner sends back an acknowledgment, and a message indicating the successful receipt of the test
alert is displayed on the console.
Step
1. Enter the following command to test the hardware-assisted takeover configuration:
cf hw_assist test
After you finish
Depending on the message received from the cf hw_assist test command, you might need to
reconfigure options so that the HA pair and the remote management card are operating.
Checking hardware-assisted takeover statistics
You can display statistics about hardware-assisted takeovers to determine how many alert events of
each type have been received from the partner.
Step
1. Enter the following command to display or clear the hardware-assisted takeover statistics,
respectively:
cf hw_assist stats
cf hw_assist stats clear
Example of hardware-assisted takeover statistics
The following example shows output from the cf hw_assist stats command on a system
that has received a variety of alerts from the partner:
# cf hw_assist: stats
Known hw_assist alerts received from partner
alert type
alerts
alert event
----------------------
-----------
num of
Managing takeover and giveback | 99
system_down
post_error
0
system_down
power_loss
0
system_down
abnormal_reboot
0
system_down
l2_watchdog_reset
0
system_down
power_off_via_rlm
0
system_down
power_cycle_via_rlm
0
system_down
reset_via_rlm
0
keep_alive
loss_of_heartbeat
0
keep_alive
periodic_message
18
test
test
6
Unknown hw_assist alerts received from partner
Partner nvramid mismatch alerts 5
Shared secret mismatch alerts 10
Unknown alerts 23
Number of times hw_assist alerts throttled: 3
Displaying the partner's name
You can display the name of the other node with the cf partner command.
Step
1. Enter the following command:
cf partner
Note: If the node does not yet know the name of its partner because the HA pair is new, this
command returns partner.
Displaying disk and array LUN information on an HA pair
To find out about the disks, array LUNs, or both on both the local and partner node, you can use the
sysconfig and aggr status commands, which display information about both nodes.
About this task
For each node, the sysconfig command output displays disks on both channel A and channel B:
•
The information about disks on channel A is the same as for storage systems not in an HA pair.
100 | High Availability and MetroCluster Configuration Guide
•
The information about disks on channel B is for hardware only; the sysconfig command
displays information about the adapters supporting the disks.
The command does not show whether a disk on channel B is a file system disk, spare disk, or
parity disk.
Step
1. Enter one of the following commands:
sysconfig -r
or
aggr status -r
What takeover and giveback are
Takeover is the process in which a node takes over the storage of its partner. Giveback is the process
in which that storage is returned to the partner. Both processes can be initiated manually or
configured for automatic initiation.
When takeovers occur
Takeovers can be initiated manually or occur automatically when a failover event happens,
depending on how you configure the HA pair. In some cases, takeovers occur automatically
regardless of configuration.
Takeovers can occur under the following conditions:
•
•
•
A takeover is manually initiated.
A node is in an HA pair with the default configuration for immediate takeover on panic, and that
node undergoes a software or system failure that leads to a panic.
By default, the node automatically performs a giveback to return the partner to normal operation
after the partner has recovered from the panic and booted up.
A node that is in an HA pair undergoes a system failure (for example, a loss of power) and cannot
reboot.
Note: If the storage for a node also loses power at the same time, a standard takeover is not
possible. For MetroCluster configurations, you can initiate a forced takeover in this situation.
•
•
•
•
•
One or more network interfaces that are configured to support failover become unavailable.
A node does not receive heartbeat messages from its partner.
This could happen if the partner experienced a hardware or software failure that did not result in a
panic but still prevented it from functioning correctly.
You halt one of the nodes without using the -f parameter.
You reboot one of the nodes without using the -f parameter.
Hardware-assisted takeover is enabled and triggers a takeover when the remote management
device (RLM or Service Processor) detects failure of the partner node.
Managing takeover and giveback | 101
What happens during takeover
When a takeover occurs, the unimpaired partner node takes over the functions and disk drives of the
failed node by creating an emulated storage system.
The emulated system performs the following tasks:
•
•
Assumes the identity of the failed node
Accesses the failed node’s disks, array LUNs, or both, and serves its data to clients
The partner node maintains its own identity and its own primary functions, but also handles the
added functionality of the failed node through the emulated node.
Note: When a takeover occurs, existing CIFS sessions are terminated. A graceful shutdown of the
CIFS sessions is not possible, and some data loss could occur for CIFS users.
If the node doing the takeover panics
If the node that is performing the takeover panics within 60 seconds of initiating takeover, the
following events occur:
•
•
•
The node that panicked reboots.
After it reboots, the node performs self-recovery operations and is no longer in takeover mode.
Failover is disabled.
What happens after takeover
After a takeover occurs, you view the surviving partner as having two identities, its own and its
partner’s, that exist simultaneously on the same storage system. Each identity can access only the
appropriate volumes and networks. You can send commands or log in to either storage system by
using the rsh command, allowing remote scripts that invoke storage system commands through a
Remote Shell connection to continue to operate normally.
Access with rsh
Commands sent to the failed node through a Remote Shell connection are serviced by the partner
node, as are rsh command login requests.
Access with Telnet
If you log in to a failed node through a Telnet session, you will see a message alerting you that your
storage system failed and to log in to the partner node instead. If you are logged in to the partner
node, you can access the failed node or its resources from the partner node by using the partner
command.
What happens during giveback
The local node returns ownership of the aggregates and volumes to the partner node after any issues
on the partner node are resolved or maintenance is complete. In addition, the local node returns
102 | High Availability and MetroCluster Configuration Guide
ownership when the partner node has booted up and giveback is initiated either manually or
automatically.
When the failed node is functioning again, the following events can occur:
•
•
•
You issue a cf giveback command that terminates the emulated node on the partner.
The failed node resumes normal operation, serving its own data.
The HA pair resumes normal operation, with each node ready to take over for its partner if the
partner fails.
Configuring automatic takeover
You can control when automatic takeovers occur by setting the appropriate options.
Reasons for automatic takeover
You can set options to control whether automatic takeovers occur due to different system errors. In
some cases, automatic takeover occurs by default unless you disable the option, and in some cases
automatic takeover cannot be prevented.
Takeovers can happen for several reasons. Some system errors must cause a takeover; for example,
when a system in an HA pair loses power, it automatically fails over to the other node.
However, for some system errors, a takeover is optional, depending on how you set up your HA pair.
The following table outlines which system errors can cause a takeover to occur, and whether you can
configure the HA pair for that error.
System error
Option used to configure
Default value Notes
A node undergoes a
system failure and
cannot reboot.
cf.takeover.on_failure
set to on
On
A node undergoes a
software or system
failure leading to a
panic.
cf.takeover.on_panic set to On
on
You should leave this
option enabled unless
instructed otherwise by
technical support.
Managing takeover and giveback | 103
System error
Option used to configure
Default value Notes
A node reboots
cf.takeover.on_reboot set
to on
On, unless FC
or iSCSI is
licensed.
Note: In
releases
prior to
Data
ONTAP
8.0, the
system
would only
take over
the partner
after reboot
if the
partner
took longer
than 90
seconds to
boot.
All the network
interface cards (NICs)
or interface groups
enabled for negotiated
failover on a node
failed.
cf.takeover.on_network_
interface_failure set to
on,
cf.takeover.on_network_
interface_failure.policy
set to all_nics
By default,
takeover on
network
failure is
disabled.
To enable a network
interface for negotiated
failover, you use the
ifconfig if_name
nfo command. For
more information, see
the Data ONTAP
MultiStore
Management Guide for
7-Mode.
One or more of the
NICs or interface
groups enabled for
negotiated failover
failed.
Note: If interfaces
fail on both nodes in
the HA pair,
takeover won't
occur.
By default,
takeover on
network
cf.takeover.on_network_
interface_failure.policy failure is
disabled.
set to any_nic
cf.takeover.on_network_
interface_failure set to on
To enable a network
interface or interface
group for negotiated
failover, you use the
ifconfig if_name
nfo command. For
more information, see
the Data ONTAP
MultiStore
Management Guide for
7-Mode.
104 | High Availability and MetroCluster Configuration Guide
System error
Option used to configure
Default value Notes
A node fails within 60
seconds of booting up.
cf.takeover.on_short_upt
ime set to on
On
A node cannot send
heartbeat messages to
its partner.
n/a
You cannot prevent
this condition from
causing a takeover.
You halt one of the
nodes without using
the -f flag.
n/a
You cannot prevent
this condition from
causing a takeover. If
you include the -f
flag, the takeover is
prevented.
You initiate a takeover
manually using the cf
takeover command.
n/a
You cannot prevent
this condition from
causing a takeover.
Changing the value of
this option on one node
automatically updates
the option on the
partner node.
Related tasks
Enabling and disabling automatic takeover of a panicked partner on page 107
Commands for performing a manual takeover
You need to know the commands you can use when initiating a takeover. You can initiate a takeover
on a node in an HA pair to perform maintenance on that node while still serving the data on its disks,
array LUNs, or both to users.
Command
Description
cf takeover
Initiates a takeover of the partner of the local
node. Takeover is aborted if a core dump is in
progress on the partner (if the
cf.takeover.on_panic option is set to off).
The takeover starts either after the partner halts
successfully or after a timeout.
cf takeover -f
Initiates an immediate takeover of the partner of
the local node regardless of whether the partner
node is dumping its core. The partner node is not
allowed to halt gracefully.
Managing takeover and giveback | 105
Command
Description
cf forcetakeover
Tells the HA monitor to ignore some
configuration problems that would otherwise
prevent a takeover, such as unsynchronized
NVRAM due to a faulty HA interconnect
connection. It then initiates a takeover of the
partner of the local node.
cf forcetakeover -d
Initiates a takeover of the local partner even in
the absence of a quorum of partner mailbox
disks or partner mailbox LUNs.
The cf forcetakeover -d command is valid
only if the remote_syncmirror option is
enabled.
Attention: You should only use the -d option
after you verify that the partner is down. The
-d option is used in conjunction with RAID
mirroring to recover from disasters in which
one partner is not available. For more
information, see the Data ONTAP Data
Protection Online Backup and Recovery
Guide for 7-Mode.
cf takeover -n
Initiates a takeover for a nondisruptive upgrade.
For more information, see the Data ONTAP
Upgrade and Revert/Downgrade Guide for 7Mode.
Halting a node without takeover
You can halt the node and prevent its partner from taking over.
About this task
You can halt the node and prevent its partner from taking over. For example, you might need to
perform maintenance on both the storage system and its disks and want to avoid an attempt by the
partner node to write to those disks.
Step
1. Enter the following command:
halt -f
106 | High Availability and MetroCluster Configuration Guide
Rebooting a node without takeover
You can reboot the node and prevent its partner from taking over, overriding the
cf.takeover.on_reboot option.
Step
1. Enter the following command:
reboot -f
Enabling and disabling takeover
You might want to use the cf disable command to disable takeover if you are doing maintenance
that typically causes a takeover. You can reenable takeover with the cf enable command after you
finish maintenance.
Step
1. Enter the following command:
cf enable|disable
Use cf enable to enable takeover or cf disable to disable takeover.
Note: You can enable or disable takeover from either node.
Enabling and disabling takeover on reboot
The takeover on reboot option enables you to control whether an automatic takeover occurs when a
node reboots. This automatic takeover, and the automatic giveback that follows after the reboot is
complete, can reduce the outage during which the storage belonging to the rebooting system is
unavailable.
About this task
If this option is enabled and a takeover occurs because of a reboot, then an automatic giveback is
performed after the partner has booted. This giveback occurs even if the
cf.giveback.auto.enable option is set to off. However, if a node takes over its partner due to a
reboot and that node itself reboots before it can execute a giveback, it performs automatic giveback
only if cf.giveback.auto.enable is set to on.
If the cf.takeover.on_reboot is off and a node is rebooted then the partner will not take over
immediately. But the partner could take over later if the node takes more than 180 seconds to boot.
Note: If the reboot -f command is used, then the partner does not take over under any
circumstances, even if the reboot timer expires.
Managing takeover and giveback | 107
Step
1. Enter the following command:
options cf.takeover.on_reboot on
The default is on, unless FC or iSCSI is licensed, in which case the default is off.
Note: If you enter this command on one node, the value applies to both nodes.
This option is persistent across reboots.
Enabling and disabling automatic takeover of a panicked partner
Data ONTAP is configured by default to initiate a takeover immediately if the partner node panics.
This shortens the time between the initial failure and the time that service is fully restored because
the takeover can be quicker than the recovery from the panic, although the subsequent giveback
causes another brief outage.
About this task
•
•
•
If you enter this command on one node, the value applies to both nodes.
The setting of this option is persistent across reboots.
By default, Data ONTAP will initiate an automatic giveback after a takeover on panic.
The cf.giveback.auto.after.panic.takeover option can be used to disable this
automatic giveback.
Steps
1. Verify that controller takeover is enabled by entering the following command:
cf enable
2. Enable or disable automatic takeover on panic by entering the following command:
options cf.takeover.on_panic {on|off}
on enables immediate takeover of a panicked node. This is the default value.
off disables immediate takeover of a panicked node. If you disable this option, normal takeover
procedures apply: if a node panics and stays down without sending messages to its partner for 15
seconds, the partner then automatically takes over the failed node.
Specifying the time period before takeover
You can specify how long (in seconds) a partner in an HA pair can be unresponsive before the other
partner takes over.
About this task
Both partners do not need to have the same value for this option. Thus, you can have one partner that
takes over more quickly than the other.
108 | High Availability and MetroCluster Configuration Guide
Note: If your HA pair is failing over because one of the nodes is too busy to respond to its partner,
increase the value of the cf.takeover.detection.seconds option on the partner.
Step
1. Enter the following command:
options cf.takeover.detection.seconds number_of_seconds
The valid values for number_of_seconds are 10 through 180; the default is 15.
Note: If the specified time is less than 15 seconds, unnecessary takeovers can occur, and a core
might not be generated for some system panics. Use caution when assigning a takeover time of
less than 15 seconds.
Enabling or disabling negotiated failover for a network interface
You can enable or disable negotiated failover for a network interface to trigger automatic takeover if
the interface experiences a persistent failure. You can use the nfo option of the ifconfig command
to enable or disable negotiated failover.
About this task
You can specify the nfo option for an interface group. However, you cannot specify the nfo option
for any underlying physical interface of the interface group.
Steps
1. To enable takeover during interface failure, enter the following command:
options cf.takeover.on_network_interface_failure on
2. To enable or disable negotiated failover, enter the following command:
ifconfig interface_name {nfo|-nfo}
interface_name is the name of the network interface.
nfo enables negotiated failover.
-nfo disables negotiated failover.
Example
To enable negotiated failover on the interface e8 of an HA configuration, enter the following
command:
ifconfig e8 nfo
Note: The nfo option is persistent across reboots after it is enabled on an interface.
Managing takeover and giveback | 109
Takeover of vFiler units and the vFiler limit
The vFiler limit, set with the vfiler limit command, determines how many vFiler units can exist
on a system. In an HA pair, if the two systems have different vFiler limits, some vFiler units might
not be taken over if a takeover occurs.
When performing a takeover, a system can take over only the number of vFiler units that were
specified by that system's vFiler limit. For example, if the limit is set to 5, the system can only take
over five vFiler units from the partner. If the partner that is being taken over has a higher vFiler limit,
some vFiler units cannot be taken over successfully.
For more information about setting the vFiler limit, see the Data ONTAP MultiStore Management
Guide for 7-Mode.
Managing an HA pair in takeover mode
You manage an HA pair in takeover mode by performing a number of management actions.
Determining why takeover occurred
You can use the cf status command to determine why a takeover occurred.
Step
1. At the takeover prompt, enter the following command:
cf status
Result
This command can display the following information:
•
•
•
Whether controller failover is enabled or disabled
Whether a takeover is imminent due to a negotiated failover
Whether a takeover occurred, and the reason for the takeover
Statistics in takeover mode
Explains differences in system statistics when in takeover mode.
In takeover mode, statistics for some commands differ from the statistics in normal mode in the
following ways:
•
•
Each display reflects the sum of operations that take place on the takeover node plus the
operations on the failed node.
The display does not differentiate between the operations on the takeover node and the operations
on the failed node.
The statistics displayed by each of these commands are cumulative.
110 | High Availability and MetroCluster Configuration Guide
•
•
After giving back the failed partner’s resources, the takeover node does not subtract the statistics
it performed for the failed node in takeover mode.
The giveback does not reset (zero out) the statistics.
To get accurate statistics from a command after a giveback, you can reset the statistics as
described in the man page for the command you are using.
Note: You can have different settings on each node for SNMP options, but any statistics gathered
while a node was taken over do not distinguish between nodes.
Managing emulated nodes
An emulated node is a software copy of the failed node that is hosted by the takeover node. You
access the emulated node in partner mode by using the partner command.
Management exceptions for emulated nodes
The management of disks and array LUNs and some other tasks are different when you are managing
an emulated node.
You manage an emulated node as you do any other storage system, including managing disks or
LUNs, with the following exceptions, which are described in greater detail later in this section:
•
•
•
An emulated node can access only its own disks or LUNs.
Some commands are unavailable.
Some displays differ from normal displays.
Accessing the emulated node from the takeover node
You access the emulated node from the takeover node in takeover mode with the partner
command.
About this task
You can issue the partner command in two forms:
•
•
Using the partner command without an argument
This toggles between partner mode, in which you manage the emulated node, and takeover mode,
in which you manage the takeover node.
Using the partner command with a Data ONTAP command as an argument
This executes the command on the emulated node in partner mode and then returns to takeover
mode.
Managing takeover and giveback | 111
Accessing the remote node using the partner command without arguments
You can use the partner command to toggle between the partner mode, in which commands are
executed on the partner node, and takeover mode.
Step
1. From the takeover prompt, enter the following command:
partner
Result
The prompt changes to the partner-mode prompt, which has the following form:
emulated_node/takeover_node>
Example of the change to partner mode
The following example shows the change from takeover mode to partner mode and back:
filer1(takeover)> partner
Login from console: filer2
Thu Aug 20 16:44:39 GMT [filer1: rc]: Login from console: filer2
filer2/filer1> partner
Logoff from console: filer2
filer1(takeover)> Thu Aug 20 16:44:54 GMT [filer1: rc]: Logoff from
console: filer2
filer1(takeover)>
Accessing the takeover node with the partner command containing arguments
You use the partner command with a Data ONTAP command as an argument, so you can execute
single commands on the takeover node without entering partner mode.
Step
1. From the takeover prompt, enter the following command:
partner command
command is the command you want to initiate on the emulated node.
Example of issuing the partner command with an argument
filer1(takeover)>partner cf status
filer2 has been taken over by filer1.
filer1(takeover)>
112 | High Availability and MetroCluster Configuration Guide
Accessing the emulated node remotely using Remote Shell
You can access the emulated node remotely using a Remote Shell (rsh) connection. You cannot
access the emulated node using Secure Shell (ssh) or Telnet.
Step
1. Enter the following command:
rsh failed_node command
failed_node is the name of the failed node.
command is the Data ONTAP command you want to run.
Example of an rsh command
In the following example, filer2 is the failed node:
rsh filer2 df
Emulated node command exceptions
Almost all the commands that are available on a takeover node are available on an emulated node.
Some commands, however, are either unavailable or behave differently in emulated mode.
Unavailable commands
The following commands are not available on an emulated node:
•
•
•
•
•
•
•
•
•
•
•
•
•
•
cf disable
cf enable
cf forcegiveback
cf forcetakeover
cf giveback
cf takeover
date
halt
ifconfig partner
ifconfig -partner
ifconfig mtusize
rdate
reboot
timezone
Managing takeover and giveback | 113
Commands with different behaviors
Command
Difference in emulated mode
ifconfig interface
•
•
•
Displays emulated interface mappings based
on the failed node's /etc/rc file rather than
the takeover node interface mappings.
Displays emulated interface names rather
than the interface names of the takeover
node.
Displays only interfaces that have been
configured, unlike the takeover node, which
displays all interfaces, configured or not.
mt
Uses the tape devices on the takeover node
because the failed node has no access to its tape
devices.
netstat -i
Appends a plus sign (+) to shared interfaces. A
shared interface is one that has two IP addresses
assigned to it: an IP address for the node in
which it physically resides and an IP address for
its partner node in the HA pair.
sysconfig
Displays information only about the hardware
that is attached to the takeover node. It does not
display information about the hardware that is
attached only to the failed node. For example,
the disk adapter information that the partner
sysconfig -r command displays is about the
disk adapters on the takeover node.
uptime
Displays how long the failed node has been
down and the host name of the takeover node.
aggr status
Displays information only about the hardware
that is attached to the takeover node. It does not
display information about the hardware that is
attached only to the failed node. For example,
the disk adapter information that the partner
aggr status -r command displays is about
the disk adapters on the takeover node.
114 | High Availability and MetroCluster Configuration Guide
Performing dumps and restores for a failed node
You can use the emulated node and peripheral devices attached to the takeover node to perform
dumps and restores for the failed node.
Before you begin
Any dump commands directed to the failed node’s tape drives are executed on the takeover node’s
tape drives. Therefore, any dump commands that you execute using a scheduler, such as the cron
command, succeed only under the following conditions:
•
•
The device names are the same on both nodes in the HA pair.
The dump commands for the takeover node and the emulated node are not scheduled to occur
during the same time period; the takeover node and the emulated node cannot access the tape
drives simultaneously.
About this task
Because the peripheral devices for a failed node are inaccessible, you perform dumps and restores for
a failed node by using the emulated node (available using the partner command on the takeover
node), making sure that you use a peripheral device attached to the takeover node.
For more information about performing dumps and restores, see the Data ONTAP Data Protection
Tape Backup and Recovery Guide for 7-Mode.
Step
1. Issue the backup or restore command, either in partner mode or as an argument in the
partner command.
Example
Issuing a restore command in partner mode:
node1 (takeover)> partner
node1/node2> restore [options [arguments]]
node1 (takeover)> partner
Example
Issuing a restore command as an argument in the partner command:
node1 (takeover)> partner restore [options [arguments]]
Managing takeover and giveback | 115
Giveback operations
Giveback can be implemented and configured in a number of different ways. It can also be
configured to occur automatically.
Performing a manual giveback
You can perform a normal giveback, a giveback in which you terminate processes on the partner
node, or a forced giveback.
Note: Prior to performing a giveback, you must remove failed drives in the taken-over system, as
described in the Data ONTAP Storage Management Guide for 7-Mode.
Option for shortening giveback time
You can shorten the client service outage during giveback by using the
cf.giveback.check.partner option. You should always set this option to on.
Removing failed disks prior to attempting giveback
For taken-over systems that use disks, you must remove the failed disk or disks prior to attempting to
implement giveback.
Step
1. Remove the failed disks, as described in the Data ONTAP Storage Management Guide for 7Mode.
After you finish
When all failed disks are removed or replaced, proceed with the giveback operation.
Initiating normal giveback
You can return control to a taken-over partner with the cf giveback command.
Before you begin
For fabric-attached MetroCluster configurations, the aggregates on the surviving node and the partner
node must already be rejoined to reestablish the MetroCluster configuration.
Step
1. Enter the following command on the command line of the takeover node:
cf giveback
Note: If the giveback fails, there might be a process running that prevents giveback, or there
might be failed disks in the system. Remove any failed disks and retry the command. For
116 | High Availability and MetroCluster Configuration Guide
processes that prevent giveback, you can wait and repeat the command. You can also initiate
giveback using the -f option to override the checks that prevent giveback.
After a giveback, the takeover node’s ability to take over its partner automatically is not
reenabled until the partner reboots successfully. If the partner fails to reboot, you can enter the
cf takeover command to initiate a manual takeover of the partner.
Troubleshooting if giveback fails
If the cf giveback command fails, you can check for system processes that are currently running
and might prevent giveback. You can also check that the HA interconnect is operational, and check
for any failed disks for systems using disks.
Steps
1. For systems using disks, check for and remove any failed disks, using the process described in the
Data ONTAP Storage Management Guide for 7-Mode.
2. Check for the following message on the console:
cf.giveback.disk.check.fail
Both nodes should be able to detect the same disks. This message indicates that there is a disk
mismatch: for some reason, one node is not seeing all the disks attached to the HA pair.
3. Check the HA interconnect and verify that it is correctly connected and operating.
4. Check whether any of the following processes were taking place on the takeover node at the same
time you attempted the giveback:
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Advanced mode repair operations, such as wafliron
Aggregate creation
AutoSupport collection
Backup dump and restore operations
Disks being added to a volume (vol add)
Disk ownership assignment
Disk sanitization operations
Outstanding CIFS sessions
Quota initialization
RAID disk additions
Snapshot copy creation, deletion, or renaming
SnapMirror transfers (if the partner is a SnapMirror destination)
SnapVault restorations
Storage system panics
Volume creation (traditional volume or FlexVol volume)
If any of these processes are taking place, either cancel the processes or wait until they complete,
and then retry the giveback operation.
Managing takeover and giveback | 117
5. If the cf giveback operation still does not succeed, contact support.
Related tasks
Forcing giveback on page 117
Forcing giveback
Because the takeover node might detect an error condition on the failed node that typically prevents a
complete giveback such as data not being flushed from NVRAM to the failed node’s disks, you can
force a giveback, if necessary.
About this task
You can use this procedure to force the takeover node to give back the resources of the failed node
even if the takeover node detects an error that typically prevents a complete giveback.
Note: The cf forcegiveback command should be used with caution because it can cause a loss
of data. If you cannot risk loss of data and are unable to complete the giveback, contact technical
support.
Steps
1. On the takeover node, enter the following command:
cf giveback -f
The -f parameter allows giveback to proceed as long as it would not result in data corruption or
an error on the storage system.
2. If giveback is still not successful, and if you can risk possible loss of data, enter the following
command on the takeover node:
cf forcegiveback
Attention: Use cf forcegiveback only when you cannot get cf giveback -f to succeed.
When you use this command, you risk losing any data committed to NVRAM but not to disk.
If a cifs terminate command is running, allow it to finish before forcing a giveback.
If giveback is interrupted
If the takeover node experiences a failure or a power outage during the giveback process, that process
stops and the takeover node returns to takeover mode until the failure is repaired or the power is
restored.
However, this depends upon the stage of giveback in which the failure occurred. If the node
encountered failure or a power outage during partial giveback state (after it has given back the root
aggregate), it will not return to takeover mode. Instead, the node returns to partial-giveback mode. If
this occurs, complete the process by repeating the giveback operation.
118 | High Availability and MetroCluster Configuration Guide
Configuring giveback
You can configure how giveback occurs, setting different Data ONTAP options to improve the speed
and timing of giveback.
Configuring automatic giveback
You can enable automatic giveback by using the cf.giveback.auto.enable command.
About this task
You should use the automatic giveback feature with care:
•
•
•
Do not enable automatic giveback in MetroCluster configurations.
Before the giveback operation is undertaken, you must rejoin the aggregates on the surviving
node and the partner node to reestablish the MetroCluster configuration. If automatic giveback is
enabled, this crucial step cannot be performed before the giveback.
You should leave this option disabled unless your clients are unaffected by failover, or you have
processes in place to handle repetitive failovers and givebacks.
If an automatic takeover occurred because the partner node panicked, the default behaviour is that
an automatic giveback occurs even if this option is set to off.
Step
1. Enter the following command to enable automatic giveback:
option cf.giveback.auto.enable on
The on value enables automatic giveback. The off value disables automatic giveback. This
option is off by default.
Adjusting the giveback delay time for automatic giveback
By default, there is a 600-second minimum time that a node stays in the takeover state before
performing an automatic giveback. This delay reduces the overall outage that can occur while the
taken-over partner reboots. Instead of a single longer outage, there are two brief outages (first when
the partner is taken over, the second when giveback occurs). This option affects all types of
automatic giveback but does not affect manual giveback.
Step
1. Enter the following command:
options cf.giveback.auto.delay.seconds number of seconds
The valid values for number_of_seconds are 0 to 600. The default is 600.
Attention: If cf.giveback.auto.delay.seconds is set to 0, the combined outage during
takeover and giveback results in a long total client outage.
Managing takeover and giveback | 119
Setting giveback delay time for CIFS clients
You can specify the number of minutes to delay an automatic giveback before the system terminates
CIFS clients that have open files.
About this task
During the delay, the system periodically sends notices to the affected clients. If you specify 0, CIFS
clients are terminated immediately.
This option is used only if automatic giveback is enabled.
Step
1. Enter the following command:
options cf.giveback.auto.cifs.terminate.minutes minutes
Valid values for minutes are 0 through 999. The default is 5 minutes.
Terminating long-running processes to speed automatic giveback
You can use the cf.giveback.auto.override.vetoes option to speed implementation of
automatic giveback.
When enabled, the cf.giveback.auto.override.vetoes option specifies how automatic
giveback handles long-running operations such as dump or restore. If the option is set to on, longrunning operations are terminated immediately when automatic giveback is initiated. If the option is
set to off, automatic giveback is deferred until the long-running operations are complete. This
option is used only if automatic giveback is enabled.
Setting giveback to terminate long-running processes and override vetos
You can set the automatic giveback process to terminate long-running processes and override vetos
that might prevent the giveback.
Step
1. Enter the following command:
options cf.giveback.auto.override.vetoes {on|off}
The on argument enables this option. The off argument disables this option. This option is off
by default.
120 | High Availability and MetroCluster Configuration Guide
Enabling and disabling automatic giveback after takeover due to partner panicking
Data ONTAP is configured by default to initiate an automatic giveback after an automatic takeover
that occurred due to the partner node panicking. This shortens the time between the initial failure and
the full restoration of service.
About this task
•
•
•
If you enter this command on one node, the value applies to both nodes.
The setting of this option is persistent across reboots.
This option is not affected by the setting of the cf.giveback.auto.enable option.
If cf.giveback.auto.enable is set to OFF, automatic giveback after takeover due to panic
still occurs if cf.giveback.auto.after.panic.takeover is set to ON.
Steps
1. Ensure that you enabled controller takeover by entering the following command:
cf enable
2. Enable or disable automatic giveback after takeover on panic by entering the following
command:
options cf.giveback.auto.after.panic.takeover {on|off}
on enables automatic giveback to the partner. This is the default value.
off disables automatic giveback after takeover on panic.
Troubleshooting HA issues
If takeover or giveback fails for an HA pair or you cannot enable HA, you need to check the HA
status and proceed based on messages you receive.
Steps
1. Check communication between the local and partner nodes by entering the following command
and observing the messages that the system returns:
cf status
2. Review the messages and take the appropriate action:
If the error message indicates...
Then...
An HA adapter error
Check the HA adapter cabling. Make sure that the cabling is correct
and properly seated at both ends of the cable.
Managing takeover and giveback | 121
If the error message indicates...
Then...
That the NVRAM adapter is in the Check the NVRAM slot number. Move it to the correct slot if needed.
wrong slot number
A Channel B cabling error
Check the cabling of the Channel B disk shelf loops and reseat and
tighten any loose cables.
A networking error
Check for network connectivity.
See the Data ONTAP Network Management Guide for 7-Mode for
more information.
3. If you have not already done so, run the Config Advisor tool.
4. Correct any errors or differences displayed in the output.
5. Reboot the HA pair and rerun the takeover and giveback tests.
6. If you still do not have takeover enabled, contact technical support.
Related tasks
Verifying the HA pair cabling and configuration on page 93
122 | High Availability and MetroCluster Configuration Guide
Managing DS14mk2 or DS14mk4 disk shelves in
an HA pair
You must follow specific procedures to add disk shelves to an HA pair or a MetroCluster
configuration, or to upgrade or replace disk shelf hardware in an HA pair.
If your configuration includes SAS disk shelves, see the following documents on the NetApp Support
Site:
•
•
•
For SAS disk shelf management, see the Installation and Service Guide for your disk shelf model.
For cabling SAS disk shelves in an HA pair, see the Universal SAS and ACP Cabling Guide.
For cabling SAS disk shelves in a MetroCluster configuration, see Configuring a MetroCluster
system with SAS disk shelves and FibreBridge 6500N bridges.
Adding DS14mk2 or DS14mk4 disk shelves to a multipath
HA loop
To add supported DS14mk2 or DS14mk4 disk shelves to an HA pair configured for multipath HA,
you need to add the new disk shelf to the end of a loop, ensuring that it is connected to the previous
disk shelf and to the controller.
About this task
This procedure does not apply to SAS disk shelves.
Steps
1. Confirm that there are two paths to every disk by entering the following command:
storage show disk -p
Note: If two paths are not listed for every disk, this procedure could result in a data service
outage. Before proceeding, address any issues so that all paths are redundant. If you do not
have redundant paths to every disk, you can use the nondisruptive upgrade method (failover) to
add your storage.
2. Install the new disk shelf in your cabinet or equipment rack, as described in the DiskShelf 14,
DiskShelf14mk2 FC, and DiskShelf14mk4 FC Hardware and Service Guide or DiskShelf14mk2
AT Hardware Service Guide.
3. Find the last disk shelf in the loop to which you want to add the new disk shelf.
Note: The Channel A Output port of the last disk shelf in the loop is connected back to one of
the controllers.
Managing DS14mk2 or DS14mk4 disk shelves in an HA pair | 123
Note: In Step 4 you disconnect the cable from the disk shelf. When you do this, the system
displays messages about adapter resets and eventually indicates that the loop is down. These
messages are normal within the context of this procedure. However, to avoid them, you can
optionally disable the adapter prior to disconnecting the disk shelf.
If you choose to, disable the adapter attached to the Channel A Output port of the last disk
shelf by entering the following command:
fcadmin config -d adapter
adapter identifies the adapter by name. For example: 0a.
4. Disconnect the SFP and cable coming from the Channel A Output port of the last disk shelf.
Note: Leave the other ends of the cable connected to the controller.
5. Using the correct cable for a shelf-to-shelf connection, connect the Channel A Output port of the
last disk shelf to the Channel A Input port of the new disk shelf.
6. Connect the cable and SFP you removed in Step 4 to the Channel A Output port of the new disk
shelf.
7. If you disabled the adapter in Step 3, reenable the adapter by entering the following command:
fcadmin config -e adapter
8. Repeat Step 4 through Step 7 for Channel B.
Note: The Channel B Output port is connected to the other controller.
9. Confirm that there are two paths to every disk by entering the following command:
storage show disk -p
Two paths should be listed for every disk.
Related tasks
Determining path status for your HA pair on page 125
Upgrading or replacing modules in an HA pair
In an HA pair with redundant pathing, you can upgrade or replace disk shelf modules without
interrupting access to storage.
About this task
These procedures are for DS14mk2 or DS14mk4 disk shelves.
Note: If your configuration includes SAS disk shelves, refer to the following documents on the
NetApp Support Site:
124 | High Availability and MetroCluster Configuration Guide
•
•
•
For SAS disk shelf management, see the Installation and Service Guide for your disk shelf
model.
For cabling SAS disk shelves in an HA pair, see the Universal SAS and ACP Cabling Guide.
For cabling SAS disk shelves in a MetroCluster configuration, see Configuring a MetroCluster
system with SAS disk shelves and FibreBridge 6500N bridges.
About the disk shelf modules
A disk shelf module (ESH4 or AT-FCX) in a DS14, DS14mk2, DS14mk4 FC or DS14mk2 AT
includes a SCSI-3 Enclosure Services Processor that maintains the integrity of the loop when disks
are swapped and provides signal retiming for enhanced loop stability. When upgrading or replacing a
module, you must be sure to cable the modules correctly.
The DS14, DS14mk2, DS14mk4 FC or DS14mk2 AT disk shelves support the ESH4 or AT-FCX
modules.
There are two modules in the middle of the rear of the disk shelf, one for Channel A and one for
Channel B.
Note: The Input and Output ports on module B on the DS14/DS14mk2/DS14mk4 FC shelf are the
reverse of module A.
Restrictions for changing module types
If you plan to change the type of any module in your HA pair, make sure that you understand the
restrictions.
You cannot mix ESH4 modules in the same loop with AT-FCX modules.
Best practices for changing module types
If you plan to change the type of any module in your HA pair, make sure that you review the best
practice guidelines.
•
•
Whenever you remove a module from an HA pair, you need to know whether the path you will
disrupt is redundant.
If it is, you can remove the module without interfering with the storage system’s ability to serve
data. However, if that module provides the only path to any disk in your HA pair, you must take
action to ensure that you do not incur system downtime.
When you replace a module, make sure that the replacement module’s termination switch is in the
same position as the module it is replacing.
Note: ESH4 modules are self-terminating; this guideline does not apply to ESH4 modules.
Managing DS14mk2 or DS14mk4 disk shelves in an HA pair | 125
•
•
If you replace a module with a different type of module, make sure that you also change the
cables, if necessary.
For more information about supported cable types, see the hardware documentation for your disk
shelf.
Always wait 30 seconds after inserting any module before reattaching any cables in that loop.
Testing the modules
You should test your disk shelf modules after replacing or upgrading them to ensure that they are
configured correctly and operating.
Steps
1. Verify that all disk shelves are functioning properly by entering the following command:
environ shelf
2. Verify that there are no missing disks by entering the following command:
aggr status -r
Local disks displayed on the local node should be displayed as partner disks on the partner node,
and vice-versa.
3. Verify that you can create and retrieve files on both nodes for each licensed protocol.
Determining path status for your HA pair
If you want to remove a module from your HA pair, you need to know whether the path you will
disrupt is redundant. You can use the storage show disk -p command to indicate whether the
disks have redundant paths.
About this task
If the disks have redundant paths, you can remove the module without interfering with the storage
system’s ability to serve data. However, if that module provides the only path to any of the disks in
your HA pair, you must take action to ensure that you do not incur system downtime.
Step
1. Use the storage show disk -p command at your system console.
This command displays the following information for every disk in the HA pair:
•
•
•
•
Primary port
Secondary port
Disk shelf
Bay
126 | High Availability and MetroCluster Configuration Guide
Examples for configurations with and without redundant paths
The following example shows what the storage show disk -p command output might
look like for a redundant-path HA pair consisting of FAS systems:
PRIMARY PORT SECONDARY PORT SHELF BAY
------- ---- --------- ---- ----- --0c.112
A
0b.112
B
7
0
0b.113
B
0c.113
A
7
1
0b.114
B
0c.114
A
7
2
0c.115
A
0b.115
B
7
3
0c.116
A
0b.116
B
7
4
0c.117
A
0b.117
B
7
5
0b.118
B
0c.118
A
7
6
0b.119
B
0c.119
A
7
7
0b.120
B
0c.120
A
7
8
0c.121
A
0b.121
B
7
9
0c.122
A
0b.122
B
7
10
0b.123
B
0c.123
A
7
11
0c.124
A
0b.124
B
7
12
0b.125
B
0c.125
A
7
13
Notice that every disk (for example, 0c.112/0b.112) has two ports active: one for A and one
for B. The presence of the redundant path means that you do not need to fail over one system
before removing modules from the system.
Attention: Make sure that every disk has two paths. Even in an HA pair configured for
redundant paths, a hardware or configuration problem can cause one or more disks to have
only one path. If any disk in your HA pair has only one path, you must treat that loop as if it
were in a single-path HA pair when removing modules.
The following example shows what the storage show disk -p command output might
look like for an HA pair consisting of FAS systems that do not use redundant paths:
filer1> storage show disk -p
PRIMARY PORT SECONDARY PORT SHELF
------- ---- --------- ---- ----5b.16
B
1
5b.17
B
1
5b.18
B
1
5b.19
B
1
5b.20
B
1
5b.21
B
1
5b.22
B
1
5b.23
B
1
5b.24
B
1
5b.25
B
1
5b.26
B
1
5b.27
B
1
5b.28
B
1
5b.29
B
1
BAY
--0
1
2
3
4
5
6
7
8
9
10
11
12
13
Managing DS14mk2 or DS14mk4 disk shelves in an HA pair | 127
For this HA pair, there is only one path to each disk. This means that you cannot remove a
module from the configuration, thereby disabling that path, without first performing a
takeover.
Hot-swapping a module
You can hot-swap a faulty disk shelf module, removing the faulty module and replacing it without
disrupting data availability.
About this task
When you hot-swap a disk shelf module, you must ensure that you never disable the only path to a
disk, which results in a system outage.
Attention: If there is newer firmware in the /etc/shelf_fw directory than that on the
replacement module, the system automatically runs a firmware update. This firmware update
causes a service interruption on non-multipath HA AT‑FCX installations, multipath HA
configurations running versions of Data ONTAP prior to 7.3.1, and systems with non-RoHS
AT‑FCX modules.
Steps
1. Verify that your storage system meets the minimum software requirements to support the disk
shelf modules that you are hot-swapping.
See the DiskShelf14, DiskShelf14mk2 FC, or DiskShelf14mk2 AT Hardware Service Guide for
more information.
2. Determine which loop contains the module you are removing, and determine whether any disks
are single-pathed through that loop.
3. Complete the following steps if any disks use this loop as their only path to a controller:
a) Follow the cables from the module you want to replace back to one of the nodes, called
NodeA.
b) Enter the following command at the NodeB console:
cf takeover
c) Wait for takeover to be complete and make sure that the partner node, or NodeA, reboots and
is waiting for giveback.
Any module in the loop that is attached to NodeA can now be replaced.
4. Put on the antistatic wrist strap and grounding leash.
5. Disconnect the module that you are removing from the Fibre Channel cabling.
6. Using the thumb and index finger of both hands, press the levers on the CAM mechanism on the
module to release it and pull it out of the disk shelf.
128 | High Availability and MetroCluster Configuration Guide
7. Slide the replacement module into the slot at the rear of the disk shelf and push the levers of the
cam mechanism into place.
Attention: Do not use excessive force when sliding the module into the disk shelf; you might
damage the connector.
Wait 30 seconds after inserting the module before proceeding to the next step.
8. Recable the disk shelf to its original location.
9. Check the operation of the new module by entering the following command from the console of
the node that is still running:
environ shelf
The node reports the status of the modified disk shelves.
10. Complete the following steps if you performed a takeover previously:
a) Return control of NodeA’s disk shelves by entering the following command at the console of
the takeover node:
cf giveback
b) Wait for the giveback to be completed before proceeding to the next step.
11. Test the replacement module.
12. Test the configuration.
Related concepts
Best practices for changing module types on page 124
Related tasks
Determining path status for your HA pair on page 125
Performing nondisruptive shelf replacement in a
MetroCluster configuration
In a MetroCluster configuration, you can replace the DS14mk2 FC and DS14mk4 FC disk shelves
nondisruptively. Performing nondisruptive shelf replacement (NDSR) involves preparing for the
procedure, replacing disk shelves, and verifying the disk shelves after the shelf replacement.
About this task
For performing the nondisruptive shelf replacement procedure, the two MetroCluster nodes referred
in the steps are mc-nodeA and mc-nodeB and the affected node is mc-nodeB. The disk shelf that
requires replacement is part of the loop connected to the port 9 of the FC switches on mc-nodeB.
Managing DS14mk2 or DS14mk4 disk shelves in an HA pair | 129
Steps
1. Preparing for nondisruptive shelf replacement on page 129
2. Replacing the disk shelf nondisruptively on page 130
3. Verifying the disks after the shelf replacement on page 131
Preparing for nondisruptive shelf replacement
You need to prepare the storage system before performing a nondisruptive shelf replacement
procedure.
Before you begin
All disks on loops affected by a disk shelf must be mirrored.
Steps
1. Verify that all aggregates and volumes contained in the disks on the affected loop are mirrored
and the mirroring is operational by using the aggr status and sysconfig –r commands at
both the nodes.
Example
mc-nodeB> aggr status
mc-nodeB> sysconfig –r
mc-nodeA> aggr status
mc-nodeA> sysconfig -r
2. Trigger an AutoSupport to indicate the start of the disk shelf replacement process by using the
options autosupport.doit command at both the nodes.
Example
mc-nodeB> options autosupport.doit “SHELF REPLACE: START”
mc-nodeA> options autosupport.doit “SHELF REPLACE: START”
3. Get the total number of the disks on both the nodes by using the sysconfig command.
You should save this output for comparing with the output received after the disk shelf
replacement.
130 | High Availability and MetroCluster Configuration Guide
Replacing the disk shelf nondisruptively
After preparing your storage system to perform nondisruptive shelf replacement, you can replace the
disk shelf.
Steps
1. Take the aggregate and plex on the mc-nodeB offline by using the aggr offline command.
Example
mc-nodeB> aggr offline aggr1/plex0
2. Disable switch port 9 on both the switches in the mc-nodeB by logging in as admin and using the
portdisable command.
Example
mc-nodeB_sw0> portdisable 9
mc-nodeB_sw2> portdisable 9
3. Wait until all disks missing notifications on both the nodes are complete and verify that the disks
and shelves are no longer visible by using the sysconfig and sysconfig -a commands.
Example
mc-nodeB> sysconfig
mc-nodeB> sysconfig -a
mc-nodeA> sysconfig
mc-nodeA> sysconfig –a
4. Power off the disk shelf connected to port 9 of the switches in the mc-nodeB.
5. Remove disks from the shelf.
You must ensure that you place the disks at a safe place.
6. Disconnect all FC and SFP cables from the disk shelf that are part of the loop connected to port 9.
7. Remove the disk shelf that are part of the loop connected to port 9 from the rack cabinet.
Managing DS14mk2 or DS14mk4 disk shelves in an HA pair | 131
8. Remove module A (top slot), including SFP, from the disk shelf and insert into slot A (top slot)
on replacement shelf.
If required, replace the module and the SFP.
9. Remove module B (bottom slot), including SFP, from the disk shelf and insert into slot B (bottom
slot) on replacement shelf.
If required, replace the module and the SFP.
10. Insert the replacement shelf into rack cabinet.
You must set the module speed and ensure that shelf ID is the same as the one that was replaced.
11. Cable all the connections again.
If required, replace the cables.
12. Reconnect all SFP cables between the new disk shelf and the disk shelf of the same loop
connected to port 9.
Note: You must not perform this step if the switch port 9 has a single disk shelf loop.
13. Insert the disks removed in step 5 into replacement shelf only after shelf is replaced (completely
installed with requisite install kit) into rack cabinet.
14. Power on disk shelf on the loop connected to port 9.
Verify that power is up and no alarms are reported.
Verifying the disks after the shelf replacement
After replacing the disk shelf, you must perform certain steps to ensure that the disks are operational.
Steps
1. Enable the switch port on both the switches at mc-nodeB by using the portenable command.
Example
mc-nodeB_sw0> portenable 9
mc-nodeB_sw2> portenable 9
2. Verify that all the disks and aggregates appear correctly in each offline plex by using the aggr
status, sysconfig -a, and sysconfig -r commands at mc-nodeB.
Example
mc-nodeB> aggr status -r aggr1
132 | High Availability and MetroCluster Configuration Guide
mc-nodeB> sysconfig –a
mc-nodeB> sysconfig –r
3. Verify that the disks are active and online by using the sysconfig command at both the nodes.
You should compare this output with the output generated before replacing the disk shelf to
confirm that the same number of disks appears under each FC host adapter listed as active and
online.
4. Take the aggregate and plex online on the mc-nodeB by using the aggr online command.
Example
mc-nodeB> aggr online aggr1/plex0
Note: You should wait for all aggr1 volumes on mc-nodeB to finish resyncing and return to a
mirrored state. This step might take minutes or hours.
5. Trigger an AutoSupport from both the nodes to indicate the completion of the disk shelf
replacement process.
Example
mc-nodeB> options autosupport.doit “SHELF REPLACE:FINISH”.
mc-nodeA> options autosupport.doit “SHELF REPLACE:FINISH”
133
Where to find procedures for nondisruptive
operations with HA pairs
By taking advantage of an HA pair's takeover and giveback operations, you can change hardware
components and perform software upgrades in your configuration without disrupting access to the
system's storage. You can refer to the specific documents for the required procedures.
You can perform nondisruptive operations on a system by having its partner take over the system's
storage, performing maintenance, and then giving back the storage. The following table lists where
you can find information on specific procedures:
If you want to perform this task
nondisruptively...
See the...
Upgrade Data ONTAP
Data ONTAP Upgrade and Revert/Downgrade
Guide for 7-Mode
Replace a hardware FRU component
FRU procedures for your platform
134 | High Availability and MetroCluster Configuration Guide
Overview of a MetroCluster configuration
In situations such as prolonged power outages or natural disasters, you can use the optional
MetroCluster feature of Data ONTAP to provide a quick failover to another site that contains a
nearly real time copy of the data at the disaster site.
Types of MetroCluster configurations
The MetroCluster configuration is of two types—Stretch MetroCluster configuration and Fabricattached MetroCluster configuration.
Stretch MetroCluster configurations provide data mirroring and the additional ability to initiate a
failover if an entire site becomes lost or unavailable.
Like mirrored HA pairs, fabric-attached MetroCluster configurations contain two complete, separate
copies of the data volumes or file systems that you configured as mirrored volumes or file systems in
your HA pair. The fabric-attached MetroCluster nodes can be physically distant from each other,
beyond the distance limit of a stretch MetroCluster configuration.
How mirroring works in a MetroCluster configuration
MetroCluster configurations provide data mirroring and the additional ability to initiate a failover if
an entire site becomes lost or unavailable. The MetroCluster configuration uses SyncMirror to build a
system that can continue to serve data even after complete loss of one of the sites.
Data consistency is retained, even when the data is contained in more than one aggregate.
Like mirrored HA pairs, MetroCluster configurations contain two complete copies of the specified
data volumes or file systems that you indicated as being mirrored volumes or file systems in your HA
pair. These copies are called plexes and are continually and synchronously updated every time Data
ONTAP writes data to the disks. Plexes are physically separated from each other across different
groupings of disks or array LUNs.
Note: You can have both mirrored and unmirrored volumes in a MetroCluster configuration.
However, MetroCluster configurations can preserve data only if volumes are mirrored. Unmirrored
volumes are lost if the storage where they reside is destroyed.
Unlike mirrored HA pairs, MetroCluster configurations provide the capability to force a failover
when an entire node (including the controllers and storage) is destroyed or unavailable.
Note: If you are using non-mirrored aggregates in a MetroCluster configuration, the plex must
consist of disks from Pool0. This is because using Pool1 disks for non-mirrored aggregate might
lead to unexpected takeovers due to loss of connectivity between the two sites.
Overview of a MetroCluster configuration | 135
For more information about using the SyncMirror to mirror data, see the Data ONTAP Data
Protection Online Backup and Recovery Guide for 7-Mode.
How Data ONTAP works with MetroCluster configurations
Data ONTAP divides storage across physically separated pools of disks. During configuration, Data
ONTAP identifies spare disks and divides them into separate groups called pools. These pools of
disks are physically separated from each other, allowing for high availability of mirrored volumes.
When you add a mirrored volume or add disks to one side of a mirrored volume, Data ONTAP
determines how much storage you need for the second half of the mirror, and dedicates that storage
from a separate pool to the mirrored volume.
Data ONTAP can also be configured to read from both the plexes, which in many cases improves
read performance.
Note: You can determine which side of the mirrored volume (also called a plex) is read when a
data request is received using the raid.mirror_read_plex_pref option. For more information, see the
na_options(1) man page.
Advantages of stretch MetroCluster configurations
MetroCluster configurations provide the same advantages of mirroring as mirrored HA pairs, with
the additional ability to initiate failover if an entire site becomes lost or unavailable.
For MetroCluster configuration on disks, the advantages of a stretch MetroCluster configuration are
as follows:
•
•
Your data is protected if there is a failure or loss of two or more disks in a RAID 4 aggregate or
three or more disks in a RAID-DP aggregate.
The failure of an FC-AL adapter, loop, or IOM module does not require a failover.
For MetroCluster configurations with V-Series systems using array LUNs, the advantages of a
stretch MetroCluster configuration are as follows:
•
For RAID 0 aggregates, your data is protected if there is a failure of a storage array or loss of an
array LUN on one of the storage arrays in the MetroCluster configuration.
In addition, a MetroCluster configuration provides the cf forcetakeover -d command, giving
you a single command to initiate a failover if an entire site becomes lost or unavailable. If a disaster
occurs at one of the node locations and destroys your data there, your data not only survives on the
other node, but can be served by that node while you address the issue or rebuild the configuration.
Related concepts
Recovering from a disaster by using MetroCluster configurations on page 232
136 | High Availability and MetroCluster Configuration Guide
Related references
Failover event cause-and-effect table on page 21
Configuration variations for stretch MetroCluster
configurations
Stretch MetroCluster configurations have asymmetrical and active/passive variations.
The following list describes some common configuration variations that are supported for stretch
MetroCluster configurations:
•
Asymmetrical mirroring
You can add storage to one or both nodes that is not mirrored by the other node.
Attention: Any data contained in the unmirrored storage could be lost if that site experiences a
disaster.
Note: Multiple disk failures in an unmirrored aggregate (three or more disk failures in a RAIDDP aggregate, two or more disk failures in a RAID4 aggregate) or failure of a single array
LUN in a RAID 0 aggregate cause the node to panic, resulting in a temporary data service
outage while the node reboots, a takeover occurs, or disaster recovery is performed.
You must mirror the root volumes to enable successful takeover.
Note: You must connect the unmirrored storage to both nodes, just as for mirrored storage.
You cannot have storage that is connected to only one node in an HA pair.
•
Active/passive MetroCluster configurations
In this configuration, the remote (passive) node does not serve data unless it has taken over for
the local (active) node. Mirroring the passive node’s root volume is optional. However, both
nodes must have all licenses for a MetroCluster configuration installed so that remote takeover is
possible.
Advantages of fabric-attached MetroCluster configurations
Fabric-attached MetroCluster configurations provide the same advantages of stretch MetroCluster
configurations, while also enabling the physical nodes to be physically distant from each other.
The advantages of a fabric-attached MetroCluster configuration over a stretch MetroCluster
configuration include the following:
•
•
The two halves of the configuration can be more than 500 meters apart, which provides increased
disaster protection.
For fabric-attached MetroCluster configurations with FAS systems, disk shelves and nodes are
not connected directly to each other, but are connected to a fabric with multiple data routes,
ensuring no single point of failure.
Overview of a MetroCluster configuration | 137
•
For fabric-attached MetroCluster configurations with V-Series systems using array LUNs, the
storage arrays and nodes are not connected directly to each other, but are connected to a fabric
with multiple data routes, ensuring no single point of failure.
Related concepts
Recovering from a disaster by using MetroCluster configurations on page 232
Related references
Failover event cause-and-effect table on page 21
Configuration variations for fabric-attached MetroCluster
configurations
Fabric-attached MetroCluster configurations support asymmetrical and active/passive configurations.
The following list describes some common configuration variations that are supported for fabricattached MetroCluster configurations:
•
Asymmetrical mirroring
You can add storage to one or both nodes that is not mirrored by the other node. However, any
data contained in the unmirrored storage could be lost if that site experiences a disaster.
Attention:
•
•
Multiple disk failures in an unmirrored aggregate (three or more disk failures in a RAIDDP aggregate, two or more disk failures in a RAID4 aggregate) will cause the node to
panic, resulting in a temporary data service outage while the node reboots or disaster
recovery is performed. You must mirror the root volumes to enable successful takeover.
If the system contains unmirrored aggregates using disks from remote site, issues in ISLs
can lead to multiple disk failure. Therefore, if you see ISL instability, such as ISL going
down, switch failures and so on, you must take such unmirrored aggregates offline.
Note:
•
•
•
You must connect the unmirrored storage to both nodes, just as for mirrored storage. You
cannot have storage that is connected to only one node in an HA configuration.
If the ISL links are being taken down for maintenance and your system contains unmirrored
aggregates, you must first take all unmirrored aggregates offline.
Active/passive MetroCluster configurations
In this configuration, the remote (passive) node does not serve data unless it has taken over for
the local (active) node. Mirroring the passive node’s root volume is optional. However, both
nodes must have all licenses for MetroCluster configuration installed so that remote takeover is
possible.
138 | High Availability and MetroCluster Configuration Guide
Implementing stretch MetroCluster configurations
with disks
Implementing a stretch MetroCluster configuration involves careful planning and systematic cabling.
You must be aware of requirements and configuration procedures that are specific to stretch
MetroCluster configuration.
Implementing the stretch MetroCluster configuration involves the following tasks:
•
•
•
Planning your stretch MetroCluster configuration
Cabling the stretch MetroCluster configuration
Managing the default configuration speed
Planning a stretch MetroCluster configuration using disks
Creating a detailed plan for your MetroCluster configuration helps you understand the unique
requirements for configuring a stretch MetroCluster configuration with disk. For installing a stretch
MetroCluster configuration, you should be aware of the requirements (such as the tools and
equipment) and restrictions for a stretch MetroCluster configuration.
Setup requirements and restrictions for stretch MetroCluster
configurations with disks
You must follow certain requirements and restrictions when setting up a new stretch MetroCluster
configuration with disks.
The requirement and restrictions for stretch MetroCluster configurations include those for a standard
HA pair and those for a mirrored HA pair. In addition, the following requirements apply:
•
•
•
•
•
Starting with Data ONTAP 8.1 for FAS systems and 8.2 for V-Series systems, stretch
MetroCluster configurations support SAS disk shelves when used with FibreBridge 6500N
bridges.
Your storage system must meet all the compatibility requirements for FibreBridge 6500N bridges
in the Interoperability Matrix at support.netapp.com/NOW/products/interoperability.
SAS, SATA, and Fibre Channel storage is supported on stretch MetroCluster configurations, but
both plexes of the same aggregate must use the same type of storage.
Stretch MetroCluster configuration using SAS disk shelves is supported up to 20 meters.
For stretch MetroCluster configurations using SAS disk shelves having distance greater than 20
meters require the use of the FibreBridge 6500N bridges. Each stack of SAS disk shelves requires
two FibreBridges.
For FAS systems, stacks of SAS disk shelves can be added to a MetroCluster configuration that
has DS14mk2 or DS14mk4 disk shelves.
Implementing stretch MetroCluster configurations with disks | 139
•
•
•
•
•
A stack of SAS disk shelves can contain shelves of SAS disk drives and shelves of SATA disk
drives, but each SAS disk shelf can only contain SAS or SATA disk drives; you cannot mix SAS
and SATA disk drives in the same disk shelf.
To know about the number and the type of disk shelves supported in a stack, see the
Interoperability Matrix on support.netapp.com/NOW/products/interoperability.
For stretch MetroCluster configurations using SAS disk shelves, each stack requires two Fibre
Channel ports on each controller.
For the number of SAS disk shelves and the types of SAS disk shelves supported in a stack, see
the Interoperability Matrix on support.netapp.com/NOW/products/interoperability.
Stretch MetroCluster configurations are not supported on FAS20xx systems.
The following distance limitations dictate the default speed you can set:
•
•
If the distance between the nodes is 150m and you have an 8-Gb FC-VI adapter, the default
speed is set to 8-Gb.
If you want to increase the distance to 270m or 500m, you can set the default speed to 4-Gb or
2-Gb, respectively.
• If the distance between nodes is between 150m and 270m and you have an 8-Gb FC-VI
adapter, you can set the default speed to 4-Gb.
• If the distance between nodes is between 270m and 500m and you have an 8-Gb FC-VI or 4Gb FC-VI adapter, you can set the default speed to 2-Gb.
The following Data ONTAP options must be enabled on both nodes:
•
cf.mode
You must set this option to ha.
•
cf.remote_syncmirror.enable
Set the option to on.
See the Interoperability Matrix on support.netapp.com/NOW/products/interoperability for more
information about hardware and firmware requirements for this configuration.
Related concepts
Setup requirements and restrictions for standard HA pairs on page 32
Setup requirements and restrictions for mirrored HA pairs on page 33
Hardware requirements for using disks in a MetroCluster configuration with
V-Series systems
Starting in Data ONTAP 8.2, a MetroCluster configuration with V-Series systems can include native
disks only, array LUNs on storage arrays only, or both. There are specific hardware requirements to
140 | High Availability and MetroCluster Configuration Guide
set up a MetroCluster with V-Series systems with native disks, which are the same for fabricattached and stretch MetroCluster configurations.
Hardware requirements
The following hardware is required for using native disks in a MetroCluster configuration with VSeries systems:
•
•
•
•
A pair of V-Series systems
Four FibreBridge 6500N bridges: two FibreBridge 6500N bridges per stack of disk shelves,
distributed across two sites
The disks shelves in the MetroCluster configuration are those supported by the FibreBridge
6500N bridges.
Supported firmware for the FibreBridge 6500N bridges, as shown in the Interoperability Matrix at
support.netapp.com
For a fabric-attached MetroCluster configuration, four switches
The switches must be designated in the Interoperability Matrix at support.netapp.com as being
supported for a MetroCluster configuration.
Note: If the fabric-attached MetroCluster configuration also accesses LUNs on the storage
arrays, these same switches are used for array LUN access as well. A stretch MetroCluster
configuration that also uses array LUNs would use four switches to access the array LUNs, but
the switches would not be used to access the disks.
For a MetroCluster configuration to include both disks and array LUNs, all systems in the
configuration must be V-Series systems. A MetroCluster configuration cannot include both V-Series
systems and FAS systems.
Note: Prior to Data ONTAP 8.2, MetroCluster configurations with V-Series systems could not
include V-Series systems that used disks.
See the Configuring a MetroCluster system with SAS disk shelves and FibreBridge 6500N bridges
for information about how to connect the FibreBridge 6500N and the disks shelves.
Connections between a V-Series system and a FibreBridge 6500N bridge
For this MetroCluster type...
The connection type is...
Fabric-attached
Switches
Stretch
Direct FC connections
Required connections for using uninterruptible power supplies with
MetroCluster configurations
You can use a UPS (Uninterruptible Power Supply) with your MetroCluster configurations. The UPS
enables the system to fail over gracefully if power fails for one of the nodes, or to shut down
Implementing stretch MetroCluster configurations with disks | 141
gracefully if power fails for both nodes. You must ensure that the correct equipment is connected to
the UPS.
The equipment that you need to connect to the UPS depends on how widespread a power outage you
want to protect against. Always connect both controllers, any Fibre Channel switches in use, and any
inter-switch link infrastructure (for example, a Dense Wavelength Division Multiplexing, or
DWDM) to the UPS.
You can leave the disks on the regular power supply. In this case, if power is interrupted to one site,
the controller can access the other plex until it shuts down or power is restored. If, however, power is
interrupted to both sites at the same time and the disks are not connected to the UPS, the
MetroCluster configuration cannot shut down gracefully.
Stretch MetroCluster configuration with disks
You can configure a stretch MetroCluster configuration so that each controller can access its own
storage and its partner's storage, with local storage mirrored at the partner site.
The following figure illustrates the stretch MetroCluster configuration using DS14mk2 disk shelves.
For an example of a stretch MetroCluster configuration using SAS disk shelves and FibreBridge
6500N bridge, see Configuring a MetroCluster system with SAS disk shelves and FibreBridge
6500N bridges on the NetApp Support Site. The configuration includes the following connections:
•
•
•
Connections from each controller to the user network
The MetroCluster interconnect between the two controllers
Connections from each controller to its own storage:
•
• Controller A to X
• Controller B to Y
Connections from each controller to its partner's storage:
•
• Controller A to Y
• Controller B to X
Connections from each controller to the mirrors of its storage:
•
•
Controller A to X-mirror
Controller B to Y-mirror
142 | High Availability and MetroCluster Configuration Guide
Note: This is a simplified figure that does not show disk shelf-to-disk shelf connections.
Related information
NetApp Support Site: support.netapp.com
Required documentation, tools, and equipment
Describes the NetApp documentation and the tools required to install a MetroCluster configuration.
Required documentation
You must refer to some of the flyers and guides that are required to install a new MetroCluster
configuration, or convert two stand-alone systems into a MetroCluster configuration.
NetApp hardware and service documentation is not contained within a single guide. Instead, the
field-replaceable units are documented in separate flyers at the NetApp Support Site.
The following table lists and briefly describes the documentation you might need to refer to when
preparing a new MetroCluster configuration, or converting two stand-alone systems into a
MetroCluster configuration:
Implementing stretch MetroCluster configurations with disks | 143
Manual name
Description
The appropriate system cabinet guide
Describes how to install NetApp equipment into
a system cabinet.
Site Requirements Guide
Describes the physical requirements your site
must meet to install NetApp equipment.
The appropriate disk shelf guide
Describes how to cable a disk shelf to a storage
system.
The appropriate hardware documentation for
your storage system model
Describes how to install the storage system,
connect it to a network, and bring it up for the
first time.
Diagnostics Guide
Describes the diagnostics tests that you can run
on the storage system.
Data ONTAP Upgrade and Revert/Downgrade
Guide for 7-Mode
Describes how to upgrade the storage system
and disk firmware, and how to upgrade the
storage system software.
Data ONTAP Data Protection Online Backup
and Recovery Guide for 7-Mode
Describes, among other topics, SyncMirror
technology, which is used for mirrored HA pair.
Data ONTAP System Administration Guide for
7-Mode
Describes general storage system administration.
Data ONTAP Software Setup Guide for 7-Mode Describes how to configure the software of a
new storage system for the first time.
Fabric-attached MetroCluster Systems Brocade
Switch Configuration Guide
Describes how to configure Brocade switches
for a fabric-attached MetroCluster configuration.
Configuring a MetroCluster system with SAS
disk shelves and FibreBridge 6500N bridges
Describes how to install FibreBridge 6500N
bridges as part of your MetroCluster
configuration.
Fabric-attached MetroCluster Systems Cisco
Switch Configuration Guide
Describes how to configure Cisco switches for a
fabric-attached MetroCluster configuration.
V-Series Installation Requirements and
Reference Guide
Describes how to prepare array LUNs on storage
arrays for use by Data ONTAP.
Related information
NetApp Support Site: support.netapp.com
144 | High Availability and MetroCluster Configuration Guide
Required tools
There are some tools that you need to install the MetroCluster configuration.
•
•
•
#1 and #2 Phillips screwdrivers
Hand level
Marker
Required equipment
You should receive the required equipment, such as storage system, HA interconnect adapter and so
on to install a MetroCluster configuration.
See the Hardware Universe (formerly the System Configuration Guide) at support.netapp.com/
knowledge/docs/hardware/NetApp/syscfg/index.shtml to confirm your storage system type, storage
capacity, and so on.
Note: For fabric-attached MetroCluster configurations, use the information in the Hardware
Universe labeled for MetroCluster configurations. For stretch MetroCluster configurations, use the
information in the Hardware Universe labeled “for HA Environments.”
Required equipment
Stretch MetroCluster configuration
Storage system
Two of the same type of storage systems.
Storage
See the Hardware Universe (formerly the System Configuration
Guide) at support.netapp.com/knowledge/docs/hardware/NetApp/
syscfg/index.shtml.
FibreBridge 6500N
bridges (if you are
attaching SAS disk
shelves)
Two FibreBridges are required for each stack of SAS disk shelves.
HA interconnect adapter
InfiniBand adapter (Required only for
systems that do not use an NVRAM5 or
NVRAM6 adapter, which functions as
the HA interconnect adapter.)
FC-VI adapter (Required for the 31xx,
32xx, and 62xx dual-controller systems.)
Note: When the FC-VI adapter is
installed in a 31xx, 32xx or 62xx
systems, the internal InfiniBand
interconnect is automatically
deactivated.
Fabric-attached
MetroCluster
configuration
FC-VI adapter
Implementing stretch MetroCluster configurations with disks | 145
Required equipment
Stretch MetroCluster configuration
FC-AL or FC HBA (FC
HBA for Disk) adapters
Two or four Fibre Channel HBAs.
Fabric-attached
MetroCluster
configuration
Note: The ports on the Fibre Channel HBAs are labeled 1 and 2.
However, the software refers to them as A and B. You see these
labeling conventions in the user interface and system messages
displayed on the console.
Fibre Channel switches
Not applicable
Two pairs of Brocade or
Cisco switches
Note: The Fibre
Channel switches must
be of the same type. A
mixture of switch types
(such as Brocade 300
and Brocade 5100
switches) is not allowed.
SFP (Small Form
Pluggable) modules
Not applicable
Two or four SFPs are
required per fabric,
depending on whether you
are using single or dual
inter-switch links. The
type of SFP needed
depends on the distance
between sites. One shortdistance for each switch
port used.
NVRAM adapter media
converter
Only if using fiber cabling.
Not applicable
146 | High Availability and MetroCluster Configuration Guide
Required equipment
Stretch MetroCluster configuration
Fabric-attached
MetroCluster
configuration
Cables (provided with
shipment unless otherwise
noted)
•
•
•
•
Four SC/LC (standard connector to
low-profile connector) controller-todisk shelf cables or FibreBridge
6500N
Two SC/LC IB HA adapter cables
Four SC/LC or LC/LC cables
•
Note: For information about required
•
cables, see the Interoperability Matrix
at support.netapp.com/NOW/products/
interoperability.
•
LC/LC controller-toswitch cables
SC/LC (for DS14) or
LC/LC (for DS14mk2
FC) disk shelf-toswitch cables or
FibreBridge 6500N
Two LC/LC interswitch link cables, not
provided in the
shipment
Multiple disk shelf-todisk shelf cables
Related information
NetApp Support Site: support.netapp.com
Cabling a stretch MetroCluster configuration with disks
The process of cabling a stretch MetroCluster configuration is the same as a mirrored HA pair.
However, your systems must meet the requirements for a stretch MetroCluster configuration.
About this task
Note: If you are installing the FibreBridge 6500N bridge as part of your MetroCluster
configuration, see the Configuring a MetroCluster system with SAS disk shelves and FibreBridge
6500N bridges on the NetApp Support Site at support.netapp.com for cabling procedures.
Related concepts
Configuring an HA pair on page 72
Setup requirements and restrictions for stretch MetroCluster configurations with disks on page 138
Recovering from a disaster by using MetroCluster configurations on page 232
Related tasks
Cabling a mirrored HA pair on page 52
Implementing stretch MetroCluster configurations with disks | 147
Stretch MetroCluster configuration on single-enclosure HA pairs
You can configure two stretch MetroCluster configurations between a pair of single-enclosure HA
pair systems. In this configuration, the HA pair between the two controllers in each chassis is
deactivated, and two separate, side-by-side stretch MetroCluster configurations are formed between
the four controllers.
To implement the stretch MetroCluster configuration, you must install an FC-VI adapter in each
controller to provide the HA interconnect between the systems. When the FC-VI adapter is installed
in the system, the internal InfiniBand interconnect is automatically disabled. This is different from
other stretch MetroCluster configurations, which use NVRAM adapters to provide the interconnect.
The following figure shows a stretch MetroCluster configuration on single-enclosure HA pair
systems:
Note: The FAS20xx systems do not support MetroCluster configurations.
Assigning disk pools in a stretch MetroCluster configuration
After completing the cabling of a stretch MetroCluster configuration, you must assign the attached
disk shelves to the appropriate disk pools.
About this task
You can explicitly assign disks on the attached disk shelves to the appropriate pool with the disk
assign command.
For the purpose of this procedure, Node A and Node B constitute a stretch MetroCluster
configuration having one stack of four disk shelves at each site. Each shelf stack is cabled in
multipath HA using quad-port fibre channel adapter in slot 2 of each node. The following table
shows the pool assignments for the disk shelves in the example used in this section:
148 | High Availability and MetroCluster Configuration Guide
Disk shelf...
At site...
Belongs to...
And is assigned to
that node's...
Disk shelf 1
Site A
Node A
Pool 0
Node B
Pool 1
Node B
Pool 0
Node A
Pool 1
Disk shelf 2
Disk shelf 3
Disk shelf 4
Disk shelf 5
Site B
Disk shelf 6
Disk shelf 7
Disk shelf 8
Note: Pool 0 always contains the disks that are local to (at the same site as) the storage system that
owns them.
Pool 1 always contains the disks that are remote to the storage system that owns them.
Steps
1. Boot Node A into Maintenance mode.
2. Assign the local disks to Node A pool 0 by entering the following command at the console:
disk assign -shelf 2a.shelf1 -p 0
disk assign -shelf 2b.shelf7 2a.shelf2 -p 0
This indicates that the disks in disk shelf 1 and disk shelf 2 attached to port 2a are assigned to
pool 0 of Node A.
Note: The -shelf option allows you to specify a particular shelf connected to a channel for
assigning all disks present in that particular shelf.
3. Assign the remote disks to Node A pool 1 by entering the following command at the console:
disk assign -shelf 2b.shelf7 -p 1
disk assign -shelf 2b.shelf8 -p 1
This indicates that all the disks in disk shelf 7 and disk shelf 8 attached to port 2b are assigned to
pool 1 of Node A.
4. Boot Node B into Maintenance mode.
5. Assign the local disks to Node B pool 0 by entering the following command at the console:
disk assign -shelf 2a.shelf5 -p 0
disk assign -shelf 2a.shelf6 -p 0
Implementing stretch MetroCluster configurations with disks | 149
This indicates that all the disks in disk shelf 5 and disk shelf 6 attached to port 2a are assigned to
pool 0 of Node B
6. Assign the remote disks to Node B pool 1 by entering the following command at the console:
disk assign -shelf 2b.shelf3 -p 1
disk assign -shelf 2b.shelf4 -p 1
This indicates that all the disks in disk shelf 3 and disk shelf 4 attached to port 2b are assigned to
pool 1 of Node B.
After you finish
Proceed to verify the disk paths on the system.
Managing the default configuration speed of a stretch
MetroCluster configuration
You can manage the default configuration speed of a stretch MetroCluster configuration either by
changing the default speed or by resetting the default configuration speed.
Changing the default configuration speed of a stretch MetroCluster
configuration
The distance between your nodes and the FC-VI adapter speed dictates the default configuration
speed of your stretch MetroCluster configuration. If the distance between nodes is greater than the
supported default configuration speed, you must change the default configuration speed.
Before you begin
The stretch MetroCluster system's default configuration speed must conform to the stretch
MetroCluster system setup requirements and restrictions.
About this task
You must perform these steps at both the nodes if they are configured at different speeds.
Steps
1. At the storage console prompt, halt the storage system by entering the following command:
halt
2. Reset the configuration speed by entering the following commands.
150 | High Availability and MetroCluster Configuration Guide
If you want to set
the speed to...
Then...
4 Gbps
a. Enter the following command:
setenv ispfcvi-force-4G-only? True
b. If you previously modified the speed to 2 Gbps, ensure that the 2 Gbps port
speed is not set by entering the following command:
unsetenv ispfcvi-force-2G-only?
c. Verify that your system is unconfigured for 2 Gbps by entering the following
command:
printenv ispfcvi-force-2G-only?
The storage system console displays output similar to the following:
Variable Name
Value
-------------------- -------------ispfcvi-force-2G-only? *** Undefined ***
d. Verify that your storage system is configured for 4 Gbps by entering the
following command:
printenv ispfcvi-force-4G-only?
The storage system console displays output similar to the following:
Variable Name
Value
-------------------- -------------ispfcvi-force-4G-only?
true
Implementing stretch MetroCluster configurations with disks | 151
If you want to set
the speed to...
Then...
2 Gbps
a. Enter the following command:
setenv ispfcvi-force-2G-only? True
b. If you previously modified the default speed to 4 Gbps, ensure that the 4 Gbps
speed is not set by entering the following command:
unsetenv ispfcvi-force-4G-only?
c. Verify that your storage system is unconfigured for 4 Gbps by entering the
following command:
printenv ispfcvi-force-4G-only?
The storage system console displays output similar to the following:
Variable Name
Value
-------------------- -----------------ispfcvi-force-4G-only?
*** Undefined ***
d. Verify that your storage system is configured for 2 Gbps by entering the
following command:
printenv ispfcvi-force-2G-only?
If your storage system is configured correctly, the system console displays output
similar to the following:
Variable Name
Value
-------------------- -------------------ispfcvi-force-2G-only?
true
3. Boot the storage system by entering the following command:
boot_ontap
Resetting a stretch MetroCluster configuration to the default speed
If you modified the default configuration speed in a stretch MetroCluster configuration using an FCVI adapter, you can reset the speed to the default configuration speed by using the unsetenv
command at the boot environment prompt.
About this task
The steps require you to unset the previously configured speed, but because the speed is set to default
automatically, you do not need to set the default speed explicitly.
152 | High Availability and MetroCluster Configuration Guide
Steps
1. At the storage prompt, halt the system by entering the following command:
halt
2. Reset the configuration speed.
If you want Then...
to reset the
speed from...
4 Gbps
a. Enter the following command:
unsetenv ispfcvi-force-4G-only?
b. Verify that your system is unconfigured for 4 Gbps by entering the following
command:
printenv ispfcvi-force-4G-only?
The system console displays output similar to the following:
Variable Name
Value
----------------------------ispfcvi-force-4G-only? *** Undefined ***
2 Gbps
a. Enter the following command:
unsetenv ispfcvi-force-2G-only?
b. Verify that your system is unconfigured for 2 Gbps by entering the following
command:
printenv ispfcvi-force-2G-only?
3. Boot the storage system by entering the following command:
boot_ontap
153
Implementing fabric-attached MetroCluster
configurations with disks
Implementing a fabric-attached MetroCluster configuration involves careful planning and systematic
cabling. The requirements and configuration procedures are specific to fabric-attached MetroCluster
configuration.
Implementing the fabric-attached MetroCluster configuration involves the following tasks:
•
•
•
Planning your implementation
Cabling the disk shelves and various other components such as switches, bridges and so on
Setting the preferred primary ports
Planning a fabric-attached MetroCluster configuration using
disks
Creating a detailed plan for your fabric-attached MetroCluster configuration helps you understand
the unique requirements for configuring a fabric-attached MetroCluster configuration with disk.
Installing a fabric-attached MetroCluster configuration involves connecting and configuring a
number of devices, which might be done by different people.
Setup requirements and restrictions for fabric-attached MetroCluster
configurations with disks
You must follow certain requirements and restrictions when setting up a new fabric-attached
MetroCluster configuration.
The setup requirements for a fabric-attached MetroCluster configuration with disks include those for
standard and mirrored HA configurations, with the following exceptions:
Note: See the Interoperability Matrix at support.netapp.com/NOW/products/interoperability for
more information about hardware and firmware requirements for this configuration.
Node requirements
•
The nodes must be one of the following system models configured for mirrored volume use; each
node in the pair must be the same model.
•
•
•
•
•
31xx systems
32xx systems
62xx systems
6080 systems
6040 systems
154 | High Availability and MetroCluster Configuration Guide
•
Each node requires a 4-Gbps FC-VI (Fibre Channel-Virtual Interface) adapter; the slot position is
dependent on the controller model.
Note: For information about supported cards and slot placement, see the Hardware Universe on
the NetApp Support Site.
The FC-VI adapter is also called a VI-MC or VI-MetroCluster adapter.
•
•
The 8-Gbps FC-VI adapter is supported only on the 32xx and 62xx systems.
If you want to convert a stretch MetroCluster configuration to a fabric-attached MetroCluster
configuration, and you have modified the default configuration speed of FC-VI adapters by
setting the boot environment variables to True, you must reset it to default configuration speed
before the conversion.
Disk and disk shelf requirements
•
•
•
•
•
•
•
•
•
The only Fibre Channel disk shelves supported are DS14mk2 or DS14mk4.
Starting with Data ONTAP 8.1, fabric-attached MetroCluster configurations also support SAS
disk shelves when used with FibreBridge 6500N.
Each stack of SAS disk shelves requires two FibreBridges.
Your storage system must meet all the compatibility requirements for FibreBridge 6500N in the
Interoperability Matrix at support.netapp.com/NOW/products/interoperability.
DS14mk2 or DS14mk4 disk shelves using SATA drives and AT-FCX storage is not supported.
You can connect a maximum of two DS14mk2 or DS14mk4 disk shelves to each loop.
A stack of SAS disk shelves can contain shelves of SAS disk drives and shelves of SATA disk
drives, but each SAS disk shelf can only contain SAS or SATA disk drives; you cannot have SAS
and SATA disk drives in the same disk shelf.
For more information about the number and the types of disk shelves supported in a stack, see the
Interoperability Matrix at support.netapp.com/NOW/products/interoperability .
Stacks of SAS disk shelves can be added to a fabric-attached MetroCluster configuration that has
DS14mk2 or DS14mk4 FC disk shelves.
For the number of SAS disk shelves and the types of SAS disk shelves supported in a stack, see
the Interoperability Matrix at support.netapp.com/NOW/products/interoperability.
Capacity limits
The maximum capacity for a system configured in a fabric-attached MetroCluster configuration is
the smallest of the following limits:
•
The maximum storage capacity for the node.
Note: For the maximum storage capacity, see the Hardware Universe on the NetApp Support
Site at support.netapp.com.
•
•
840 Fibre Channel disks (60 disk shelves).
In a MetroCluster configuration using both FC and SAS disk shelves), the total disk drives count
must not exceed storage capacity.
Implementing fabric-attached MetroCluster configurations with disks | 155
Fibre Channel switch requirements
•
•
•
•
•
•
Each site of the MetroCluster configuration requires two switches.
Switches must be a supported Brocade or Cisco model supplied by NetApp.
Customer-supplied switches are not supported.
The two switches within the fabric must be the same model and must be licensed for the same
number of ports.
All four switches for a particular fabric-attached MetroCluster configuration must support the
same maximum speed.
Switches must be running the correct firmware version.
Configurations using SAS storage and FibreBridge 6500N bridges supports Brocade or Cisco
switch models supplied by NetApp.
See the Interoperability Matrix at support.netapp.com/NOW/products/interoperability for more
information about supported switch models.
Related concepts
Setup requirements and restrictions for standard HA pairs on page 32
Setup requirements and restrictions for mirrored HA pairs on page 33
Configuration limitations for fabric-attached MetroCluster configurations
with disks
You must be aware of certain limitations when setting up a new fabric-attached MetroCluster
configuration.
The fabric-attached MetroCluster configuration has the following limitations:
•
•
•
DS14mk2 or DS14mk4 disk shelves using SATA and AT-FCX storage is not supported.
You cannot use the switches of the MetroCluster configuration to connect Fibre Channel tape
devices, or for FC traffic of any kind; you can connect only system controllers and disk shelves to
those switches.
The switches of the MetroCluster configuration either connect DS14mk2 or DS14mk4 disk
shelves to the controllers or FibreBridge 6500N bridge (connected to SAS or SATA disk shelves)
to the disk shelves.
You can connect a tape storage area network (SAN) to either of the nodes, but the tape SAN must
not use the switches used in a MetroCluster configuration.
Hardware requirements for using disks in a MetroCluster configuration with
V-Series systems
Starting in Data ONTAP 8.2, a MetroCluster configuration with V-Series systems can include native
disks only, array LUNs on storage arrays only, or both. There are specific hardware requirements to
156 | High Availability and MetroCluster Configuration Guide
set up a MetroCluster with V-Series systems with native disks, which are the same for fabricattached and stretch MetroCluster configurations.
Hardware requirements
The following hardware is required for using native disks in a MetroCluster configuration with VSeries systems:
•
•
•
•
A pair of V-Series systems
Four FibreBridge 6500N bridges: two FibreBridge 6500N bridges per stack of disk shelves,
distributed across two sites
The disks shelves in the MetroCluster configuration are those supported by the FibreBridge
6500N bridges.
Supported firmware for the FibreBridge 6500N bridges, as shown in the Interoperability Matrix at
support.netapp.com
For a fabric-attached MetroCluster configuration, four switches
The switches must be designated in the Interoperability Matrix at support.netapp.com as being
supported for a MetroCluster configuration.
Note: If the fabric-attached MetroCluster configuration also accesses LUNs on the storage
arrays, these same switches are used for array LUN access as well. A stretch MetroCluster
configuration that also uses array LUNs would use four switches to access the array LUNs, but
the switches would not be used to access the disks.
For a MetroCluster configuration to include both disks and array LUNs, all systems in the
configuration must be V-Series systems. A MetroCluster configuration cannot include both V-Series
systems and FAS systems.
Note: Prior to Data ONTAP 8.2, MetroCluster configurations with V-Series systems could not
include V-Series systems that used disks.
See the Configuring a MetroCluster system with SAS disk shelves and FibreBridge 6500N bridges
for information about how to connect the FibreBridge 6500N and the disks shelves.
Connections between a V-Series system and a FibreBridge 6500N bridge
For this MetroCluster type...
The connection type is...
Fabric-attached
Switches
Stretch
Direct FC connections
Required connections for using uninterruptible power supplies with
MetroCluster configurations
You can use a UPS (Uninterruptible Power Supply) with your MetroCluster configurations. The UPS
enables the system to fail over gracefully if power fails for one of the nodes, or to shut down
Implementing fabric-attached MetroCluster configurations with disks | 157
gracefully if power fails for both nodes. You must ensure that the correct equipment is connected to
the UPS.
The equipment that you need to connect to the UPS depends on how widespread a power outage you
want to protect against. Always connect both controllers, any Fibre Channel switches in use, and any
inter-switch link infrastructure (for example, a Dense Wavelength Division Multiplexing, or
DWDM) to the UPS.
You can leave the disks on the regular power supply. In this case, if power is interrupted to one site,
the controller can access the other plex until it shuts down or power is restored. If, however, power is
interrupted to both sites at the same time and the disks are not connected to the UPS, the
MetroCluster configuration cannot shut down gracefully.
Requirements for a shared-switches configuration
You must follow certain requirements when creating a shared-switches configuration.
•
•
•
•
Fabric-attached MetroCluster configuration
Brocade 5100 or 6510 switch
SAS disk shelves with FibreBridge 6500N
To know how to install and configure FibreBridge 6500N as part of your MetroCluster
configuration, see Configuring a MetroCluster system with SAS disk shelves and FibreBridge
6500N bridges on the NetApp Support Site at support.netapp.com.
A minimum of two ISLs per fabric with TI enabled.
For more information about configuring a switch in a shared-switches configuration, see the
Fabric-attached MetroCluster Systems Brocade Switch Configuration Guide on the NetApp
Support Site at support.netapp.com.
Fabric-attached MetroCluster configuration with disks
A fabric-attached MetroCluster configuration includes two pairs Brocade or Cisco Fibre Channel
switch fabrics that provide long distance connectivity between the nodes. Through the switches, each
controller can access its own storage and its partner's storage, with local storage mirrored at the
partner site.
The following figure illustrates the fabric-attached MetroCluster configuration with disks:
158 | High Availability and MetroCluster Configuration Guide
Note: This is a simplified figure that does not show disk shelf-to-disk shelf connections.
How fabric-attached MetroCluster configurations use Brocade and Cisco
Fibre Channel switches
A MetroCluster configuration for distances greater than 500 meters connects the two nodes by using
four Brocade or Cisco Fibre Channel switches in a dual-fabric configuration for redundancy.
Each site has two Fibre Channel switches, each of which is connected through an inter-switch link to
a partner switch at the other site. The inter-switch links are fiber optic connections that provide a
greater distance between nodes than other HA pairs. For more information about the switches, see the
Fabric-attached MetroCluster Systems Brocade Switch Configuration Guide for Brocade switches
and Fabric-attached MetroCluster Systems Cisco Switch Configuration Guide for Cisco switches,
available at the NetApp Support Site.
Each local switch combines with a partner switch to form a fabric. By using four switches instead of
two, redundancy is provided to avoid single-points-of-failure in the switches and their connections.
Like a stretch MetroCluster configuration, a fabric-attached MetroCluster configuration employs
SyncMirror to build a system that can continue to serve data even after complete loss of one of the
nodes and the storage at that site. Data consistency is retained, even when the data is contained in
more than one aggregate.
Related information
NetApp Support Site: support.netapp.com
Implementing fabric-attached MetroCluster configurations with disks | 159
Planning the fabric-attached MetroCluster installation
You must fill out the fabric-attached MetroCluster configuration worksheet to record specific cabling
information about your fabric-attached MetroCluster configuration. You must identify several pieces
of information that you use during configuration procedures. Recording this information can reduce
configuration errors.
Step
1. Fill in the following tables.
Each site has two Brocade Fibre Channel switches. Use the following table to record the
configured names, IP addresses, and domain IDs of these switches.
Switch
number...
At site...
1
A
2
A
3
B
4
B
Is named...
IP address...
Domain ID...
In addition to on-board ports, each site has an FC-VI adapter and two Fibre Channel HBAs that
connect the node to the switches. Use the following table to record which switch port these
adapters are connected to:
This adapter...
FC-VI adapter
FC HBA 1
FC HBA 2
At site...
Port 1 of this adapter
is...
Port 2 of this adapter
is...
Cabled to
switch...
Cabled to
switch...
Switch
port...
A
1
2
B
3
4
A
1
2
B
3
4
A
1
2
B
3
4
Switch
port...
Disk shelves or FibreBridge 6500N bridges (if you are using SAS disk shelves) at each site
connect to the Fibre Channel switches. Use the following table to record which switch port the
disk shelves or FibreBridge 6500N bridges are connected to.
160 | High Availability and MetroCluster Configuration Guide
Note: If you are using SAS or SATA disk shelves, the FibreBridges connect to the disk
shelves. Each FibreBridge 6500N bridge needs to be connected to only one switch through one
of the FC ports on the FibreBridge bridge. For cabling examples, see Configuring a
MetroCluster system with SAS disk shelves and FibreBridge 6500N bridges at the NetApp
Support Site.
Disk shelf or
FibreBridge
6500N...
At site...
Belonging to...
Connects to
switches...
1
A
Node A Pool 0
1 and 2
On switch port...
2
3
Node B Pool 1
4
5
B
Node B Pool 0
3 and 4
6
7
Node A Pool 1
8
Required documentation, tools, and equipment
Describes the NetApp documentation and the tools required to install a MetroCluster configuration.
Required documentation
You must refer to some of the flyers and guides that are required to install a new MetroCluster
configuration, or convert two stand-alone systems into a MetroCluster configuration.
NetApp hardware and service documentation is not contained within a single guide. Instead, the
field-replaceable units are documented in separate flyers at the NetApp Support Site.
The following table lists and briefly describes the documentation you might need to refer to when
preparing a new MetroCluster configuration, or converting two stand-alone systems into a
MetroCluster configuration:
Manual name
Description
The appropriate system cabinet guide
Describes how to install NetApp equipment into
a system cabinet.
Site Requirements Guide
Describes the physical requirements your site
must meet to install NetApp equipment.
The appropriate disk shelf guide
Describes how to cable a disk shelf to a storage
system.
Implementing fabric-attached MetroCluster configurations with disks | 161
Manual name
Description
The appropriate hardware documentation for
your storage system model
Describes how to install the storage system,
connect it to a network, and bring it up for the
first time.
Diagnostics Guide
Describes the diagnostics tests that you can run
on the storage system.
Data ONTAP Upgrade and Revert/Downgrade
Guide for 7-Mode
Describes how to upgrade the storage system
and disk firmware, and how to upgrade the
storage system software.
Data ONTAP Data Protection Online Backup
and Recovery Guide for 7-Mode
Describes, among other topics, SyncMirror
technology, which is used for mirrored HA pair.
Data ONTAP System Administration Guide for
7-Mode
Describes general storage system administration.
Data ONTAP Software Setup Guide for 7-Mode Describes how to configure the software of a
new storage system for the first time.
Fabric-attached MetroCluster Systems Brocade
Switch Configuration Guide
Describes how to configure Brocade switches
for a fabric-attached MetroCluster configuration.
Configuring a MetroCluster system with SAS
disk shelves and FibreBridge 6500N bridges
Describes how to install FibreBridge 6500N
bridges as part of your MetroCluster
configuration.
Fabric-attached MetroCluster Systems Cisco
Switch Configuration Guide
Describes how to configure Cisco switches for a
fabric-attached MetroCluster configuration.
V-Series Installation Requirements and
Reference Guide
Describes how to prepare array LUNs on storage
arrays for use by Data ONTAP.
Related information
NetApp Support Site: support.netapp.com
Required tools
There are some tools that you need to install the MetroCluster configuration.
•
•
•
#1 and #2 Phillips screwdrivers
Hand level
Marker
162 | High Availability and MetroCluster Configuration Guide
Required equipment
You should receive the required equipment, such as storage system, HA interconnect adapter and so
on to install a MetroCluster configuration.
See the Hardware Universe (formerly the System Configuration Guide) at support.netapp.com/
knowledge/docs/hardware/NetApp/syscfg/index.shtml to confirm your storage system type, storage
capacity, and so on.
Note: For fabric-attached MetroCluster configurations, use the information in the Hardware
Universe labeled for MetroCluster configurations. For stretch MetroCluster configurations, use the
information in the Hardware Universe labeled “for HA Environments.”
Required equipment
Stretch MetroCluster configuration
Fabric-attached
MetroCluster
configuration
Storage system
Two of the same type of storage systems.
Storage
See the Hardware Universe (formerly the System Configuration
Guide) at support.netapp.com/knowledge/docs/hardware/NetApp/
syscfg/index.shtml.
FibreBridge 6500N
bridges (if you are
attaching SAS disk
shelves)
Two FibreBridges are required for each stack of SAS disk shelves.
HA interconnect adapter
InfiniBand adapter (Required only for
systems that do not use an NVRAM5 or
NVRAM6 adapter, which functions as
the HA interconnect adapter.)
FC-VI adapter (Required for the 31xx,
32xx, and 62xx dual-controller systems.)
FC-VI adapter
Note: When the FC-VI adapter is
installed in a 31xx, 32xx or 62xx
systems, the internal InfiniBand
interconnect is automatically
deactivated.
FC-AL or FC HBA (FC
HBA for Disk) adapters
Two or four Fibre Channel HBAs.
Note: The ports on the Fibre Channel HBAs are labeled 1 and 2.
However, the software refers to them as A and B. You see these
labeling conventions in the user interface and system messages
displayed on the console.
Implementing fabric-attached MetroCluster configurations with disks | 163
Required equipment
Stretch MetroCluster configuration
Fabric-attached
MetroCluster
configuration
Fibre Channel switches
Not applicable
Two pairs of Brocade or
Cisco switches
Note: The Fibre
Channel switches must
be of the same type. A
mixture of switch types
(such as Brocade 300
and Brocade 5100
switches) is not allowed.
SFP (Small Form
Pluggable) modules
Not applicable
Two or four SFPs are
required per fabric,
depending on whether you
are using single or dual
inter-switch links. The
type of SFP needed
depends on the distance
between sites. One shortdistance for each switch
port used.
NVRAM adapter media
converter
Only if using fiber cabling.
Not applicable
Cables (provided with
shipment unless otherwise
noted)
•
•
•
•
Four SC/LC (standard connector to
low-profile connector) controller-todisk shelf cables or FibreBridge
6500N
Two SC/LC IB HA adapter cables
Four SC/LC or LC/LC cables
•
Note: For information about required
•
cables, see the Interoperability Matrix
at support.netapp.com/NOW/products/
interoperability.
•
Related information
NetApp Support Site: support.netapp.com
LC/LC controller-toswitch cables
SC/LC (for DS14) or
LC/LC (for DS14mk2
FC) disk shelf-toswitch cables or
FibreBridge 6500N
Two LC/LC interswitch link cables, not
provided in the
shipment
Multiple disk shelf-todisk shelf cables
164 | High Availability and MetroCluster Configuration Guide
Converting an HA pair to a fabric-attached MetroCluster
configuration
With the correct hardware, you can reconfigure an HA pair with DS14mk2 or DS14mk4 disk shelves
to a fabric-attached MetroCluster configuration.
About this task
•
•
•
•
If you are upgrading an existing HA pair to a MetroCluster configuration, you must upgrade disk
firmware to the latest version.
After upgrading disk firmware, you must power-cycle the affected disk drives to ensure that they
work correctly in a fabric-attached MetroCluster configuration. You can download the latest disk
firmware from the NetApp Support Site.
If you are upgrading a 31xx, 32xx, and 62xx dual controller systems, the resulting upgraded
system can only have one controller in each chassis.
If you have a chassis with two controllers, you must move one controller to a new chassis to form
the partner node of the MetroCluster configuration. You must also obtain and install the FC-VI
interconnect card on both systems.
For details about this conversion process, see TR-3548, Best Practices for MetroCluster Design
and Implementation, at the NetApp Support Site.
If you are converting an HA pair that has SAS disk shelves to a fabric-attached MetroCluster
configuration, see Configuring a MetroCluster system with SAS disk shelves and FibreBridge
6500N bridges at the NetApp Support Site.
Steps
1. Update Data ONTAP, storage system firmware, and disk firmware, as described in the Data
ONTAP Upgrade and Revert/Downgrade Guide for 7-Mode, making sure to shut down the nodes
to the boot prompt.
2. Remove any ATA drives in the configuration.
ATA drives are not supported in a MetroCluster configuration.
3. Move the NVRAM adapter and FC-VI adapter to the correct slots for your model, as shown by
the Hardware Universe at the NetApp Support Site.
4. Determine your switch and general configuration by completing the planning worksheet.
5. Set up and configure the local switches, and verify your switch licenses, as described in the
Fabric-attached MetroCluster Systems Brocade Switch Configuration Guide and Fabric-attached
MetroCluster Systems Cisco Switch Configuration Guide.
Note: The configuration and firmware requirements for Brocade and Cisco switches in a
MetroCluster environment are different from the requirements for switches used in SAN
environments. Always refer to MetroCluster documentation when installing and configuring
your switches for a MetroCluster configuration.
Implementing fabric-attached MetroCluster configurations with disks | 165
6. Cable the local node.
7. Enable the Data ONTAP licenses in the following order:
a) cf.mode
Set the mode to ha
b) cf.remote_syncmirror.enable
Set the option to on
8. Configure the local node depending on the type of HA pair:
If you are converting a...
Then...
Standard HA pair
Set up mirroring and configure the local node.
Stretch MetroCluster configuration
Configure the local node.
9. Transport the partner node, disk shelves, and switches to the remote location.
10. Set up the remote node, disk shelves, and switches.
After you finish
Configure the MetroCluster configuration.
Related concepts
Configuring an HA pair on page 72
Recovering from a disaster by using MetroCluster configurations on page 232
Related tasks
Cabling Node A on page 168
Cabling Node B on page 172
Disabling the change_fsid option in MetroCluster configurations on page 79
Related information
NetApp Support Site: support.netapp.com
166 | High Availability and MetroCluster Configuration Guide
Cabling a fabric-attached MetroCluster configuration with
disks
You cable the fabric-attached MetroCluster configuration so that the controller and the disk shelves
at each site are connected to Brocade or Cisco switches. In turn, the switches at one site are
connected through inter-switch links to the switches at the other site.
Before you begin
•
•
•
To cable a fabric-attached MetroCluster configuration, you must be familiar with HA
configuration, the Brocade or Cisco command-line interface, and synchronous mirroring.
You must also be familiar with the characteristics of fabric-attached MetroCluster configurations.
You must also have the following information:
•
•
Correct Brocade or Cisco licenses for each switch
Unique domain IDs for each of the switches
Note: You can use the switch numbers (1, 2, 3, and 4) as the switch Domain ID.
•
Ethernet IP address for both the switches and nodes
Note: The switches ship with a default IP address (10.77.77.77), which you can use if the
switches are not attached to a network.
•
•
Ethernet subnetmask
Gateway address
About this task
A fabric-attached MetroCluster configuration involves two nodes at physically separated sites. To
differentiate between the nodes, the guide refers to the two nodes as Node A and Node B.
Note: If you are using SAS disk shelves, the disk shelves connect to FibreBridge 6500N bridges.
To see an example of a cabled fabric-attached MetroCluster system with FibreBridges and SAS
disk shelves, see the Configuring a MetroCluster system with SAS disk shelves and FibreBridge
6500N bridges at the NetApp Support Site at support.netapp.com.
The following illustration shows a fabric-attached MetroCluster configuration:
Implementing fabric-attached MetroCluster configurations with disks | 167
Steps
Configuring the switches on page 168
Cabling Node A on page 168
Cabling Node B on page 172
Assigning disk pools on page 176
Verifying aggregate status on page 177
Setting up a shared-switches configuration on page 178
Fabric-attached MetroCluster configuration on single-enclosure HA pairs on page 180
Configuration differences for fabric-attached MetroCluster configurations on single-enclosure HA
pairs on page 181
1.
2.
3.
4.
5.
6.
7.
8.
Related concepts
Setup requirements and restrictions for fabric-attached MetroCluster configurations with disks on
page 153
Configuring an HA pair on page 72
Recovering from a disaster by using MetroCluster configurations on page 232
Related tasks
Disabling the change_fsid option in MetroCluster configurations on page 79
168 | High Availability and MetroCluster Configuration Guide
Configuring the switches
To configure the switches, you must refer to the Fabric-attached MetroCluster Brocade Switch
Configuration Guide for your Brocade switch model. The Brocade switch configuration for a
MetroCluster configuration is different than the one used for a SAN configuration.
Step
1. To configure your Brocade switches, see the Fabric-attached MetroCluster Systems Brocade
Switch Configuration Guide for your switch model at the NetApp Support Site.
Note: The configuration and firmware requirements for Brocade switches in a MetroCluster
environment are different from the requirements for switches used in SAN environments.
Always refer to the Interoperability Matrix at support.netapp.com/NOW/products/
interoperability, when installing and configuring your MetroCluster switches.
After you finish
Proceed to configure Node A.
Related information
NetApp Support Site: support.netapp.com
Cabling Node A
To cable the local node (Node A), you need to attach the controller and the disk shelves to the
switches, connect the HA interconnect to the switches, and ensure that the disk shelves in the
configuration belong to the correct pools.
About this task
If you are using SAS disk shelves, the SAS disk shelves connect to the FibreBridge 6500N bridges
and the bridges connect to the switches.
Steps
1. Cabling the controller on page 168
2. Cabling the shelves on page 169
3. Cabling the FC-VI adapter and inter-switch link on page 171
Cabling the controller
You must cable the Fibre Channel ports on the controller to the Brocade or Cisco switches.
About this task
The following illustration shows the cabling of the controller to the Brocade switches:
Implementing fabric-attached MetroCluster configurations with disks | 169
Steps
1. Determine which Fibre Channel ports on your system that you want to use and create a list
showing the order you want to use them.
Note: The numbers in the example refer to the preferred order of usage, not the port ID. For
example, Fibre Channel port 1 might be port 0a on the controller.
2. Cable the first two Fibre Channel ports of Node A to the same numbered ports on Switch 1 and
Switch 2, for example, port 1.
3. Cable the second two Fibre Channel ports of Node A to the same numbered ports on Switch 1 and
Switch 2, for example, port 2.
Note: The switches in the example are 16-port switches.
After you finish
Proceed to cable disk shelves to the switches.
Related concepts
Determining which Fibre Channel ports to use for Fibre Channel disk shelf connections on page 46
Cabling the shelves
You must cable the DS14mk2 or DS14mk4 disk shelf loops on Node A directly to the switches.
About this task
To cable SAS disk shelves and FibreBridge 6500N, see Configuring a MetroCluster system with
SAS disk shelves and FibreBridge 6500N bridges on the NetApp Support Site.
170 | High Availability and MetroCluster Configuration Guide
Note: You can cable a maximum of two disk shelves on each loop.
Steps
1. Connect the Node A pool 0 disk shelves to the switches by completing the following substeps:
a) Connect the Input port of the A module on disk shelf 1 to any available port on Switch 2 other
than ports 0, 4, 8, and 12.
In the example, switch port 3 is used.
b) Connect the Input port of the B module on disk shelf 1 to the same port on Switch 1.
In the example, switch port 3 is used.
c) Connect disk shelf 1 to disk shelf 2 by connecting the Output ports of the module of disk shelf
1 to the Input ports of the corresponding module of the next disk shelf.
d) If your disk shelf modules have terminate switches, set them to Off on all but the last disk
shelf in the disk pool, then set the terminate switches on the last disk shelf to On.
Implementing fabric-attached MetroCluster configurations with disks | 171
Note: ESH4 modules are self-terminating and therefore do not have a terminate switch.
2. Connect the Node B pool 1 disk shelves to the switches by completing the following substeps:
a) Connect the Input port of the module Channel A on disk shelf 3 to any available port on
Switch 2 other than ports 0, 4, 8, and 12.
The example uses switch port 5.
b) Connect the Input port of the module Channel B on disk shelf 3 to the same port on Switch 1.
The example uses switch port 5.
c) Connect disk shelf 3 to disk shelf 4 by connecting the Output ports of the module of disk shelf
3 to the Input ports of the corresponding module of the next disk shelf.
d) If your disk shelf modules have terminate switches, set them to Off on all but the last disk
shelf in the disk pool, then set the terminate switches on the last disk shelf to On.
3. If you have more than one loop, connect the other loops in the same manner.
After you finish
Proceed to cable the FC-VI adapter and inter-switch connections.
Cabling the FC-VI adapter and inter-switch link
You must cable the HA interconnect and inter-switch link on Node A.
About this task
Steps
1. Connect one port of the FC-VI adapter on switch 1 and the second port to the same port on switch
2.
In the example we are using ports 0, 4, 8, and 12, for the FC-VI and inter-switch link connections.
Note: There should be one FC-VI adapter connection for each switch. You must ensure that
you have the FC-VI adapter in the correct slot for your system, as shown in the Hardware
172 | High Availability and MetroCluster Configuration Guide
Universe (formerly the System Configuration Guide) at support.netapp.com/knowledge/docs/
hardware/NetApp/syscfg/index.shtml.
2. Connect an inter-switch link cable to a port on each switch.
In the example, we are using ports 8 and 12 on switch 1 and switch 2 for the inter-switch links.
Note: If you are using dual inter-switch links, traffic isolation must be configured on the
switches.
After you finish
Proceed to cable Node B.
Related information
NetApp Support Site: support.netapp.com
Cabling Node B
To cable the remote node (Node B), you need to attach the controller and the disk shelves to the
switches, connect the HA interconnect to the switches, and ensure that the disk shelves in the
configuration belong to the correct pools.
About this task
If you are using SAS disk shelves, the SAS disk shelves connect to the FibreBridge 6500N bridge
and the bridges connect to the switches.
Steps
1. Cabling the controller on page 172
2. Cabling the shelves on page 173
3. Cabling the FC-VI adapter and inter-switch link on page 175
Cabling the controller
You must cable the Fibre Channel ports on the controller to the Brocade or Cisco switches.
About this task
The following illustration shows the cabling of the controller to the Brocade switches:
Implementing fabric-attached MetroCluster configurations with disks | 173
Steps
1. Determine the Fibre Channel ports on your system that you want to use and create a list showing
the order you want to use them.
Note: The numbers in the example refer to the preferred order of usage, not the port ID. For
example, Fibre Channel port 1 might be port 0a on the controller.
2. Cable the first two Fibre Channel ports of Node B to the same numbered ports Switch 3 and
Switch 4, for example, port 1.
3. Cable the second two Fibre Channel ports of Node B to the same numbered ports Switch 3 and
Switch 4, for example, port 2.
After you finish
Proceed to cable disk shelves to the switches.
Related concepts
Determining which Fibre Channel ports to use for Fibre Channel disk shelf connections on page 46
Cabling the shelves
You must cable the DS14mk2 or DS14mk4 disk shelf loops on Node B directly to the switches.
About this task
To cable SAS disk shelves and FibreBridge 6500N bridges, see Configuring a MetroCluster system
with SAS disk shelves and FibreBridge 6500N bridges on the NetApp Support Site.
174 | High Availability and MetroCluster Configuration Guide
Note: You can cable a maximum of two disk shelves on each loop.
Steps
1. Connect the Node B pool 0 disk shelves to the switches by completing the following substeps:
a) Connect the Input port of the A module on disk shelf 5 to any available port on Switch 4 that
is not reserved for the FC-VI and inter-switch link connections.
The example uses switch port 3.
b) Connect the Input port of the B module on disk shelf 5 to the same port on Switch 3.
The example uses switch port 3.
c) Connect disk shelf 5 to disk shelf 6 by connecting the Output ports of the module of disk shelf
5 to the Input ports of the corresponding module of the next disk shelf.
d) If your disk shelf modules have terminate switches, set them to Off on all but the last disk
shelf in the disk pool, then set the terminate switches on the last disk shelf to On.
Implementing fabric-attached MetroCluster configurations with disks | 175
Note: ESH4 modules are self-terminating and therefore do not have a terminate switch.
2. Connect the Node B pool 1 disk shelves to the switches by completing the following substeps:
a) Connect the Input port of the module Channel A on disk shelf 7 to any available port on
Switch 4 that is not reserved for the FC-VI and inter-switch link connections.
The example uses switch port 5.
b) Connect the Input port of the module Channel B on disk shelf 7 to the same port on Switch 3.
The example uses switch port 5.
c) Connect disk shelf 7 to disk shelf 8 by connecting the Output ports of the module of disk shelf
7 to the Input ports of the corresponding module of the next disk shelf.
d) If your disk shelf modules have terminate switches, set them to Off on all but the last disk
shelf in the disk pool, then set the terminate switches on the last disk shelf to On.
3. If you have more than one loop, connect the other loops in the same manner.
After you finish
Proceed to cable the FC-VI adapter and inter-switch connections.
Related information
NetApp Support Site: support.netapp.com
Cabling the FC-VI adapter and inter-switch link
You must cable the HA interconnect and inter-switch link on Node B.
About this task
Steps
1. Connect one port of the FC-VI adapter to another port that you have reserved for the FC-VI and
inter-switch link connections.
176 | High Availability and MetroCluster Configuration Guide
In the example, port 0 on switch 1 and port 0 on switch 2 is used.
Note: There should be one FC-VI adapter connection for each switch. Make sure that you have
the FC-VI adapter in the correct slot for your system, as shown in the Hardware Universe at the
NetApp Support Site.
2. Connect an inter-switch link cable to a port on each switch.
The example uses port 8 and port 12 on switch 1 and switch 2 for the inter-switch links.
Note: If using dual inter-switch links, traffic isolation must be configured on the switches.
After you finish
Proceed to assign disks to disk pools.
Related information
NetApp Support Site: support.netapp.com
Assigning disk pools
You must assign the attached disk shelves to the appropriate pools.
About this task
You can explicitly assign disks on the attached disk shelves to the appropriate pool with the disk
assign command. Using wildcards in the command enables you to assign all the disks on a disk
shelf with one command.
The following table shows the pool assignments for the disk shelves in the example used in this
section.
Disk shelf...
At site...
Belongs to...
And is assigned to
that node's...
Disk shelf 1
Site A
Node A
Pool 0
Node B
Pool 1
Node B
Pool 0
Node A
Pool 1
Disk shelf 2
Disk shelf 3
Disk shelf 4
Disk shelf 5
Site B
Disk shelf 6
Disk shelf 7
Disk shelf 8
Implementing fabric-attached MetroCluster configurations with disks | 177
Note: Pool 0 always contains the disks that are local to (at the same site as) the storage system that
owns them.
Pool 1 always contains the disks that are remote to the storage system that owns them.
Steps
1. Boot Node A into Maintenance mode.
2. Assign the local disks to Node A pool 0 by entering the following command at the console:
disk assign switch2:port3.* -p 0
This indicates that the disks attached to port 3 of switch 2 are assigned to pool 0. The asterisk (*)
indicates that all disks attached to the port are assigned.
3. Assign the remote disks to Node A pool 1 by entering the following command at the console:
disk assign switch4:port5.* -p 1
This indicates that the disks attached to port 5 of switch 4 are assigned to pool 1. The asterisk (*)
indicates that all disks attached to the port are assigned.
4. Boot Node B into Maintenance mode.
5. Assign the local disks to Node B pool 0 by entering the following command at the console:
disk assign switch4:port3.* -p 0
This indicates that the disks attached to port 3 of switch 4 are assigned to pool 0. The asterisk (*)
indicates that all disks attached to the port are assigned.
6. Assign the remote disks to Node B pool 1 by entering the following command at the console:
disk assign switch2:port5.* -p 1
This indicates that the disks attached to port 5 of switch 2 are assigned to pool 1. The asterisk (*)
indicates that all disks attached to the port are assigned.
After you finish
Proceed to verify the disk paths on the system.
Verifying aggregate status
You must use this procedure to verify your disk paths for configurations using DS14mk2 or
DS14mk4 or SAS shelves.
Steps
1. Boot Node A into normal mode, if necessary.
2. Enter the following command to confirm that your aggregates and volumes are operational and
mirrored:
178 | High Availability and MetroCluster Configuration Guide
aggr status
See the Data ONTAP 7-Mode Storage Management Guide for information about the aggr
status command.
3. Repeat steps 1 and 2 on Node B.
Setting up a shared-switches configuration
In a shared-switches configuration, two fabric-attached MetroCluster configurations share the same
four switches and the ISLs between them.
Shared-switches configuration are cost-effective because you can use the four Brocade switches
between two MetroCluster configurations.
Related concepts
Requirements for a shared-switches MetroCluster configuration with array LUNs on page 188
Cabling the FC-VI adapter and ISL in a shared-switches configuration
Cabling in a shared-switches configuration involves cabling the FC-VI and HBA ports to the
Brocade switches, and cabling the FibreBridge 6500N bridges to the Brocade switches and the SAS
disk shelves.
About this task
In a shared-switches configuration, four nodes join together to form two fabric-attached MetroCluster
configurations. In the illustration, FMC1-1 and FMC1-2 form one fabric-attached MetroCluster
configuration, FMC1. FMC2-1 and FMC2-2 form another fabric-attached MetroCluster
configuration, FMC2.
The following illustration shows the shared-switches configuration:
Implementing fabric-attached MetroCluster configurations with disks | 179
Site A
Site B
FMC1-1
3 1 2 0 8
9
S1
7 6
FB2-1
vol
5
FB2-2
vol
FB2-3
FB2-4
7
S2
6
FCVI_TI_S1S3
17
18
17
18
4
4
FB1-1
FB1-2
vol
vol
FB1-3
5
FB1-4
8 0 2 1 3
9
S3
5
FB1-5
FB1-6
vol
vol
FB1-7
FB1-8
6
7
FB2-5
FB2-6
vol
FB2-7
vol
FibreBridge
6500N
FB2-8
6
5 4
7
17
17
S4
18
18
2 0 83 1 9
9 1 3 8 0 2
FCVI_TI_S2S4
FMC2-1
Site A
FMC1-2
4
FMC2-2
Site B
Steps
1. Connect the ports of the FC-VI adapter to primary and secondary switches for both the storage
systems in the MetroCluster configuration.
In the illustration, the FC-VI adapter from storage system FMC1-1 connects to primary switch S1
and secondary switch S2 through port 0.
The FC-VI adapter from storage system FMC2-1 connects to primary switch S2 and secondary
switch S1 through port 1.
2. Connect the ports of the HBA to the primary and secondary switches.
In the illustration, the HBA from storage system FMC1-1 connects to primary switch S1 and
secondary switch S2 through ports 2 and 8.
The HBA from storage system FMC2-1 connects to primary switch S2 and secondary switch S1
through ports 3 and 9.
180 | High Availability and MetroCluster Configuration Guide
3. If you are using the FibreBridge 6500N bridge, complete the following steps:
For details about installing FibreBridge 6500N bridge as part of your MetroCluster configuration,
see Configuring a MetroCluster system with SAS disk shelves and FibreBridge 6500N bridges,
available at the NetApp Support Site.
a) Connect the FibreBridge 6500N bridge to the switches.
In the illustration, the FibreBridge bridge FB1-1 of the FMC1-1 connects to switch S1 through
port 5.
b) Connect the FibreBridge 6500N bridge to the disk shelves.
4. Repeat steps 1 through 3 at another site.
5. When cabling at both the sites are complete, connect an ISL cable to a port on each switch.
In the illustration, the fabric containing switches S1 and S3 has two ISLs and the fabric
containing switches S2 and S4 also has two ISLs.
The switch S1 connects to switch S3 and switch S2 connects to switch S4 through ports 17 and
18.
After you finish
Configure the traffic isolation zone on the switches. In the illustration, FCVI_TI_S1S3 refers to the
traffic isolation zone created using ports 0, 1, and 17 of switches S1 and S3. Similarly,
FCVI_TI_S2S4 refers to the traffic isolation zone created using ports 0, 1, and 17 of switches S2 and
S4.For more information, see the Fabric-attached MetroCluster Systems Brocade Switch
Configuration Guide.
After cabling the shared-switches configuration, you must define the preferred primary port.
Related information
NetApp Support Site: support.netapp.com
Fabric-attached MetroCluster configuration on single-enclosure HA pairs
You can configure a fabric-attached MetroCluster configuration between a pair of single-enclosure
HA pair systems. In this configuration, the HA pair between the two controllers in each chassis is
deactivated, and two separate, side-by-side MetroCluster configurations are formed between the four
controllers.
When the system detects the presence of an FC-VI adapter, which connects the controller to the
switch fabric, the internal InfiniBand connection is automatically deactivated.
The following figure shows a fabric-attached MetroCluster configuration on single-enclosure HA
pair systems:
Implementing fabric-attached MetroCluster configurations with disks | 181
Note: This is a simplified figure that does not show disk shelf-to-disk shelf connections.
Configuration differences for fabric-attached MetroCluster configurations
on single-enclosure HA pairs
When configuring a fabric-attached MetroCluster configuration between single-enclosure HA pair
(systems with two controllers in the same chassis), you get two separate MetroCluster configurations.
A single-enclosure HA pair can be connected to another HA pair to create two separate fabricattached MetroCluster configurations. The internal InfiniBand connection in each system is
automatically deactivated when the FC-VI card is installed in the controller.
You must cable each fabric-attached MetroCluster configuration separately by using the normal
procedures for each and assign the storage appropriately.
Related concepts
Fabric-attached MetroCluster configuration on single-enclosure HA pairs on page 180
182 | High Availability and MetroCluster Configuration Guide
Setting preferred primary port in a MetroCluster
configuration
After cabling a MetroCluster configuration, you can define the primary port between the four
switches to carry the traffic. You can set the preferred primary port in a shared-switches
configuration by using the ic primary set command.
About this task
By default, the primary port is 0.
Steps
1. To set the primary port, enter the following command:
ic primary set 0|1 [-r]
You must use the -r option for the primary port to take effect immediately.
Note:
•
•
If the port that you have set as the preferred primary port is down, the other port will be
used. Once the port is up, it will be set as the preferred primary port.
If the FC-VI link between the preferred primary ports fails, the fabric-attached
MetroCluster configuration fails over to the secondary FC-VI link. When the primary link
is restored, the fabric-attached MetroCluster configuration again uses the primary link. The
failover temporarily causes the logs to be unsynchronized.
Example
If you want to set the preferred primary port to 1, enter the following command:
ic primary set 1 -r
In a shared-switches configuration, if you have set the preferred primary port of FMC1 (FMC1-1
and FMC1-2) to 1, the preferred primary port for FMC2 (FMC2-1 and FMC2-2) is set to 0.
2. To view the current primary port, enter the following command:
ic primary show
The primary port is displayed.
Implementing fabric-attached MetroCluster configurations with disks | 183
Removing the preferred primary port in a fabric-attached
MetroCluster configuration
You can remove the port that you assigned as primary port in a fabric-attached MetroCluster
configuration.
Step
1. To remove the primary port, enter the following command:
ic primary unset
184 | High Availability and MetroCluster Configuration Guide
Implementing a stretch or fabric-attached
MetroCluster configuration with array LUNs
Prior to Data ONTAP 8.2, stretch and fabric-attached MetroCluster configurations with V-Series
systems could include only array LUNs. Starting in Data ONTAP 8.2, MetroCluster configurations
with V-Series systems can include array LUNs only, native disks only, or both.
Careful planning, a systematic configuration process, and testing is key to a successful
implementation of a MetroCluster configuration. There are requirements and configuration and
testing procedures that are unique to setting up a MetroCluster configuration with V-Series systems
using array LUNs.
If your MetroCluster configuration includes both array LUNs and native disks, complete the setup
and testing of one type of storage before starting configuration of the other type of storage.
Note: MetroCluster configurations are supported only with Data ONTAP operating in 7-Mode.
Related concepts
Planning for a MetroCluster configuration with array LUNs on page 184
Connecting devices in a MetroCluster configuration with array LUNs on page 198
Setting up Data ONTAP after connecting devices in a MetroCluster configuration with array
LUNs on page 215
Testing a MetroCluster configuration with array LUNs on page 215
Planning for a MetroCluster configuration with array LUNs
Creating a detailed plan for your MetroCluster configuration helps you understand the unique
requirements for a MetroCluster configuration that uses LUNs on storage arrays. A plan also helps
you communicate with other people involved in the installation. Installing a MetroCluster
configuration involves connecting and configuring a number of devices, which might be done by
different people.
Related concepts
Implementation overview for a MetroCluster configuration with array LUNs on page 185
Requirements for a MetroCluster configuration with array LUNs on page 186
Recommended fabric-attached MetroCluster configuration with array LUNs on page 188
Recommended stretch MetroCluster configuration with array LUNs on page 192
Cabling guidelines for a MetroCluster configuration with array LUNs on page 195
Planning zoning for a MetroCluster configuration with array LUNs on page 196
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 185
Implementation overview for a MetroCluster configuration with array LUNs
Implementing a MetroCluster configuration to use LUNs from storage arrays requires planning the
implementation, installing hardware, connecting multiple devices, configuring Data ONTAP, and
testing the MetroCluster configuration to ensure that it is operating correctly.
The following tasks must be completed, in the order shown, to set up a MetroCluster configuration to
work with storage arrays. Storage array configuration is performed by the storage array administrator
or the storage array vendor. Zoning is often performed by a switch administrator.
1. Planning your MetroCluster implementation.
2. Setting up the storage array to present array LUNs to Data ONTAP and configuring the
parameters that Data ONTAP requires to for a storage array to work with Data ONTAP.
3. Installing the FC-VI adapter on each V-Series system, if it is not installed already.
Note: New V-Series systems that are ordered for a MetroCluster configuration are shipped
with the correct FC-VI adapter. If you are configuring a MetroCluster configuration with
existing V-Series systems, you might need to install a new adapter or move an existing adapter
to another slot. The Hardware Universe (formerly the System Configuration Guide) at
support.netapp.com/knowledge/docs/hardware/NetApp/syscfg/index.shtml contains
information about the required FC-VI adapter for your platform and the slot to install it in.
4. Connecting the local V-Series system to the fabric.
5. Connecting the remote V-Series system to the fabric.
6. Connecting the switched fabric.
7. Connecting the storage array to the fabric.
8. Configuring zoning.
9. Assigning array LUNs to specific V-Series systems.
10. Configuring Data ONTAP features.
11. Testing the MetroCluster configuration.
Related concepts
Connecting devices in a MetroCluster configuration with array LUNs on page 198
Setting up Data ONTAP after connecting devices in a MetroCluster configuration with array
LUNs on page 215
Testing a MetroCluster configuration with array LUNs on page 215
Related tasks
Connecting the local V-Series systems in a MetroCluster configuration on page 199
Connecting the remote V-Series systems in a MetroCluster configuration on page 203
Connecting the switch fabric in a MetroCluster configuration with array LUNs on page 205
186 | High Availability and MetroCluster Configuration Guide
Connecting the fabric and storage array in a MetroCluster configuration with array LUNs on page
207
Configuring zoning in a MetroCluster configuration with array LUNs on page 213
Requirements for a MetroCluster configuration with array LUNs
There are some unique requirements for setting up a MetroCluster configuration with array LUNs.
The Interoperability Matrix at support.netapp.com contains the latest policies for MetroCluster
configurations that use array LUNs. It includes information about supported storage arrays, switches,
and V-Series systems. The Interoperability Matrix is the final authority for information about
requirements and restrictions for MetroCluster configurations that use array LUNs.
Requirements for V-Series systems
•
•
•
Only V-Series systems (not FAS systems) can be deployed in a MetroCluster configuration that
uses array LUNs.
The V-Series platform must be identified in the Interoperability Matrix as supported for
MetroCluster configurations.
A system with a two controllers in the same enclosure, such as a 32xx system, requires an FC-VI
adapter.
TheHardware Universe contains information about adapter requirements for different models.
Note: In releases prior to Data ONTAP 8.2, MetroCluster configurations with V-Series systems
could not use native disks. Starting in Data ONTAP 8.2, a MetroCluster configuration with VSeries systems can include native disks, storage arrays, or both.
Note: Connecting a single V-Series FC initiator port to multiple target ports is not supported in a
MetroCluster configuration. This functionality is supported for non MetroCluster configurations
starting in Data ONTAP 8.2, as described in the V-Series Installation Requirements and Reference
Guide.
Requirements for storage arrays
•
•
The storage arrays must be identified in the Interoperability Matrix as supported for MetroCluster
configurations.
The storage arrays in the MetroCluster configuration must be symmetric, which means the
following:
•
•
•
The two storage arrays must be in the same vendor family.
The V-Series Implementation Guide for Third-Party Storage contains details about storage
array families.
A pool of array LUNs for the mirrored storage must be created on each of the two storage
arrays.
The array LUNs must be the same size.
Disk types (SATA or SAS) used for mirrored storage must be the same on both storage arrays.
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 187
•
•
Storage arrays that provide tiered configurations (for example, Hitachi) must use the same
tiered topology on each side of the MetroCluster configuration.
The root volume must be mirrored for successful takeover to occur.
Requirements for FC switches
•
•
•
The switches and switch firmware must be identified in the Interoperability Matrix as supported
for MetroCluster configurations.
Each fabric must have two switches.
Each V-Series system must be connected to storage using redundant components so that there is
redundancy in case of device failure.
The Interoperability Matrix contains the latest information about switch requirements and supported
switches. The configuration and firmware requirements for switches in a MetroCluster environment
might differ from those in other V-Series system configurations.
Zoning requirements
•
•
Single-initiator to single-target zoning is recommended.
Single-initiator to single-target zoning limits each zone to a single V-Series FC initiator port. It
also improves discovery and boot time because the V-Series FC initiators do not attempt to
discover each other.
FC-VI ports on the FC adapters must be zoned end-to-end across the fabric.
The Interoperability Matrix contains specific guidelines for FC-VI.
Requirements for ISLs
Data ONTAP supports using one or two ISLs, depending on the configuration. The Interoperability
Matrix contains information about the number of ISLs for specific configurations.
For specific information about Traffic Isolation zoning for FC-VI traffic only, see the “Brocade's
Zoning Feature” section in the Fabric-attached MetroCluster Systems Brocade Switch Configuration
Guide on the NetApp Support Site at support.netapp.com. Regular zoning guidelines apply for VSeries systems and storage arrays.
SyncMirror requirements
•
•
SyncMirror is required for a MetroCluster configuration.
Licenses are no longer requried for SyncMirror. You need to enable SyncMirror with the
cf.mode and cf.remote_syncmirror.enable options.
Two separate storage arrays are required for the mirrored storage.
Note: Using two different storage arrays for mirroring is optional in configurations that are not
MetroCluster configurations.
•
Two sets of LUNs are required—one set for the aggregate on the local storage array (pool0) and
another set at the remote storage array for the mirror of the aggregate (the other plex of the
aggregate, pool1).
188 | High Availability and MetroCluster Configuration Guide
The Data ONTAP Data Protection Online Backup and Recovery Guide for 7-Mode provides more
details about planning for setting up SyncMirror for a MetroCluster configuration that uses array
LUNs.
Requirements for a shared-switches MetroCluster configuration with array
LUNs
There are specific requirements for creating a shared-switches MetroCluster configuration with array
LUNs.
You need the following to create a shared-switches MetroCluster configuration with array LUNs:
•
•
In releases prior to Data ONTAP 8.2, four Brocade 5100 switches. In Data ONTAP 8.2 and later,
four Brocade 5100 switches or 6510 switches.
Two storage arrays at each site.
One storage array at one site is used for the first MetroCluster configuration and the other storage
array at the site is used for the second MetroCluster configuration.
Note: A single storage array at each site is supported if the target ports on the storage array and
the array LUNs are not shared between the two MetroCluster configurations.
•
A minimum of two ISLs per fabric with TI enabled.
For more information about configuring Brocade switches, see the Fabric-attached MetroCluster
Systems Brocade Switch Configuration Guide.
Recommended fabric-attached MetroCluster configuration with array LUNs
When configuring a fabric-attached MetroCluster configuration with V-Series systems and storage
arrays, you should follow the best practice recommendations for how to connect the hardware
components. The Interoperability Matrix at support.netapp.com contains information about the
hardware components that are supported for a V-Series MetroCluster configuration without a PVR.
The following illustration shows the components and best practice configuration of a fabric-attached
MetroCluster configuration that uses array LUNs from storage arrays. The fabric-attached
MetroCluster configuration shown provides the same single-point-of-failure protections that are
available for all mirrored HA pairs.
Note: V-Series FC initiator port names differ among some platforms. Your platform's port names
might not match the port names shown in the illustration.
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 189
Site A (Local)
vs1
0a
1
0b
FC-VI
2
3
Switch 1
4 6
5
Site B (Remote)
vs2
0c
0d
0a
6
13
11 12
Switch 2
15
14
0b
0d
18
16 17
Switch 4
20
19
21
ISL (Fabric 2)
2A
2A
1A
LUNs 1-4
LUNs 1-4
2B
1B
0c
7 8
Switch 3
9 11 10
ISL (Fabric 1)
16
1A
FC-VI
2B
1B
Storage array 1
Storage array 2
The following sections describe the connections for the fabric-attached MetroCluster configuration in
the previous illustration.
V-Series interconnections
The V-Series systems in the sample MetroCluster illustration are configured to be an HA pair. They
are interconnected by connecting ports A and B on each system's FC-VI adapter to alternate switch
fabrics.
V-Series system
Switch
Fabric
vs1: Port A
vs1: Port B
Switch 2, Port 12
Switch 1, port 2
2
1
vs2: Port A
vs2: Port B
Switch 4, port 17
Switch 3, port 7
2
1
If your V-Series systems are dual-enclosure systems with two controllers in the same chassis (such as
a 32xx system), you connect the top controller in one system to the top controller in the other system
190 | High Availability and MetroCluster Configuration Guide
through the fabric. You connect the bottom controllers between the fabric in the same way. These
two separate connections result in two separate fabric-attached MetroCluster configurations.
Note: The internal InfiniBand connection in each system is automatically deactivated when the
FC-VI adapter is installed in the controller.
The following illustration shows the connections of the V-Series controllers to the switch fabric for
dual-enclosure systems with two controllers in the same chassis.
LAN
Controller C
Controller A
Brocade Switch Fabric
PSU
Two dedicated fabrics
with 1 or 2 inter-switch
links and two switches per fabric
PSU
Brocade Switch Fabric
Two dedicated fabrics
with 1 or 2 inter-switch
links and two switches per fabric
Controller B
PSU
PSU
Controller D
Site 2
Site 1
Inter-Switch Link connections (ISLs)
Fabric-attached MetroCluster configurations use a switched fabric to connect the local half of the
configuration to the remote half of the configuration. The sample MetroCluster illustration shows the
following:
•
•
Switches 1 and 3 are connected to each other (Fabric 1).
The first fabric in the MetroCluster configuration begins from Switch 1 on Site A (local) and is
completed by connecting the ISL cable to the first switch on Site B (remote)—Switch 3.
Switches 2 and 4 are also connected (Fabric 2).
The second fabric is created using Switch 2 on Site A (local), connected through a second ISL
cable, to the second switch on Site B (remote)—Switch 4.
The following table lists the ISLs in this configuration.
ISL connection
Switch
Fabric
Port 5 on switch 1
Port 10 on switch 3
Fabric 1
Port 14 on switch 2
Port 19 on switch 4
Fabric 2
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 191
V-Series system-to-switch connections
The best practice connections in the recommended MetroCluster illustration eliminate a single point
of failure in the following ways:
•
•
FC initiator ports on the same FC controller chip (for example, port 0a and 0b) connect to
alternate fabrics.
Multiple paths and zones ensure that FC initiator ports on the same controller chips access the
array LUN from different V-Series systems and switch fabrics.
The following table lists the connections from the V-Series system to the switch.
V-Series system port
Switch
vs1: FC port 0a
Switch 1: Port 1
vs1: FC port 0b
Switch 2: Port 11
vs1: FC port 0c
Switch 1: Port 3
vs1: FC port 0d
Switch 2: Port 13
vs2: FC port 0a
Switch 3: Port 6
vs2: FC port 0b
Switch 4: Port 16
vs2: FC port 0c
Switch 3: Port 8
vs2: FC port 0d
Switch 4: Port 18
Storage array-to-switch connections
Best practice connections from the storage array to the switch are as follows:
•
•
•
Ports 1A and 2A on each storage array connect to alternate fabrics.
Ports 1B and 2B on each storage array connect to alternate fabrics.
V-Series systems are configured to access any array LUN on two storage array paths (1A and 2A
or 1B and 2B).
Storage array port
Switch
Fabric
Array 1: Port 1A
Switch 1: Port 4
1
Array 1: Port 2A
Switch 2: Port 15
2
Array 1: Port 1B
Switch 1: Port 6
1
Array 1: Port 2B
Switch 2: Port 16
2
Array 2: Port 1A
Switch 3: Port 9
1
Array 2: Port 2A
Switch 4: Port 20
2
192 | High Availability and MetroCluster Configuration Guide
Storage array port
Switch
Fabric
Array 2:Port 1B
Switch 3: Port 11
1
Array 2:Port 2B
Switch 4: Port 21
2
Recommended stretch MetroCluster configuration with array LUNs
When configuring a stretch MetroCluster configuration with V-Series systems and storage arrays,
you should follow the best practice recommendations for how to connect the hardware components.
The Interoperability Matrix at support.netapp.com contains information about the hardware
components that are supported for a V-Series MetroCluster configuration.
The following illustration shows the components and best practice for a stretch MetroCluster
configuration with storage arrays.
Note: V-Series FC initiator port names differ between some platforms. Your platform's port names
might not match the port names shown in the illustration. The Hardware Universe contains
information that helps you determine in which slot to install the FC-VI adapter for your platform.
Site A (Local)
vs1
0a
0b
0c
FC-VI
1
2
Switch 1
3 5 4
Site B (Remote)
vs2
0d
10
Switch 2
11
12
9
0a
0b
5
6
Switch 3
7 9 8
2A
14
Switch 4
17
2A
1A
LUNs 1-4
2B
Storage array 1
0d
ISL (Fabric 2)
LUNs 1-4
1B
13
0c
15
ISL (Fabric 1)
13
1A
FC-VI
2B
1B
Storage array 2
16
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 193
V-Series system interconnections
In a stretch MetroCluster configuration, the V-Series systems are connected directly.
The sample illustration shows the connections between a dual-enclosure HA pair for systems that can
contain only one controller in the enclosure.
For V-Series systems that are dual-enclosure systems with two controllers in the same chassis, such
as 31xx systems, you install a V-Series system at each site. To implement the stretch MetroCluster
configuration, you must install an FC-VI adapter in each controller to provide the HA interconnect
between the systems.
Note: When the FC-VI adapter is installed in the system, the internal InfiniBand interconnect is
automatically disabled.
The following illustration shows the FC-VI port connections between HA pair systems such as the
31xx. Each enclosure contains two controllers. You configure two separate MetroCluster
configurations among the four controllers.
In the stretch MetroCluster illustration, one MetroCluster configuration is between Controller A and
Controller C.
From Controller A
To Controller C
FC-VI port A
FC-VI port A
FC-VI port B
FC-VI port B
In the stretch MetroCluster illustration, the other MetroCluster configuration is between Controller B
and Controller D.
From Controller B
To Controller D
FC-VI port A
FC-VI port A
FC-VI port B
FC-VI port B
Some older V-Series models use NVRAM to connect the systems in a MetroCluster configuration.
The following table lists the NVRAM connections on V-Series systems that connect to each other
through NVRAM.
194 | High Availability and MetroCluster Configuration Guide
vs1
vs2
NVRAM port L02 Ph2
NVRAM port L02 Ph2
NVRAM port L01 Ph1
NVRAM port L01 Ph1
Inter-Switch Link connections (ISLs)
Stretch MetroCluster configurations use a switched fabric to connect the local half of the
configuration to the remote half of the configuration. In the stretch MetroCluster illustration,
Switches 1 and 3 are connected to each other (Fabric 1). Switches 2 and 4 are also connected (Fabric
2).
•
•
The first fabric in the MetroCluster configuration begins from Switch 1 on Site A (local) and is
completed by connecting the ISL cable to the first switch on Site B (remote)—Switch 3.
The second fabric is created using Switch 2 on Site A (local), connected through a second ISL
cable to the second switch on Site B (remote)—Switch 4.
ISL connection
Switch
Fabric
Port 4 on switch 1
Port 8 on switch 3
Fabric 1
Port 11 on switch 2
Port 15 on switch 4
Fabric 2
V-Series system-to-switch connections
The best practice connections in the stretch MetroCluster with storage arrays illustration eliminate a
single-point-of-failure in the following ways:
•
•
FC initiator ports on the same FC controller chip (for example, port 0a and 0b) connect to
alternate fabrics.
Multiple paths and zones ensure that FC initiator ports on the same controller chips access the
array LUN from different V-Series systems and switch fabrics.
The following table lists the connections from the V-Series system to the switch.
V-Series system port
Switch
vs1: FC port 0a
Switch 1: Port 1
vs1: FC port 0b
Switch 2: Port 9
vs1: FC port 0c
Switch 1: Port 2
vs1: FC port 0d
Switch 2: Port 10
vs2: FC port 0a
Switch 3: Port 5
vs2: FC port 0b
Switch 4: Port 13
vs2: FC port 0c
Switch 3: Port 6
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 195
V-Series system port
Switch
vs2: FC port 0d
Switch 4: Port 14
Storage array-to-switch connections
The best practice connections from the storage array to the switch are as follows:
•
•
•
Ports 1A and 2A on each storage array connect to alternate fabrics.
Ports 1B and 2B on each storage array connect to alternate fabrics.
V-Series systems are configured to access any array LUN on two storage array paths (1A and 2A
or 1B and 2B).
Storage array port
Switch
Fabric
Array 1: Port 1A
Switch 1: Port 3
1
Array 1: Port 2A
Switch 2: Port 12
2
Array 1: Port 1B
Switch 1: Port 5
1
Array 1: Port 2B
Switch 2: Port 13
2
Array 2: Port 1A
Switch 3: Port 7
1
Array 2: Port 2A
Switch 4: Port 16
2
Array 2: Port 1B
Switch 3: Port 9
1
Array 2: Port 2B
Switch 4: Port 17
2
Related concepts
Connecting devices in a MetroCluster configuration with array LUNs on page 198
Cabling guidelines for a MetroCluster configuration with array LUNs
There are a number of cabling guidelines that you need to review before connecting the devices in a
MetroCluster configuration with array LUNs.
Cabling guidelines for a dual-enclosure HA pair with multiple controllers in the same
enclosure
Some V-Series models, for example, 31xx systems, support two controllers in the same enclosure.
For a MetroCluster configuration, you must have two such systems. You configure the two systems
into a pair of MetroCluster configurations by connecting the FC-VI adapter between the two top
controllers and then connecting the FC-VI adapter between the bottom controllers
In such a configuration, the internal InfiniBand connections between the controllers in the same
enclosure are automatically deactivated. Therefore, the two controllers in the enclosure are no longer
in an HA pair with each other. Each controller is connected through FC-VI connections to another
196 | High Availability and MetroCluster Configuration Guide
controller of the same type, so the four controllers form two independent MetroCluster
configurations.
Cabling guidelines for partner-to-partner connections
The cabling for the MetroCluster partners is different on each node of the HA pair. Use the following
guidelines to plan your cabling for partner-to-partner connections.
•
•
•
Each node in the HA pair must have a different name so that you can distinguish them.
One node needs to be identified as being in Site A and the other as being in Site B.
For example, the local partner could be V-Series system 1, Site A, and the remote partner could
be V-Series system 2, Site B.
Each port on the node must be connected to the same fabric.
For example, if Port A of the FC-VI adapter on the local node is connected to Switch 2 and Port
A of the FC-VI adapter on the remote node is connected to Switch 4, then Switch 2 and Switch 4
must be connected by the ISL, thereby connecting the nodes to the same fabric.
Cabling guidelines for V-Series system-to-switch connections
•
FC ports on the same channel controller chip cannot be connected to the same switch.
One port must be connected to one switch and the other port must be connected to the other
switch. For example, if onboard port 0a is connected to Switch 3, you cannot connect onboard
port 0b to Switch 3; port 0b must be connected to Switch 4.
Note: Connecting both onboard ports of the pair to the same switch port number can simplify
cabling and management of the MetroCluster configuration. For example, if port 0a is
connected to Port 1, Switch 1, connect port 0b to Port 1, Switch 2.
•
All switches within a fabric must be the same switch model and have the same number of
licensed ports.
Cabling guidelines for Inter-Switch connections
You can connect an Inter-Switch Link (ISL) to any available switch port.
Cabling guidelines for V-Series system-to-storage array connections
Cabling between the V-Series systems and the storage arrays must be redundant and conform to the
required number of paths (two).
Planning zoning for a MetroCluster configuration with array LUNs
Switch zoning defines paths between connected nodes based on the node’s unique WWN. Sketching
out the zoning plan enables you to correct errors before zones are configured, and makes it easier to
communicate the zoning information to the person who configures the switches.
You can use the following example as a reference when determining your zoning plan. The example
shows single-initiator zoning for a fabric-attached MetroCluster configuration. The lines in the
following example represent zones rather than connections; each line is labeled with its zone number.
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 197
Site A (Local)
vs1
0a
z1
0b
FC-VI
z3
1
Site B (Remote)
vs2
0c
0d
z5
0a
z7
z2
2
3
Switch 1
4
5
0b
FC-VI
z4
0c
z6
7 8
Switch 3
9
10
0d
z8
6
13
11 12
Switch 2
14
15
Fabric 1
18
16 17
Switch 4
20
19
Fabric 2
z1
z7
z6
1A
LUNs 1-4
1B
Storage array 1
z4 z2
z5
2A
1A
2B
1B
z3
LUNs 1-4
z8
2A
2B
Storage array 2
In the sample illustration, four array LUNs are allocated on each storage array for the MetroCluster
configuration. LUNs of equal size are provisioned on the storage arrays at both sites, which is a
SyncMirror requirement. Each FC initiator port on each V-Series system has a path to each V-Series
LUN on the storage arrays. The ports on the storage array are redundant, and are configured as
follows:
•
Storage array 1
•
• Ports 1A and 2A are a redundant port pair.
• Ports 1B and 2B are a redundant port pair.
• In each port pair, both ports can access LUNs 1 through 4 because they are alternate paths.
Storage array 2
•
•
•
Ports 1A and 2A are a redundant port pair.
Ports 1B and 2B are a redundant port pair.
In each port pair, both ports can access LUNs 1 through 4 because they are alternate paths.
Switches are zoned so that there are only two paths to each array LUN, one unique path from each VSeries FC initiator port through each switch. If there are multiple connections between a V-Series
198 | High Availability and MetroCluster Configuration Guide
system and the switch, the best practice recommendation is to put each connection into a separate
zone.
The following table shows the zones for this example:
Zone
V-Series FC initiator port
Storage array port
z1
vs1:Port 0a
Storage array 1:Port 1A
z5
vs1:Port 0c
Storage array 2:Port 1B
z3
vs1:Port 0b
Storage array 2:Port 2B
z7
vs1:Port 0d
Storage array 1:Port 2A
z2
vs2:Port 0a
Storage array 2:Port 1A
z6
vs2:Port 0c
Storage array 1:Port 1B
z8
vs2:Port 0d
Storage array 2:Port 2A
z4
vs2:Port 0b
Storage array 1:Port 2B
Switch 1
Switch 2
Switch 3
Switch 4
Related tasks
Configuring zoning in a MetroCluster configuration with array LUNs on page 213
Testing zoning of FC-VI ports in a MetroCluster configuration with array LUNs on page 215
Connecting devices in a MetroCluster configuration with
array LUNs
You should plan the cabling required for your MetroCluster configuration before you start
connecting the devices. It is helpful to have a port-to-port connectivity diagram to use for reference
while you are connecting the devices in a MetroCluster configuration with array LUNs.
Note: The instructions for connecting devices in a MetroCluster configuration use the term FC-VI
adapter. Alternative terms for FC-VI adapter are VI-MC adapter and VI-MetroCluster adapter.
Steps
1. Connecting the local V-Series systems in a MetroCluster configuration on page 199
2. Connecting the remote V-Series systems in a MetroCluster configuration on page 203
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 199
3. Connecting the switch fabric in a MetroCluster configuration with array LUNs on page 205
4. Connecting the fabric and storage array in a MetroCluster configuration with array LUNs on page
207
5. Setting up a shared-switches configuration on page 209
6. Setting preferred primary port in a MetroCluster configuration on page 212
7. Removing the preferred primary port in a fabric-attached MetroCluster configuration on page 213
8. Configuring zoning in a MetroCluster configuration with array LUNs on page 213
9. Changing the configuration speed of a stretch MetroCluster configuration on page 214
Related concepts
Requirements for a MetroCluster configuration with array LUNs on page 186
Recommended fabric-attached MetroCluster configuration with array LUNs on page 188
Recommended stretch MetroCluster configuration with array LUNs on page 192
Cabling guidelines for a MetroCluster configuration with array LUNs on page 195
Setting up Data ONTAP after connecting devices in a MetroCluster configuration with array
LUNs on page 215
Testing a MetroCluster configuration with array LUNs on page 215
Connecting the local V-Series systems in a MetroCluster configuration
The first stage in connecting the devices in a V-Series MetroCluster configuration is to connect a
local V-Series system to the fabric.
Before you begin
•
•
MetroCluster configuration planning must be complete.
Storage array LUNs must already be created and presented to Data ONTAP.
About this task
Use the following fabric-attached MetroCluster illustration as a reference for connecting the local VSeries systems in a MetroCluster configuration. Differences for stretch MetroCluster configurations
are noted in the procedure.
200 | High Availability and MetroCluster Configuration Guide
Site A (Local)
vs1
0a
1
4
0b
FC-VI
Site B (Remote)
vs2
0c
2
3
Switch 1
6
0d
0a
6
13
11 12
Switch 2
14
15
ISL (Fabric 1)
5
0b
2A
2A
1A
LUNs 1-4
2B
Storage array 1
0d
18
16 17
Switch 4
20
19
21
ISL (Fabric 2)
LUNs 1-4
1B
0c
7 8
Switch 3
9 11 10
16
1A
FC-VI
2B
1B
Storage array 2
Note: A port pair consists of two FC initiator ports that are used to access the same set of array
LUNs. For example, V-Series FC initiators 0a and 0d are accessing the same LUNs. See the VSeries Installation Requirements and Reference Guide for specific information about port pairs for
your platform.
Steps
1. On the local V-Series system, locate the NVRAM or FC-VI interconnect module.
2. Take one of the following actions to connect the local V-Series system to the remote V-Series
system.
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 201
For a...
Do this...
Fabric-attached
configuration
a. Cable FC-VI Port A to one fabric (Port A to Switch 2 in the previous
illustration).
b. Cable FC-VI Port B to the alternate fabric (Port B to Switch 1 in the previous
illustration).
c. For dual-enclosure systems, repeat substeps a and b for the bottom
controllers.
Stretch configuration
with an FC-VI adapter
Connect the cables between the FC-VI ports directly (point-to-point).
Stretch configuration
with an NVRAM
adapter
a. Install a copper-to-fiber converter in L01 Ph1, and cable L01 Ph1 on the local
node to L01 Ph1 on the remote node.
b. Install a copper-to-fiber converter in L02 Ph2, and cable L02 Ph2 on the local
node to L02 Ph2 on the remote node.
Dual-enclosure HA
pair systems with an
FC-VI adapter
a. Connect port A of the FC-VI adapter on the top controller of the local site
(vs1) to port A of the corresponding FC-VI adapter at the remote site (vs2).
b. Connect port B of the FC-VI adapter on the top controller of the local site
(vs1) to port B of the corresponding FC-VI adapter at the remote site (vs2).
c. Repeat substeps a and b for connecting the FC-VI adapter on the bottom
controller.
3. Connect the V-Series FC initiator ports to the switched fabric:
a) Identify one of the two FC initiator port pairs on the V-Series system.
b) Cable one port of the FC initiator port pair to one fabric.
c) Cable the other port of the pair to the alternate fabric.
In the sample illustration, these are ports 0a and 0b.
d) Identify the other onboard FC initiator port pair on the V-Series system.
e) Cable one port of the pair to one fabric.
f) Cable another port of the pair to the alternate fabric.
In the sample illustration, these are ports 0c and 0d.
4. Connect the V-Series system to a tape backup device through a separate FC initiator port or SCSI
tape adapter.
5. Connect a console cable to the console port on the V-Series system, then connect the console
cable to the adapter.
Use the RJ-45 to DB-9 adapter that is included with your system.
6. Install the cable management tray:
a) Pinch the arms of the tray and fit the holes in the arms through the motherboard tray pins.
202 | High Availability and MetroCluster Configuration Guide
b) Push the cables into the cable holders, thread the adapter cables through the top rows of the
cable holders, and then thread the port cables through the lower cable holders.
7. Connect the V-Series system to the Ethernet network by plugging the network cable into the
networking port.
If you are connecting more than one network cable to the network, connect to the ports
sequentially. Use the cable management tray to direct all the cabling from your system.
8. Connect the remote management device from the back of the V-Series system to the network
using an Ethernet cable.
The network switch port for the remote management device connection must negotiate down to
10/100 or autonegotiate.
9. If applicable, turn on any tape backup devices.
10. For each power supply on the V-Series system, take the following steps:
a)
b)
c)
d)
Ensure that the power switch is in the Off (0) position.
Connect the socket end of the power cord to the power plug on the power supply.
Secure the power cord with the retaining adjustable clip on the power supply.
Plug the other end of the power cord into a grounded electrical outlet.
Note: To obtain power supply redundancy, you must connect the second power supply to a
separate AC circuit.
11. Start a communications program.
You must use some form of communications program to perform initial network setup and VSeries configuration. You can start a communications program through the remote management
device or through the console after connecting to the serial port.
After you finish
Connect the remote V-Series system to the fabric.
Related concepts
Recommended fabric-attached MetroCluster configuration with array LUNs on page 188
Recommended stretch MetroCluster configuration with array LUNs on page 192
Cabling guidelines for a MetroCluster configuration with array LUNs on page 195
Related tasks
Connecting the remote V-Series systems in a MetroCluster configuration on page 203
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 203
Connecting the remote V-Series systems in a MetroCluster configuration
After connecting the local V-Series systems to the fabric, you need to connect the remote systems to
the fabric.
Before you begin
The local V-Series systems must be connected to the fabric.
About this task
Use the following illustration of a fabric-attached MetroCluster configuration as a reference for
connecting the remote V-Series systems. Differences for stretch MetroCluster configurations are
noted in the procedure.
Note: A port pair consists of two FC initiator ports that are used to access the same set of array
LUNs. For example, V-Series FC initiators 0a and 0d are accessing the same LUNs. See the VSeries Installation Requirements and Reference Guide for specific information about port pairs.
Site A (Local)
vs1
0a
1
0b
FC-VI
Site B (Remote)
vs2
0c
0d
2
3
Switch 1
4 6
0a
6
13
11 12
Switch 2
14
15
ISL (Fabric 1)
5
0b
7 8
Switch 3
9 11 10
2A
2A
1A
LUNs 1-4
2B
Storage array 1
18
16 17
Switch 4
ISL (Fabric 2)
LUNs 1-4
1B
0d
19 21
16
1A
0c
FC-VI
2B
1B
Storage array 2
20
204 | High Availability and MetroCluster Configuration Guide
Steps
1. Connect the local V-Series system to the remote V-Series system.
For a...
Do this...
Fabric-attached configuration
a. Cable FC-VI Port A to one fabric (Port A to Switch 4 in the sample
illustration).
b. Cable FC-VI Port B to the alternate fabric (Port B to Switch 3 in the
sample illustration).
Stretch configuration with an
FC-VI card
Use a point-to-point connection.
2. Connect the V-Series FC initiator ports to the switched fabric:
a) Identify one of the two FC initiator port pairs on the V-Series system.
b) Cable one port of the FC initiator port pair to one fabric.
c) Cable the other port of the pair to the alternate fabric.
In the sample illustration, these are ports 0a and 0b.
d) Identify the other FC initiator port pair on the V-Series system.
e) Cable one port of the pair to one fabric.
f) Cable another port of the pair to the alternate fabric.
In the sample illustration, these are ports 0c and 0d.
3. Connect the V-Series system to a tape backup device through a separate FC initiator port or SCSI
tape adapter.
4. Connect a console cable to the console port on the V-Series system.
Use the RJ-45 to DB-9 adapter that is included with your system. Connect the console cable to
the adapter.
5. Install the cable management tray.
a) Pinch the arms of the tray and fit the holes in the arms through the motherboard tray pins.
b) Push the cables into the cable holders, thread the adapter cables through the top rows of the
cable holders, and then thread the port cables through the lower cable holders.
6. Connect the V-Series system to the Ethernet network by plugging the network cable into the
networking port.
If you are connecting more than one network cable to the network, connect to the ports
sequentially. Use the cable management tray to direct all the cabling from your system.
7. Connect the remote management device from the back of the V-Series system to the network
using an Ethernet cable.
The network switch port for the remote management device connection must negotiate down to
10/100 or autonegotiate.
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 205
8. If applicable, turn on any tape backup devices.
9. For each power supply on the V-Series system, take the following steps:
a)
b)
c)
d)
Ensure that the power switch is in the Off (0) position.
Connect the socket end of the power cord to the power plug on the power supply.
Secure the power cord with the retaining adjustable clip on the power supply.
Plug the other end of the power cord into a grounded electrical outlet.
Note: To obtain power supply redundancy, you must connect the second power supply to a
separate AC circuit.
10. Start a communications program.
You must use some form of communications program to perform initial network setup and VSeries configuration. You can start a communications program through the remote management
device or through the console after connecting to the serial port.
After you finish
Connect the switch fabric in the MetroCluster configuration.
Related concepts
Recommended fabric-attached MetroCluster configuration with array LUNs on page 188
Recommended stretch MetroCluster configuration with array LUNs on page 192
Cabling guidelines for a MetroCluster configuration with array LUNs on page 195
Related tasks
Connecting the switch fabric in a MetroCluster configuration with array LUNs on page 205
Connecting the switch fabric in a MetroCluster configuration with array
LUNs
Connecting the switch fabric involves connecting the ISL cables and applying power to the switches.
Before you begin
The local and remote V-Series systems must be connected to the fabric.
About this task
The following illustration provides a reference for connecting the switch fabric.
206 | High Availability and MetroCluster Configuration Guide
Site A (Local)
vs1
0a
1
4
0b
FC-VI
2
3
Switch 1
6
5
Site B (Remote)
vs2
0c
0d
0a
6
13
11 12
Switch 2
15
14
0b
7 8
Switch 3
9 11 10
ISL (Fabric 1)
2A
0d
18
16 17
Switch 4
20
19
21
2A
1A
LUNs 1-4
LUNs 1-4
2B
1B
0c
ISL (Fabric 2)
16
1A
FC-VI
Storage array 1
2B
1B
Storage array 2
Steps
1. Connect the switched fabric:
a) If you are configuring a new switch fabric, set the domain ID to a unique domain ID for each
switch in the fabric.
See your switch documentation for details.
b) Connect an ISL cable to a switch on one fabric and to another switch on the same fabric.
In the sample illustration, Fabric 1, Switch 1, Port 5 connects to Fabric 1, Switch 3, Port 10.
c) Connect an ISL cable on a switch on the alternate fabric to another switch on the alternate
fabric.
In the sample illustration, Fabric 2, Switch 2, Port 14 connects to Fabric 2, Switch 4, Port 19.
Note: You must install a long distance SFP adapter in each port that you use to connect an ISL
cable. You might also need to install an additional switch license to provide ISL support.
2. Make sure that all switch IDs are set, and then turn on each switch 10 minutes apart from one
another.
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 207
After you finish
Connect the fabric and the storage arrays.
Related concepts
Recommended fabric-attached MetroCluster configuration with array LUNs on page 188
Recommended stretch MetroCluster configuration with array LUNs on page 192
Cabling guidelines for a MetroCluster configuration with array LUNs on page 195
Related tasks
Connecting the fabric and storage array in a MetroCluster configuration with array LUNs on page
207
Connecting the fabric and storage array in a MetroCluster configuration
with array LUNs
All storage arrays, regardless of model, must be configured to allow Data ONTAP to access a
specific LUN on two (primary and secondary) storage array ports.
Before you begin
•
•
•
The local and remote V-Series systems must be connected to the fabric.
The fabric must be connected and the switches must be powered up.
Your storage array documentation verified that the ports you plan to use do not access more than
the number of array LUNs and host groups that are supported for that storage array model.
About this task
The following illustration provides a reference for connecting the fabric and the storage arrays.
208 | High Availability and MetroCluster Configuration Guide
Site A (Local)
vs1
0a
1
4
0b
FC-VI
2
3
Switch 1
6
5
Site B (Remote)
vs2
0c
0d
0a
6
13
11 12
Switch 2
15
14
0b
FC-VI
7 8
Switch 3
9 11 10
ISL (Fabric 1)
2A
2A
LUNs 1-4
2B
Storage array 1
18
16 17
Switch 4
19 21 20
1A
LUNs 1-4
1B
0d
ISL (Fabric 2)
16
1A
0c
2B
1B
Storage array 2
Steps
1. Connect the ports on the storage array at Site A:
a) Connect storage array controller 1A to any port on one fabric.
In the sample illustration, this is Switch 1, Port 4, Fabric 1.
b) Connect controller 2A to any port on the alternate fabric.
In the sample illustration this is Switch 2, Port 15, Fabric 2.
c) Connect storage array controller 1B to any port on one fabric.
In the sample illustration, this is Switch 1, Port 6, Fabric 1.
d) Connect controller 2B to any port on the alternate fabric.
In the sample illustration this is Switch 2, Port 16, Fabric 2.
e) Connect additional controller ports and fabrics, as required by your MetroCluster
configuration.
2. Connect the ports on the storage array at Site B:
a) Connect storage array controller 1A to any port on one fabric.
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 209
In the sample illustration, this is Switch 3, Port 9, Fabric 1.
b) Connect controller 2A to any port on the alternate fabric.
In the sample illustration, this is Switch 4, Port 20, Fabric 2.
c) Connect additional controller ports and fabric, as required by your MetroCluster
configuration.
d) Connect storage array controller 1B to any port on one fabric.
In the sample illustration, this is Switch 3, Port 11, Fabric 1.
e) Connect controller 2B to any port on the alternate fabric.
In the sample illustration, this is Switch 4, Port 21, Fabric 2.
After you finish
Next configure zoning.
Related concepts
Recommended fabric-attached MetroCluster configuration with array LUNs on page 188
Recommended stretch MetroCluster configuration with array LUNs on page 192
Cabling guidelines for a MetroCluster configuration with array LUNs on page 195
Related tasks
Configuring zoning in a MetroCluster configuration with array LUNs on page 213
Setting up a shared-switches configuration
In a shared-switches configuration, two fabric-attached MetroCluster configurations share the same
four switches and the ISLs between them.
Shared-switches configuration are cost-effective because you can use the four Brocade switches
between two MetroCluster configurations.
Cabling the FC-VI adapter and ISL in a shared-switches MetroCluster configuration with
array LUNs
Cabling in a shared-switches MetroCluster configuration with array LUNs involves cabling the FCVI and HBA ports to the switches, and cabling the switches to the storage arrays.
About this task
In a shared-switches configuration, four nodes join together to form two fabric-attached MetroCluster
configurations. In the illustration, the MetroCluster configurations are as follows:
•
•
FMC1-1 and FMC1-2 form one fabric-attached MetroCluster configuration named FMC1.
FMC2-1 and FMC2-2 form another fabric-attached MetroCluster configuration named FMC2.
210 | High Availability and MetroCluster Configuration Guide
Site A
Site B
FMC1-1
11
10
2 0
1
3
S1
7
FMC1-2
9
8
11
10
FCVI_TI_S1S3
17
18
0 2
17
18
5
1
3
S3
4
5
7
6
1A
2A
1A
2A
1A
2A
1A
2A
1B
2B
1B
2B
1B
2B
1B
2B
Storage array
1-1
8
7
5
S2
2 0
9 17
18
3 1
10
Storage array
2-2
Storage array
2-1
Storage array
1-2
5
1 3
FCVI_TI_S2S4
11
FMC2-1
Site A
4
6
17
18
11
0
7
S4
2
10
FMC2-2
Site B
Steps
1. Connect the ports of the FC-VI adapter to primary and secondary switches for both V-Series
systems in the MetroCluster configuration.
The following table shows the FC-VI adapter connections for the sample configuration:
V-Series system FC-VI adapter...
Connects to these switches...
FMC1-1
Primary switch S1 and secondary switch S2
through port 0
FMC1-2
Primary switch S3 and secondary switch S4
through port 0
FMC2-1
Primary switch S2 and secondary switch S1
through port 1
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 211
V-Series system FC-VI adapter...
Connects to these switches...
FMC2-2
Primary switch S3 and secondary switch S4
through port 1
2. Connect the V-Series FC initiator ports to the primary and secondary switches.
The FC initiator port connections for the sample configuration are shown in the following table:
V-Series FC initiator ports
Connect to these switches...
FMC 1-1
Primary switch S1 and secondary switch S2
through port 2 and port 10
FMC 1-2
Primary switch S3 and secondary switch S4
through port 2 and port 10
FMC 2-1
Primary switch S2 and secondary switch S1
through port 3 and port 11
FMC 2-2
Primary switch S3 and secondary switch S4
through port 3 and port 11
3. Connect the storage arrays to the switches.
The following table shows the switch-to-storage-array connections for FMC1:
Storage array and port
Switch and port
1-1: Port 1A
Switch 1: Port 7
1-1: Port 1B
Switch 1: Port 8
1-1: Port 2A
Switch 2: Port 7
1-1: Port 2B
Switch 2: Port 8
1-2: Port 1A
Switch 1: Port 5
1-2: Port 1B
Switch 1: Port 9
1-2: Port 2A
Switch 2: Port 5
1-2: Port 2B
Switch 2: Port 9
4. Repeat steps 1 through 3 at another site.
The following table shows the switch-to-storage-array connections for FMC2:
Storage array and port
Switch and port
2-1: Port 1A
Switch 3: Port 4
2-1: Port 1B
Switch 3: Port 5
212 | High Availability and MetroCluster Configuration Guide
Storage array and port
Switch and port
2-1: Port 2A
Switch 4: Port 6
2-1: Port 2B
Switch 4: Port 5
2-2: Port 1A
Switch 3: Port 6
2-2: Port 1B
Switch 3: Port 7
2-2: Port 2A
Switch 4: Port 4
2-2: Port 2B
Switch 4: Port 7
5. When cabling at both the sites is complete, connect an ISL (inter-switch link) cable to a port on
each switch.
In the illustration, the fabric containing switches S1 and S3 has two ISLs and the fabric
containing switches S2 and S4 also has two ISLs.
Switch S1 connects to switch S3 and switch S2 connects to switch S4 through ports 17 and 18.
After you finish
After you finish cabling the devices, you need to configure traffic isolation on the switches. This
configuration is required when you are using dual ISLs. See the Fabric-attached MetroCluster
Systems Brocade Switch Configuration Guide or the Fabric-attached MetroCluster Systems Cisco
Switch Configuration Guide for more information.
Related tasks
Setting preferred primary port in a MetroCluster configuration on page 182
Setting preferred primary port in a MetroCluster configuration
After cabling a MetroCluster configuration, you can define the primary port between the four
switches to carry the traffic. You can set the preferred primary port in a shared-switches
configuration by using the ic primary set command.
About this task
By default, the primary port is 0.
Steps
1. To set the primary port, enter the following command:
ic primary set 0|1 [-r]
You must use the -r option for the primary port to take effect immediately.
Note:
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 213
•
•
If the port that you have set as the preferred primary port is down, the other port will be
used. Once the port is up, it will be set as the preferred primary port.
If the FC-VI link between the preferred primary ports fails, the fabric-attached
MetroCluster configuration fails over to the secondary FC-VI link. When the primary link
is restored, the fabric-attached MetroCluster configuration again uses the primary link. The
failover temporarily causes the logs to be unsynchronized.
Example
If you want to set the preferred primary port to 1, enter the following command:
ic primary set 1 -r
In a shared-switches configuration, if you have set the preferred primary port of FMC1 (FMC1-1
and FMC1-2) to 1, the preferred primary port for FMC2 (FMC2-1 and FMC2-2) is set to 0.
2. To view the current primary port, enter the following command:
ic primary show
The primary port is displayed.
Removing the preferred primary port in a fabric-attached MetroCluster
configuration
You can remove the port that you assigned as primary port in a fabric-attached MetroCluster
configuration.
Step
1. To remove the primary port, enter the following command:
ic primary unset
Configuring zoning in a MetroCluster configuration with array LUNs
Switch zoning defines paths between connected nodes based on the node’s unique WWN. Singleinitiator zoning is recommended for a V-Series configuration.
Before you begin
•
•
•
•
Zoning must be determined for each switch.
The local and remote V-Series systems must be connected to the fabric.
The switch fabric must be connected and the switches must be powered up.
The fabric and storage array must be connected.
About this task
For a MetroCluster configuration with array LUNs, the FC-VI ports on the FC adapters must be
zoned end-to-end across the fabric.
214 | High Availability and MetroCluster Configuration Guide
Steps
1. Gather the WWPNs for the FC-VI ports on the V-Series system FC-VI adapter.
The WWPNs for the FC-VI ports are needed when you set up the zones for the FC-VI ports. The
WWPNs for the FC-VI ports are available from the following sources:
•
•
The switch
Data ONTAP sysconfig -M output
Gather WWPNs from entries such as the following: !Qlogic 2352 FCVI Cluster
Interconnect Adapter<adapter_WWPN>
2. Zone FC-VI port “a” on the local V-Series system to the FC-VI port “a” on the remote V-Series
system.
3. Zone FC-VI port “b” on the local V-Series system to the FC-VI port “b” on the remote V-Series
system.
After you finish
•
•
If necessary, change the configuration speed of your stretch MetroCluster configuration.
Set up Data ONTAP on the V-Series systems so that they can access storage on the storage
arrays, and so that you can take advantage of Data ONTAP features.
Related concepts
Planning zoning for a MetroCluster configuration with array LUNs on page 196
Related tasks
Testing zoning of FC-VI ports in a MetroCluster configuration with array LUNs on page 215
Changing the configuration speed of a stretch MetroCluster configuration
If the distance between nodes in a stretch MetroCluster configuration is greater than the supported
default configuration speed, you must change the default configuration speed. If you modified the
default configuration speed in a stretch MetroCluster configuration that uses an FC-VI adapter, you
can reset the speed to the default configuration speed.
The default maximum speed between nodes in a stretch MetroCluster configuration depends on the
speed at which the FC-VI adapter operates (for example, 4 GB or 8 GB). The maximum supported
distance is included in the Interoperability Matrix at support.netapp.com.
The procedures for changing the configuration speed of a stretch MetroCluster configuration are the
same for MetroCluster configurations with FAS systems and MetroCluster configurations with VSeries systems.
Related tasks
Changing the default configuration speed of a stretch MetroCluster configuration on page 149
Resetting a stretch MetroCluster configuration to the default speed on page 151
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 215
Setting up Data ONTAP after connecting devices in a
MetroCluster configuration with array LUNs
After connecting the devices in the MetroCluster configuration, you need to set up the V-Series
systems to use the storage on the storage array. You also need to set up any desired Data ONTAP
features.
Perform the following tasks after connecting the devices in your MetroCluster configuration with VSeries systems.
1. Set up the V-Series systems as described in the Data ONTAP Software Setup Guide for 7-Mode
for V-Series systems using only array LUNs.
2. Create one or more mirrored aggregates.
See the Data ONTAP Data Protection Online Backup and Recovery Guide for 7-Mode.
3. Test the MetroCluster configuration, as described in the Data ONTAP High Availability and
MetroCluster Configuration Guide for 7-Mode.
4. Verify network and protocol setup, as described in the Data ONTAP File Access and Protocols
Management Guide for 7-Mode.
5. Configure other Data ONTAP features as desired, for example, data protection features.
See the relevant Data ONTAP guide for the feature you want to set up.
Testing a MetroCluster configuration with array LUNs
It is important to test that your MetroCluster configuration is properly set up before putting it into a
production environment.
Steps
1. Testing zoning of FC-VI ports in a MetroCluster configuration with array LUNs on page 215
2. Verifying proper setup at MetroCluster sites with storage arrays on page 216
3. Simulating a disaster recovery in a MetroCluster configuration with array LUNs on page 217
Testing zoning of FC-VI ports in a MetroCluster configuration with array
LUNs
To test zoning of the FC-VI ports, you need to check that you have not crossed the FC-VI “a” ports
and “b” ports in your zones.
Related tasks
Configuring zoning in a MetroCluster configuration with array LUNs on page 213
216 | High Availability and MetroCluster Configuration Guide
Verifying proper setup at MetroCluster sites with storage arrays
After installing a MetroCluster configuration with storage arrays, you need to test the paths, FC
ports, and switch configuration at each MetroCluster site to ensure that they are set up correctly.
Steps
1. Enter the following command to display path information from each V-Series system to the array
LUNs:
storage show disk -p
You should see two paths to each array LUN.
2. To test FC initiator ports, enter the following commands for each FC initiator port in the
MetroCluster configuration:
a) Enter privilege mode:
priv set advanced
b) Display the state of the HBA port:
fcadmin link_state
c) Take an FC port offline to simulate a port failure or a cable pull:
fcadmin offline portname
Example
fcadmin offline 0a
d) Display disk path information:
storage show disk -p
The display should show only one path.
e) Bring the FC port online:
fcadmin online portname
Example
fcadmin online 0a
f) Verify that both paths are online:
storage show disk -p
3. Simulate a switch failure or a storage array controller failure for each switch fabric by entering
the following commands:
a) Take all FC ports offline on both V-Series systems that are attached to one fabric:
fcadmin offline portname
Implementing a stretch or fabric-attached MetroCluster configuration with array LUNs | 217
Example
fcadmin offline 0a
fcadmin offline 0b
fcadmin offline 0c
fcadmin offline 0d
If your V-Series model has more than four FC initiator ports, take the remaining initiator ports
offline.
b) Verify that all HBAs are disabled:
fcadmin link_state
c) Initiate a takeover from site A:
cf takeover
d) When site B is in takeover mode, initiate a giveback from Site A:
cf giveback
Simulating a disaster recovery in a MetroCluster configuration with array
LUNs
Testing for proper setup of a MetroCluster configuration with array LUNs includes simulating a
disaster recovery. Simulating a site failure and recovery involves disabling and degrading mirrors.
About this task
You should never simulate disaster site failure and recovery in production environments without
prior planning and downtime.
Steps
1. Disable the HA interconnect between the V-Series systems.
2. Power down one of the V-Series systems.
The system you power down simulates the site of the disaster.
3. On the surviving system, complete the following steps:
a) Enter the following command to activate a forced, manual takeover:
cf forcetakeover -d
b) Enter the following command to validate that the forced takeover has occurred:
cf status
c) Enter the following commands to validate the status of the aggregates and of the volumes that
they include:
aggr status
vol status
218 | High Availability and MetroCluster Configuration Guide
The display should show an online status for aggregates and volumes, but mirrors, which are
disabled, should be displayed as “degraded.”
4. Reconnect the HA interconnect.
5. On the surviving system, enter the following commands:
a) Enter the following command to rejoin the aggregates:
aggr mirror surviving_aggregate -v victim_aggregate
Note: Data ONTAP uses parentheses to indicate the degraded aggregate. For example,
aggr0_b indicates the aggregate that is not degraded, and aggr0_b(1) is degraded.
b) Enter the following command to validate that forced takeover has occurred:
cf status
c) Enter the following command to validate aggregate status:
aggr status
The display should show an online and mirrored status for aggregates.
d) Enter the following command to validate volume status:
vol status
The display should show an online and mirrored status for volumes.
6. Power up the V-Series system that you powered down to simulate the disaster.
7. On the surviving system, enter the following command to reactivate the site that simulated the
disaster:
cf giveback
8. To validate that the HA pair, aggregates, and mirrors are online and operational, enter the
following commands at both sites:
cf status
aggr status
vol status
The display should indicate that the HA pair is enabled, and that aggregates and volumes are
online and mirrored.
219
Implementing a MetroCluster configuration with
both disks and array LUNs
Starting in Data ONTAP 8.2, a MetroCluster configuration with V-Series systems can include both
native disks and LUNs from storage arrays (or just one of these types of storage). Only MetroCluster
configurations with V-Series systems support both types of storage.
Prior to Data ONTAP 8.2, support for MetroCluster configurations with V-Series systems was
limited to systems that used only array LUNs.
Planning a MetroCluster configuration with disks and array
LUNs
When planning a MetroCluster configuration with V-Series systems using both disks and array
LUNs, you need to consider the requirements for using each type of storage in the configuration and
the additional guidelines for using both types of storage in a MetroCluster configuration.
Guidelines for implementing a MetroCluster configuration with disks and
array LUNs
When planning your MetroCluster configuration to use disks and array LUNs, you need to follow the
guidelines for setting up a MetroCluster configuration for each type of storage plus the additional
guidelines for setting up a MetroCluster configuration with both types of storage.
The basic requirements for using disks and array LUNs in a MetroCluster configuration with VSeries systems are the same whether you are using just one type of storage or both. Likewise,
installation, configuration, and testing of each type of storage is the same as if you were using just
one type of storage.
However, there are additional guidelines to consider if you are using both disks and array LUNs in
your MetroCluster configuration with V-Series systems.
Note: A MetroCluster configuration with both disks and array LUNs is not supported prior to Data
ONTAP 8.2, and it is supported only with V-Series systems in the configuration.
Consider the information in the following table when planning your configuration:
Consideration
Guideline
Order of setting up access to the
storage
You can set up access to either type of storage first. You
should complete all setup for that type of storage and
verify that it is set up correctly before setting up the other
type of storage.
220 | High Availability and MetroCluster Configuration Guide
Consideration
Guideline
Root volume location
•
•
If you are setting up a new MetroCluster deployment
with both disks and array LUNs, it is recommended
that you create the root volume on native disks.
If you are adding native disks to an existing
MetroCluster configuration with V-Series systems, the
root volume can remain on an array LUN.
FC initiator port usage
You need to use different FC initiator ports to connect to
a FibreBridge 6500N bridge than you use to connect to
the switches that connect to the storage arrays.
V-Series platform in the MetroCluster
configuration
In a V-Series MetroCluster configuration, four FC
connections are required for native disks shelves
connected through the FibreBridge 6500N and four FC
connections are required for the FC connections to target
ports on the storage arrays. Therefore, a minimum of
eight FC initiator ports are required for a V-Series system
to connect to both native disks and array LUNs. Not all
V-Series platforms support eight FC initiator ports. See
the Interoperability Matrix at support.netapp.com to
determine whether your platform can support eight FC
initiator ports.
Switches and switch port usage
For stretch MetroCluster configurations, only direct
connection to a FibreBridge 6500N bridge is supported.
Switches are used to connect to the storage arrays.
For fabric-attached MetroCluster configurations, the rules
for switches are as follows:
•
•
Reverting
(Transitioning a storage system to a
Data ONTAP release in an earlier
family)
The same switches are used to connect to the storage
arrays and the FibreBridge 6500N bridges.
The switch ports used to connect to a FibreBridge
6500N bridge must be different from the switch ports
used to connect to a storage array.
You cannot revert a system set up with a mixed disk and
array LUN MetroCluster configuration because a mixed
disk and array LUN MetroCluster configuration is not
supported prior to Data ONTAP 8.2.
Related concepts
Planning a fabric-attached MetroCluster configuration using disks on page 153
Implementing a MetroCluster configuration with both disks and array LUNs | 221
Setup requirements and restrictions for fabric-attached MetroCluster configurations with disks on
page 153
Planning a stretch MetroCluster configuration using disks on page 138
Hardware requirements for using disks in a MetroCluster configuration with V-Series systems on
page 139
Setup requirements and restrictions for stretch MetroCluster configurations with disks on page 138
Planning for a MetroCluster configuration with array LUNs on page 184
Requirements for a MetroCluster configuration with array LUNs on page 186
Recommended fabric-attached MetroCluster configuration with array LUNs on page 188
Recommended stretch MetroCluster configuration with array LUNs on page 192
Supported fabric-attached MetroCluster configuration with disks and array
LUNs
When configuring a MetroCluster configuration with V-Series systems and disks and storage arrays,
you need to follow the best practice recommendations for how to connect the hardware components.
The key points to remember about connections in a fabric-attached MetroCluster configuration with
disks and storage arrays are as follows:
•
•
•
You use the same switches to connect a V-Series system to the FibreBridge 6500N bridges and
the storage array.
You use different switch ports for the connections to the FibreBridge 6500N bridges than for the
connections to the storage array.
You use different V-Series FC initiator ports for the switch ports connecting to the FibreBridge
6500N bridges than for the switch ports connecting to the storage array.
If your V-Series system does not have enough FC initiator ports for separate connections to the
bridges and storage array, you must add some FC initiator ports.
The Interoperability Matrix at support.netapp.com contains information about the hardware
components that are supported for a MetroCluster configuration with storage arrays. Configuring a
MetroCluster system with SAS disk shelves and FibreBridge 6500N bridges contains information
about the connections between the FibreBridge 6500N bridges and the disk shelves.
The following block diagram shows a fabric-attached MetroCluster configuration that uses only array
LUNs.
222 | High Availability and MetroCluster Configuration Guide
vs1
FC-VI
6
0a
1
0b
0d
FC-VI
FC-VI
2
Switch 1
(Fabric 1)
3 4
0c
vs2
1
6
5
1
2
6
Switch 3
(Fabric 2)
5
5
0a
0c
0d
5
6
2
1
Switch 4
(Fabric 2)
3 4
Storage array 1
FC-VI
2
Switch 2
(Fabric 1)
3 4
0b
3
4
Storage array 2
The following block diagram shows the same fabric-attached MetroCluster, but with disks as well.
Notice that the same switches are used for access to the storage arrays and the FibreBridge 6500N
bridges, but different switch ports are used for the bridges and the storage arrays. Also notice that
different FC initiator ports are used for the two types of storage.
Implementing a MetroCluster configuration with both disks and array LUNs | 223
vs1
vs2
FC-VI
FC-VI
0a
0b
0c
1
6
0d
2
Switch 1
(Fabric 1)
9
3 4
1a
1b
1c
1d
7 8
5
1
0a
6
2
7
8 6
Switch 3
(Fabric 2)
9
5
0b
0c
1 2
Switch 2
(Fabric 1)
9
3 4
0d
1a
1c
1d
FC-VI
7 8
2 7 8 6
1
5
5
1b
3 4
Switch 4
(Fabric 2)
9
FibreBridge
6500N
3
4
FibreBridge
6500N
Disk shelf
...
...
Disk shelf
Disk shelf
Storage array 1
Disk shelf
Storage array 2
Supported stretch MetroCluster configuration with disks and array LUNs
When configuring a MetroCluster configuration with V-Series systems and disks and array LUNs,
you need to follow the best practice recommendations for how to connect the hardware components.
The key points to remember about connections in a stretch MetroCluster configuration with disks and
storage arrays are as follows:
•
•
•
The V-Series system connects directly to the FibreBridge 6500N bridges, but connects to the
storage array through switches.
Connecting the V-Series system to the FibreBridge 6500N bridges through switches is not
supported.
Different V-Series FC initiator ports must be used for the switch ports connecting to the
FibreBridge 6500N bridges than the FC initiator ports for connecting to the storage array.
If your V-Series system does not have enough FC initiator ports for separate connections to the
bridges and storage array, you must add some FC initiator ports.
The Interoperability Matrix at support.netapp.com contains information about the hardware
components that are supported for a MetroCluster configuration with storage arrays. Configuring a
224 | High Availability and MetroCluster Configuration Guide
MetroCluster system with SAS disk shelves and FibreBridge 6500N bridges contains information
about the connections between the FibreBridge 6500N bridges and the disk shelves.
The following block diagram shows a stretch MetroCluster configuration with just array LUNs.
vs1
vs2
HA interconnect
0a
1
0b
0c
0d
0a
2
1
Switch 1
(Fabric 1)
3 4
5
1
5
2
Switch 3
(Fabric 2)
0c
0d
2
Switch 2
(Fabric 1)
3 4
5
0b
2
1
5
Switch 4
(Fabric 2)
3 4
Storage array 1
3
4
Storage array 2
The following block diagram shows the connectivity when FibreBridge 6500N bridges and disk
shelves are added to the stretch MetroCluster configuration from the previous illustration. The
connections for accessing array LUNs are purposely grayed out to make it easier for you to see the
connections for the FibreBridge bridges.
Implementing a MetroCluster configuration with both disks and array LUNs | 225
vs1
vs2
HA interconnect
0a
0b
1
0c
0d
1a
1b
1c
1d
2
Switch 1
(Fabric 1)
3 4
5
1
0a
0b
1
2
5
2
Switch 3
(Fabric 2)
0c
Switch 2
(Fabric 1)
3 4
0d
1a
1c
1d
2
1
5
5
1b
Switch 4
(Fabric 2)
3 4
3
FibreBridge
6500N
4
FibreBridge
6500N
Disk shelf
...
...
Disk shelf
Disk shelf
Storage array 1
Disk shelf
Storage array 2
As the illustration shows, the stretch MetroCluster uses direct FC connections between the V-Series
systems and the FibreBridge 6500N bridges. Eight FC initiator ports are required on each V-Series
system, four for the FibreBridge 6500N bridges and four for the connections to the storage array via
the switches.
Details about how to connect the V-Series systems to the storage arrays are not discussed here. See
the information in the Data ONTAP High Availability and MetroCluster Configuration Guide for 7Mode for how to connect the V-Series systems to the storage arrays.
Testing a MetroCluster configuration with disks and array
LUNs
You should test your MetroCluster configuration in stages, testing that each type of storage is
working properly before going onto installation for the next type of storage.
226 | High Availability and MetroCluster Configuration Guide
Related concepts
Testing a MetroCluster configuration with array LUNs on page 215
227
Performing nondisruptive shelf replacement in a
MetroCluster configuration
In a MetroCluster configuration, you can replace the DS14mk2 FC and DS14mk4 FC disk shelves
nondisruptively. Performing nondisruptive shelf replacement (NDSR) involves preparing for the
procedure, replacing disk shelves, and verifying the disk shelves after the shelf replacement.
About this task
For performing the nondisruptive shelf replacement procedure, the two MetroCluster nodes referred
in the steps are mc-nodeA and mc-nodeB and the affected node is mc-nodeB. The disk shelf that
requires replacement is part of the loop connected to the port 9 of the FC switches on mc-nodeB.
Steps
1. Preparing for nondisruptive shelf replacement on page 227
2. Replacing the disk shelf nondisruptively on page 228
3. Verifying the disks after the shelf replacement on page 230
Preparing for nondisruptive shelf replacement
You need to prepare the storage system before performing a nondisruptive shelf replacement
procedure.
Before you begin
All disks on loops affected by a disk shelf must be mirrored.
Steps
1. Verify that all aggregates and volumes contained in the disks on the affected loop are mirrored
and the mirroring is operational by using the aggr status and sysconfig –r commands at
both the nodes.
Example
mc-nodeB> aggr status
mc-nodeB> sysconfig –r
228 | High Availability and MetroCluster Configuration Guide
mc-nodeA> aggr status
mc-nodeA> sysconfig -r
2. Trigger an AutoSupport to indicate the start of the disk shelf replacement process by using the
options autosupport.doit command at both the nodes.
Example
mc-nodeB> options autosupport.doit “SHELF REPLACE: START”
mc-nodeA> options autosupport.doit “SHELF REPLACE: START”
3. Get the total number of the disks on both the nodes by using the sysconfig command.
You should save this output for comparing with the output received after the disk shelf
replacement.
Replacing the disk shelf nondisruptively
After preparing your storage system to perform nondisruptive shelf replacement, you can replace the
disk shelf.
Steps
1. Take the aggregate and plex on the mc-nodeB offline by using the aggr offline command.
Example
mc-nodeB> aggr offline aggr1/plex0
2. Disable switch port 9 on both the switches in the mc-nodeB by logging in as admin and using the
portdisable command.
Example
mc-nodeB_sw0> portdisable 9
mc-nodeB_sw2> portdisable 9
3. Wait until all disks missing notifications on both the nodes are complete and verify that the disks
and shelves are no longer visible by using the sysconfig and sysconfig -a commands.
Performing nondisruptive shelf replacement in a MetroCluster configuration | 229
Example
mc-nodeB> sysconfig
mc-nodeB> sysconfig -a
mc-nodeA> sysconfig
mc-nodeA> sysconfig –a
4. Power off the disk shelf connected to port 9 of the switches in the mc-nodeB.
5. Remove disks from the shelf.
You must ensure that you place the disks at a safe place.
6. Disconnect all FC and SFP cables from the disk shelf that are part of the loop connected to port 9.
7. Remove the disk shelf that are part of the loop connected to port 9 from the rack cabinet.
8. Remove module A (top slot), including SFP, from the disk shelf and insert into slot A (top slot)
on replacement shelf.
If required, replace the module and the SFP.
9. Remove module B (bottom slot), including SFP, from the disk shelf and insert into slot B (bottom
slot) on replacement shelf.
If required, replace the module and the SFP.
10. Insert the replacement shelf into rack cabinet.
You must set the module speed and ensure that shelf ID is the same as the one that was replaced.
11. Cable all the connections again.
If required, replace the cables.
12. Reconnect all SFP cables between the new disk shelf and the disk shelf of the same loop
connected to port 9.
Note: You must not perform this step if the switch port 9 has a single disk shelf loop.
13. Insert the disks removed in step 5 into replacement shelf only after shelf is replaced (completely
installed with requisite install kit) into rack cabinet.
14. Power on disk shelf on the loop connected to port 9.
Verify that power is up and no alarms are reported.
230 | High Availability and MetroCluster Configuration Guide
Verifying the disks after the shelf replacement
After replacing the disk shelf, you must perform certain steps to ensure that the disks are operational.
Steps
1. Enable the switch port on both the switches at mc-nodeB by using the portenable command.
Example
mc-nodeB_sw0> portenable 9
mc-nodeB_sw2> portenable 9
2. Verify that all the disks and aggregates appear correctly in each offline plex by using the aggr
status, sysconfig -a, and sysconfig -r commands at mc-nodeB.
Example
mc-nodeB> aggr status -r aggr1
mc-nodeB> sysconfig –a
mc-nodeB> sysconfig –r
3. Verify that the disks are active and online by using the sysconfig command at both the nodes.
You should compare this output with the output generated before replacing the disk shelf to
confirm that the same number of disks appears under each FC host adapter listed as active and
online.
4. Take the aggregate and plex online on the mc-nodeB by using the aggr online command.
Example
mc-nodeB> aggr online aggr1/plex0
Note: You should wait for all aggr1 volumes on mc-nodeB to finish resyncing and return to a
mirrored state. This step might take minutes or hours.
5. Trigger an AutoSupport from both the nodes to indicate the completion of the disk shelf
replacement process.
Performing nondisruptive shelf replacement in a MetroCluster configuration | 231
Example
mc-nodeB> options autosupport.doit “SHELF REPLACE:FINISH”.
mc-nodeA> options autosupport.doit “SHELF REPLACE:FINISH”
232 | High Availability and MetroCluster Configuration Guide
Recovering from a disaster by using MetroCluster
configurations
In situations such as prolonged power outages or natural disasters, you can use the MetroCluster
feature of Data ONTAP to provide a quick failover to another site that contains a nearly real time
copy of the data at the disaster site.
Conditions that constitute a disaster
The disaster recovery procedure is an extreme measure that you should use only if the failure disrupts
all communication from one MetroCluster site to the other for a prolonged period of time.
The following are examples of disasters that could cause such a failure:
•
•
•
•
Fire
Earthquake
Prolonged power outages at a site
Prolonged loss of connectivity from clients to the storage systems at a site
Ways to determine whether a disaster occurred
You should declare a disaster only after using predefined procedures to verify that service cannot be
restored.
It is critical that you follow a predefined procedure to confirm that a disaster occurred. The procedure
should include determining the status of the disaster site by:
•
Using external interfaces to the storage system, such as the following:
•
•
• ping command to verify network connectivity
• Remote shell
• FilerView administration tool
Using network management tools to verify connectivity to the disaster site
Physically inspecting the disaster site, if possible
You should declare a disaster only after verifying that service cannot be restored.
Failures that do not require disaster recovery
There are some failures that do not require any disaster recovery such as, a failure of the
interconnect, failure of cables and so on. If you can reestablish the MetroCluster connection after
fixing the problem, you should not perform the disaster recovery procedure.
You should not perform the disaster recovery procedure for the following failures:
•
A failure of the HA interconnect between the two sites, which can be caused by the following:
Recovering from a disaster by using MetroCluster configurations | 233
•
•
•
•
With this type of failure, both nodes remain running. Automatic takeover is disabled because
Data ONTAP cannot synchronize the NVRAM logs. After you fix the problem and reestablish
the connection, the nodes resynchronize their NVRAM logs and the MetroCluster configuration
returns to normal operation.
The storage from one site (site A) is not accessible to the node at the other site (site B), which can
be caused by the following:
•
•
•
•
•
•
Failure of the interconnect cable
Failure of one of the FC-VI adapters
If using switches, a failure of the SFP connecting a node to the switch
Failure of any of the cables connecting the storage at one site to the node at the other site or
switch
If using switches, failure of any of the SFPs connecting the storage to the switch or the node
to the switch
Failure of the Fibre Channel adapter on the node
Failure of a storage disk shelf (disk shelf module; power; access to disk shelves; and so on)
With this type of failure, you see a mailbox disk invalid message on the console of the
storage system that cannot see the storage. After you fix the problem and reestablish the
connection, the MetroCluster configuration returns to normal operation.
If you are using switches, the inter-switch link between each pair of switches fails.
With this type of failure, both nodes remain running. You see a mailbox disk invalid
message because a storage system at one site cannot see the storage system at the other site. You
also see a message because the two nodes cannot communicate with each other. After you fix the
problem and reestablish the connection, the nodes resynchronize their NVRAM logs and the
MetroCluster configuration returns to normal operation.
If you are using FibreBridge 6500N bridge, a failure of the FibreBridge 6500N bridge.
Recovering from a disaster
After determining that there is a disaster, you should take steps to recover access to data, fix
problems at the disaster site, and re-create the MetroCluster configuration.
About this task
Complete the following tasks in the order shown.
Attention: If for any reason the primary node has data that was not mirrored to the secondary node
prior to the execution of the cf forcetakeover –d command, data could be lost. Do not
resynchronize the original disks of the primary site for a SnapLock volume until an additional
backup has been made of those disks to ensure availability of all data. This situation could arise,
for example, if the link between the sites was down and the primary node had data written to it in
the interim before the cf forcetakeover –d command was issued.
234 | High Availability and MetroCluster Configuration Guide
For more information about backing up data in SnapLock volumes using SnapMirror, see the Data
ONTAP Archive and Compliance Management Guide for 7-Mode.
Steps
1.
2.
3.
4.
5.
6.
Restricting access to the disaster site node on page 234
Forcing a node into takeover mode on page 235
Remounting volumes of the failed node on page 235
Recovering LUNs of the failed node on page 236
Fixing failures caused by the disaster on page 237
Reestablishing the MetroCluster configuration on page 238
Restricting access to the disaster site node
You must restrict access to the disaster site node to prevent the node from resuming service. If you
do not restrict access, you risk the possibility of data corruption.
About this task
You can restrict access to the disaster site node in the following ways:
•
•
Turning off power to the disaster site node.
Manually fencing off the node.
Steps
1. Restricting access to the node by turning off power on page 234
2. Restricting access to the node by fencing off on page 234
Restricting access to the node by turning off power
This is the preferred method for restricting access to the disaster site node. You can perform this task
at the disaster site or remotely, if you have that capability.
Step
1. Switch off the power at the back of the storage system.
Restricting access to the node by fencing off
You can use manual fencing as an alternative to turning off power to the disaster site node. The
manual fencing method restricts access using software and physical means.
Steps
1. Disconnect the HA interconnect and Fibre Channel adapter cables of the node at the surviving
site.
Recovering from a disaster by using MetroCluster configurations | 235
2. Use the appropriate fencing method depending on the type of failover you are using:
If you are using...
Then fencing is achieved by...
Application
failover
Using any application-specified method that either prevents the application from
restarting at the disaster site or prevents the application clients from accessing the
application servers at the disaster site. Methods can include turning off the
application server, removing an application server from the network, or any other
method that prevents the application server from running applications.
IP failover
Using network management procedures to ensure that the storage systems at the
disaster site are isolated from the external public network.
Forcing a node into takeover mode
If a disaster has occurred, you can force the surviving node into takeover mode, so that the surviving
node serves the data of the failed node.
Step
1. Enter the following command on the surviving node:
cf forcetakeover -d
Result
Data ONTAP causes the following to occur:
•
•
The surviving node takes over the functions of the failed node.
The mirrored relationships between the two plexes of mirrored aggregates are broken, thereby
creating two unmirrored aggregates.
This is called splitting the mirrored aggregates.
The overall result of using the cf forcetakeover -d command is that a node at the surviving site
is running in takeover mode with all the data in unmirrored aggregates.
Remounting volumes of the failed node
If the cf.takeover.change_fsid option is set to on, you must remount the volumes of the failed
node because the volumes are accessed through the surviving node.
About this task
For more information about mounting volumes, see the Clustered Data ONTAP File Access and
Protocols Management Guide.
Note: You can disable the change_fsid option to avoid the necessity of remounting the volumes.
236 | High Availability and MetroCluster Configuration Guide
Steps
1. On an NFS client at the surviving site, create a directory to act as a mount point by entering the
following command.
mkdir directory_path
Example
mkdir /n/toaster/home
2. Mount the volume by entering the following command.
mount volume_name
Example
mount toaster:/vol/vol0/home /n/toaster/home
Related tasks
Disabling the change_fsid option in MetroCluster configurations on page 79
Recovering LUNs of the failed node
You must actively track whether LUNs are online or offline in a MetroCluster configuration. If the
cf.takeover.change_fsid option is set to on, and there is a disaster, all LUNs in the aggregates
that were mirrored at the surviving site are offline. You can’t determine if they were online prior to
the disaster unless you track their state.
About this task
If you have a MetroCluster configuration, you must actively track the state of LUNs (track whether
they are online or offline) on the node at each site. If there is a failure to a MetroCluster
configuration that qualifies as a disaster and the node at one site is inaccessible, all LUNs in the
aggregates that were mirrored at the surviving site are offline. There is no way to distinguish the
LUNs that were offline before the disaster from the LUNs that were online before the disaster unless
you have been tracking their status.
When you recover access to the failed node’s LUNs, it is important to bring back online only the
LUNs that were online before the disaster. To avoid igroup mapping conflicts, do not bring a LUN
online if it was offline before the disaster. For example, suppose you have two LUNs with IDs of 5
mapped to the same igroup, but one of these LUNs was offline before the disaster. If you bring the
previously offline LUN online first, you cannot bring the second LUN online because you cannot
have two LUNs with the same ID mapped to the same host.
Note: You can disable the change_fsid option to avoid the necessity of remounting the volumes.
Steps
1. Identify the LUNs that were online before the disaster occurred.
Recovering from a disaster by using MetroCluster configurations | 237
2. Make sure that the LUNs are mapped to an igroup that contains the hosts attached to the
surviving node.
For more information about mapping LUNs to igroups, see your Data ONTAP SAN
Administration Guide for 7-Mode.
3. On the surviving node, enter the following command:
lun online lun-path ...
lun-path is the path to the LUN you want to bring online. You can specify more than one path
to bring multiple LUNs online.
Example
lun online /vol/vol1/lun5
Example
lun online /vol/vol1/lun3 /vol/vol1/lun4
Note: After you bring LUNs back online, you might have to perform some application or hostside recovery procedures. For example, the File System Identifiers (FSIDs) are rewritten,
which can cause the LUN disk signatures to change. For more information, see the
documentation for your application and for your host operating system.
Fixing failures caused by the disaster
You need to fix the failures caused by the disaster, if possible. For example, if a prolonged power
outage to one of the MetroCluster sites caused the failure, restoring the power fixes the failure.
About this task
You cannot fix failures if the disaster causes a site to be destroyed. For example, a fire or an
earthquake could destroy one of the MetroCluster sites. In this case, you fix the failure by creating a
new partner for a MetroCluster configuration at a different site.
Step
1. Fix the failures at the disaster site.
After you finish
After the node at the surviving site can see the disk shelves at the disaster site, Data ONTAP renames
the mirrored aggregates that were split, and the volumes they contain, by adding a number in
parenthesis to the name. For example, if a volume name was vol1 before the disaster and the split, the
renamed volume name could be vol1(1).
238 | High Availability and MetroCluster Configuration Guide
Reestablishing the MetroCluster configuration
You can reestablish a MetroCluster configuration after a disaster, depending on the state of the
mirrored aggregate at the time of the takeover.
About this task
Depending on the state of a mirrored aggregate before you forced the surviving node to take over its
partner, you use one of two procedures to reestablish the MetroCluster configuration:
•
•
If the mirrored aggregate was in a normal state before the forced takeover, you can rejoin the two
aggregates to reestablish the MetroCluster configuration.
This is the most typical case.
If the mirrored aggregate was in an initial resynchronization state (level-0) before the forced
takeover, you cannot rejoin the two aggregates.
You must re-create the synchronous mirror to reestablish the MetroCluster configuration.
Rejoining the mirrored aggregates to reestablish a MetroCluster configuration
You must rejoin the mirrored aggregates if the mirrored aggregate was in a normal state before the
forced takeover.
About this task
Attention: If you attempt a giveback operation prior to rejoining the aggregates, you might cause
the node to boot with a previously failed plex, resulting in a data service outage.
Steps
1. Validate that you can access the remote storage by entering the following command:
aggr status -r
2. Turn on power to the node at the disaster site.
After the node at the disaster site boots, it displays the following message:
Waiting for Giveback...
3. Determine which aggregates are at the surviving site and which aggregates are at the disaster site
by entering the following command:
aggr status
Aggregates at the disaster site show plexes that are in a failed state with an out-of-date status.
Aggregates at the surviving site show plexes as online.
4. If aggregates at the disaster site are online, take them offline by entering the following command
for each online aggregate:
aggr offline disaster_aggr
disaster_aggr is the name of the aggregate at the disaster site.
Recovering from a disaster by using MetroCluster configurations | 239
Note: An error message appears if the aggregate is already offline.
5. Re-create the mirrored aggregates by entering the following command for each aggregate that
was split:
aggr mirror aggr_name -v disaster_aggr
aggr_name is the aggregate on the surviving site’s node.
disaster_aggr is the aggregate on the disaster site’s node.
The aggr_name aggregate rejoins the disaster_aggr aggregate to reestablish the MetroCluster
configuration.
6. Verify that the mirrored aggregates have been re-created by entering the following command:
aggr status -r
The giveback operation only succeeds if the aggregates have been rejoined.
7. Enter the following command at the partner node:
cf giveback
The node at the disaster site reboots.
Example of rejoining aggregates
The following example shows the commands and status output when you rejoin aggregates to
reestablish the MetroCluster configuration.
First, the aggregate status of the disaster site’s storage after reestablishing access to the partner
node at the surviving site is shown:
filer1> aggr status -r
Aggregate mir (online, normal) (zoned checksums)
Plex /mir/plex5 (online, normal, active)
RAID group /filer1/plex5/rg0 (normal)
RAID Disk Device HA SHELF BAY
--------- ------ ------------parity
8a.2
8a
0
2
data
8a.8
8a
1
0
CHAN
----FC:B
FC:B
Used (MB/blks)
-------------34500/70656000
34500/70656000
Phys (MB/blks)
------------35003/71687368
35003/71687368
Aggregate mir(1) (failed, out-of-date) (zoned checksums)
Plex /mir(1)/plex1 (offline, normal, out-of-date)
RAID group /mir(1)/plex1/rg0 (normal)
RAID Disk Device HA SHELF BAY
--------- ------ ------------parity
6a.0
6a
0
0
data
6a.1
6a
0
1
CHAN
----FC:B
FC:B
Used (MB/blks)
-------------34500/70656000
34500/70656000
Phys (MB/blks)
------------35003/71687368
35003/71687368
Plex /mir(1)/plex5 (offline, failed, out-of-date)
Next, the mirror is reestablished using the aggr mirror -v command.
240 | High Availability and MetroCluster Configuration Guide
Note: The node at the surviving site is called filer1; the node at the disaster site is called
filer2.
filer1> aggr mirror mir -v mir(1)
This will destroy the contents of mir(1). Are you sure? y
Mon Nov 18 15:36:59 GMT [filer1:
raid.mirror.resync.snapcrtok:info]: mir: created mirror
resynchronization snapshot mirror_resync.1118153658(filer2)
Mon Nov 18 15:36:59 GMT [filer1: raid.rg.resync.start:notice]: /mir/
plex6/rg0: start resynchronization (level 1)
Mon Nov 18 15:36:59 GMT [filer1: raid.mirror.resync.start:notice]: /
mir: start resynchronize to target /mir/plex6
After the aggregates rejoin, the synchronous mirrors of the MetroCluster configuration are
reestablished:
filer1> aggr status -r mir
Aggregate mir (online, mirrored) (zoned checksums)
Plex /mir/plex5 (online, normal, active)
RAID group /mir/plex5/rg0 (normal)
RAID Disk Device HA SHELF BAY
--------- ------ ------------parity
8a.2
8a
0
2
data
8a.8
8a
1
0
CHAN
----FC:B
FC:B
Used (MB/blks)
-------------34500/70656000
34500/70656000
Phys (MB/blks)
------------35003/71687368
35003/71687368
Plex /mir/plex6 (online, normal, active)
RAID group /mir/plex6/rg0 (normal)
RAID Disk Device HA SHELF BAY
--------- ------ ------------parity
6a.0
6a
0
0
data
6a.1
6a
0
1
CHAN
----FC:B
FC:B
Used (MB/blks)
-------------34500/70656000
34500/70656000
Phys (MB/blks)
------------35003/71687368
35003/71687368
Re-creating mirrored aggregates to return a MetroCluster configuration to normal
operation
To return the MetroCluster configuration to normal operation, you must re-create the mirrored
aggregates.
Steps
1. Validate that you can access the remote storage by entering the following command:
aggr status -r
Note: A (level-0 resync in progress) message indicates that a plex cannot be rejoined.
2. Turn on the power to the node at the disaster site.
After the node at the disaster site boots up, it displays the following:
Recovering from a disaster by using MetroCluster configurations | 241
Waiting for giveback...(Press Ctrl-C to abort wait)
3. If the aggregates at the disaster site are online, take them offline by entering the following
command for each aggregate that was split:
aggr offline disaster_aggr
disaster_aggr is the name of the aggregate at the disaster site.
Note: An error message appears if the aggregate is already offline.
4. Destroy every target plex that is in a level-0 resync state by entering the following command:
aggr destroy plex_name
5. Re-create the mirrored aggregates by entering the following command for each aggregate that
was split:
aggr mirror aggr_name -v disaster_aggr
aggr_name is the aggregate on the surviving site’s node.
disaster_aggr is the aggregate on the disaster site’s node.
The aggr_name aggregate rejoins the disaster_aggr aggregate to reestablish the MetroCluster
configuration.
6. Enter the following command at the partner node:
cf giveback
The node at the disaster site reboots.
Example of re-creating a mirrored aggregate
The following example shows the commands and status output when re-creating aggregates to
reestablish the MetroCluster configuration.
The following output shows the aggregate status of the disaster site’s storage after
reestablishing access to the partner at the surviving site:
filer1>aggr status -r
Aggregate mir1 (online, normal) (zoned checksums)
Plex /mir1/plex0 (online, normal, active)
RAID group /mir1/plex0/rg0 (normal)
RAID Disk Device HA SHELF BAY
--------- ------ ------------parity
8a.3
8a
0
3
data
8a.4
8a
0
4
data
8a.6
8a
0
6
data
8a.5
8a
0
5
CHAN
----FC:B
FC:B
FC:B
FC:B
Used (MB/blks)
-------------34500/70656000
34500/70656000
34500/70656000
34500/70656000
Phys (MB/blks)
------------35003/71687368
35003/71687368
35003/71687368
35003/71687368
Aggregate mir1(1) (failed, partial) (zoned checksums)
Plex /mir1(1)/plex0 (offline, failed, inactive)
242 | High Availability and MetroCluster Configuration Guide
Plex /mir1(1)/plex6 (online, normal, resyncing)
RAID group /mir1(1)/plex6/rg0 (level-0 resync in progress)
RAID Disk
--------parity
data
data
data
Device
-----6a.6
6a.2
6a.3
6a.5
HA SHELF BAY
------------6a 0
6
6a 0
2
6a 0
3
6a 0
5
CHAN
----FC:B
FC:B
FC:B
FC:B
Used (MB/blks)
-------------34500/70656000
34500/70656000
34500/70656000
34500/70656000
Phys (MB/blks)
------------35003/71687368
35003/71687368
35003/71687368
35003/71687368
The mir1(1)/plex6 plex shows that a level-0 resynchronization was in progress; therefore, an
attempt to rejoin the plexes fails, as shown in the following output:
filer1> aggr mirror mir1 -v mir1(1)
aggr mirror: Illegal mirror state for aggregate 'mir1(1)'
Because the mir1(1)/plex6 plex had a level-0 resynchronization in progress, the mir1(1)
aggregate must be destroyed and the mir aggregate remirrored to reestablish a synchronous
mirror, as shown in the following output:
filer1> aggr mirror mir1 -v mir1(1)
aggr mirror: Illegal mirror state for aggregate 'mir1(1)'
filer1> aggr destroy mir1(1)
Are you sure you want to destroy this aggregate? y
Aggregate 'mir1(1)' destroyed.
filer1> aggr mirror mir1
Creation of a mirror plex with 4 disks has been initiated. The
disks need to be zeroed before addition to the aggregate. The
process has been initiated and you will be notified via the system
log as disks are added.
243
Feature update record
This record provides the history of changes made to this guide. When a change is made, it applies to
the release in which it was implemented and all subsequent releases, unless otherwise specified.
Feature updates
Feature first implemented in
Feature release date
•
•
•
Updates for FAS920
Update for NVRAM5
Illustration updates
Data ONTAP 6.5.1
May 2004
•
Data ONTAP 7.0
Updates for NVRAM5
support in FAS900 series
HA pairs, except for
MetroCluster.
Failover event cause-andeffect table.
Declaration of Conformity
update.
Addition of controller
failover and single-point-offailure analysis.
•
•
•
•
•
FAS30xx information.
Corrections were made to
the Upgrading an LRC to
Data ONTAP 7.0.1
November 2004
April 2005
ESH/ESH2/AT-FCX
procedure.
•
Incorporation of the Cluster Data ONTAP 7.1
Administration chapter from
the Data ONTAP 7-Mode
System Administration
Guide and the Disaster
Protection Using
MetroCluster appendix from
the Data ONTAP Data
Protection Online Backup
and Recovery Guide.
June 2005
244 | High Availability and MetroCluster Configuration Guide
Feature updates
Feature first implemented in
Feature release date
•
Updated MetroCluster
information for FAS30xx
Data ONTAP 7.1
October 2005
•
Updated module
replacement information.
Fixed problem in Brocade
switch configuration
information.
Data ONTAP 7.1
December 2005
Updated and extended HA
pair information.
Moved Brocade switch
configuration to Brocade
Switch Description Page.
Moved from cluster to
Data ONTAP 7.1.1
June 2006
Data ONTAP 7.2 RC1
February 2006
•
•
•
•
•
•
•
active/active configuration.
Added information about
Multipath Storage for HA
pairs.
Generalized standard and
mirrored HA pair cabling
instructions.
Updated standard and
mirrored HA pair cabling
instructions to include
FAS60xx.
Feature update record | 245
Feature updates
•
•
•
•
•
Feature first implemented in
Changed name of document Data ONTAP 7.2 RC3
from Cluster Installation
Data ONTAP 7.2.1
Added information about
Multipath Storage for HA
pairs.
•
Data ONTAP 7.2.2
Added quad-port, 4-Gb
Fibre Channel HBA, ESH4
module, DS14mk4 FC disk
shelf information.
Added information to
explain that automatic
giveback should not be used
in MetroClusters.
Updated Multipath Storage
information.
Updated MetroCluster
disaster recovery
information.
Corrected failover and
single-point-of-failure table.
•
•
•
May 2006
and Administration Guide
to Active/Active
Configuration Guide.
Added FAS60xx
information
Updated and extended HA
pairs configuration
information.
Moved Brocade switch
configuration to Brocade
Switch Description Page.
Moved from cluster to
active/active configuration.
•
•
Feature release date
November 2006
March 2007
246 | High Availability and MetroCluster Configuration Guide
Feature updates
Feature first implemented in
Feature release date
•
Data ONTAP 7.2.3
June 2007
•
•
•
•
•
•
•
•
•
•
•
Added procedures for
configuring fabric-attached
MetroClusters on systems
using software-based disk
management.
Added procedure for
unconfiguring an active/
active pair and returning to
stand-alone operation.
Added support for 504 disks Data ONTAP 7.2.4
in MetroClusters.
Added support for the
FAS6040 and FAS6080
systems.
Added support for the
change_fsid option.
Added procedure for
removing an HA pair.
November 2007
Changed name of document Data ONTAP 8.0 RC1
from Active/Active
Configuration Guide to
June 2009
High-Availability
Configuration Guide.
Moved from active/active
configuration to High
Availability pair.
Added information about
configuration on V-Series
systems.
Added support for 672 disks
in MetroClusters.
Added MetroCluster
support for the Brocade 300
and 5100 switches.
Added references to the
DS4243 disk shelf
documentation.
Data ONTAP 8.0 GA
January 2010
Feature update record | 247
Feature updates
•
•
•
•
•
•
•
•
•
•
•
•
Feature first implemented in
Feature release date
Added support for 32xx and Data ONTAP 8.0.1
62xx systems.
Added support for the 8Gbps FC-VI adapter on
MetroCluster
configurations.
August 2010
Added support for 32xx and Data ONTAP 7.3.5
62xx systems.
Added support for the 8Gbps FC-VI adapter on
MetroCluster
configurations.
October 2010
Added more SAS disk shelf Data ONTAP 8.0.2
information and references Data ONTAP 7.3.5.1
to SAS disk shelf
documentation.
Updated multipath HA as a
requirement.
Removed procedures for
non-multipath HA cabling.
Replaced the term
Multipath Storage with
multipath HA for
consistency with other
documentation.
May 2011
Data ONTAP 8.1 RC1
Added support for
MetroCluster with SAS disk
shelves.
Added support for
MetroCluster sharedswitches configurations.
Added procedures for
MetroCluster installation
with third-party storage.
Added support for FAS2240
systems.
September 2011
248 | High Availability and MetroCluster Configuration Guide
Feature updates
Feature first implemented in
Feature release date
•
Removed references to
ESH2 and LRC modules.
Data ONTAP 8.1 GA
December 2011
•
Changed title to Data
Data ONTAP 8.1.1 RC1
May 2012
Data ONTAP 8.1.2 RC1
October 2012
•
ONTAP High Availability
and MetroCluster
Configuration Guide for 7Mode.
Added MetroCluster
support for Cisco switches.
•
Added support for
automatic giveback after
takeover-on-panic.
•
Added support for 3220 and Data ONTAP 8.1.2 GA
3250 systems.
•
Added support for the
cf.mode option.
Added coverage for the
MetroCluster configuration
requirement for both the
cf.mode and
•
Data ONTAP 8.2 RC1
November 2012
April 2013
cf.remote_syncmirror.
enable options.
•
Added support for native
disks in addition to LUNs
on storage arrays for VSeries MetroCluster
configurations.
•
Data ONTAP 8.2 GA
Added coverage for disk
pool assignment in a stretch
MetroCluster configuration.
May 2013
249
Copyright information
Copyright © 1994–2013 NetApp, Inc. All rights reserved. Printed in the U.S.
No part of this document covered by copyright may be reproduced in any form or by any means—
graphic, electronic, or mechanical, including photocopying, recording, taping, or storage in an
electronic retrieval system—without prior written permission of the copyright owner.
Software derived from copyrighted NetApp material is subject to the following license and
disclaimer:
THIS SOFTWARE IS PROVIDED BY NETAPP "AS IS" AND WITHOUT ANY EXPRESS OR
IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE,
WHICH ARE HEREBY DISCLAIMED. IN NO EVENT SHALL NETAPP BE LIABLE FOR ANY
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
NetApp reserves the right to change any products described herein at any time, and without notice.
NetApp assumes no responsibility or liability arising from the use of products described herein,
except as expressly agreed to in writing by NetApp. The use or purchase of this product does not
convey a license under any patent rights, trademark rights, or any other intellectual property rights of
NetApp.
The product described in this manual may be protected by one or more U.S. patents, foreign patents,
or pending applications.
RESTRICTED RIGHTS LEGEND: Use, duplication, or disclosure by the government is subject to
restrictions as set forth in subparagraph (c)(1)(ii) of the Rights in Technical Data and Computer
Software clause at DFARS 252.277-7103 (October 1988) and FAR 52-227-19 (June 1987).
250 | High Availability and MetroCluster Configuration Guide
Trademark information
NetApp, the NetApp logo, Network Appliance, the Network Appliance logo, Akorri,
ApplianceWatch, ASUP, AutoSupport, BalancePoint, BalancePoint Predictor, Bycast, Campaign
Express, ComplianceClock, Cryptainer, CryptoShred, CyberSnap, Data Center Fitness, Data
ONTAP, DataFabric, DataFort, Decru, Decru DataFort, DenseStak, Engenio, Engenio logo, E-Stack,
ExpressPod, FAServer, FastStak, FilerView, Flash Accel, Flash Cache, Flash Pool, FlashRay,
FlexCache, FlexClone, FlexPod, FlexScale, FlexShare, FlexSuite, FlexVol, FPolicy, GetSuccessful,
gFiler, Go further, faster, Imagine Virtually Anything, Lifetime Key Management, LockVault, Mars,
Manage ONTAP, MetroCluster, MultiStore, NearStore, NetCache, NOW (NetApp on the Web),
Onaro, OnCommand, ONTAPI, OpenKey, PerformanceStak, RAID-DP, ReplicatorX, SANscreen,
SANshare, SANtricity, SecureAdmin, SecureShare, Select, Service Builder, Shadow Tape,
Simplicity, Simulate ONTAP, SnapCopy, Snap Creator, SnapDirector, SnapDrive, SnapFilter,
SnapIntegrator, SnapLock, SnapManager, SnapMigrator, SnapMirror, SnapMover, SnapProtect,
SnapRestore, Snapshot, SnapSuite, SnapValidator, SnapVault, StorageGRID, StoreVault, the
StoreVault logo, SyncMirror, Tech OnTap, The evolution of storage, Topio, VelocityStak, vFiler,
VFM, Virtual File Manager, VPolicy, WAFL, Web Filer, and XBB are trademarks or registered
trademarks of NetApp, Inc. in the United States, other countries, or both.
IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of International Business
Machines Corporation in the United States, other countries, or both. A complete and current list of
other IBM trademarks is available on the web at www.ibm.com/legal/copytrade.shtml.
Apple is a registered trademark and QuickTime is a trademark of Apple, Inc. in the United States
and/or other countries. Microsoft is a registered trademark and Windows Media is a trademark of
Microsoft Corporation in the United States and/or other countries. RealAudio, RealNetworks,
RealPlayer, RealSystem, RealText, and RealVideo are registered trademarks and RealMedia,
RealProxy, and SureStream are trademarks of RealNetworks, Inc. in the United States and/or other
countries.
All other brands or products are trademarks or registered trademarks of their respective holders and
should be treated as such.
NetApp, Inc. is a licensee of the CompactFlash and CF Logo trademarks.
NetApp, Inc. NetCache is certified RealSystem compatible.
251
How to send your comments
You can help us to improve the quality of our documentation by sending us your feedback.
Your feedback is important in helping us to provide the most accurate and high-quality information.
If you have suggestions for improving this document, send us your comments by email to
doccomments@netapp.com. To help us direct your comments to the correct division, include in the
subject line the product name, version, and operating system.
You can also contact us in the following ways:
•
•
•
•
NetApp, Inc., 495 East Java Drive, Sunnyvale, CA 94089 U.S.
Telephone: +1 (408) 822-6000
Fax: +1 (408) 822-4501
Support telephone: +1 (888) 463-8277
252 | High Availability and MetroCluster Configuration Guide
Index
A
active/active configurations
status messages 95
active/passive configuration 36
adapters
quad-port Fibre Channel HBA 46, 53
aggregates
ownership change 29, 101
re-creating mirrored after disaster 240
rejoining after disaster 238
root 31
automatic giveback
enhancing speed of 119
automatic takeover
disabling 107
enabling 107
automatic takeover reasons 102
B
background disk firmware update 30
best practices
HA configuration 31
bring up
manually setting options for 76
Brocade
Cisco
fabric-attached MetroCluster configuration
uses Brocade and Cisco switches 158
using fabric-attached MetroCluster
configuration 158
using fabric-attached MetroCluster configuration
158
Brocade switch configuration 168
C
cabinets
preparing for cabling 45
cable 43, 144, 162
cabling
Channel A
for mirrored HA pairs 55
for standard HA pairs 47
Channel B
for mirrored HA pairs 57
for standard HA pairs 49
cluster interconnect for fabric-attached
MetroCluster configurations
with software-based disk ownership 175
cross-cabled cluster interconnect 52, 61
cross-cabled HA interconnect 51, 61
error message, cross-cabled cluster interconnect 51,
52, 61
fabric-attached MetroCluster configurations 166
FC-VI adapter for fabric-attached MetroCluster
configurations
with software-based disk ownership 175
HA interconnect for fabric-attached MetroCluster
configurations 171
HA interconnect for standard HA pair 51, 61
HA interconnect for standard HA pair, 32xx systems
52, 61
HA pairs 41
local controller in fabric-attached MetroCluster
configuration
with software-based disk ownership 168
local disk shelves in fabric-attached MetroCluster
configuration
with software-based disk ownership 169
local node 168
Node A 168
Node B 172
preparing equipment racks for 45
preparing system cabinets for 45
remote controller in fabric-attached MetroCluster
configuration
with software-based disk ownership 172
remote disk shelves in fabric-attached MetroCluster
configuration 173
remote node 172
requirements 43, 144, 162
stretch MetroCluster configurations 146
with software-based disk ownership 173
cabling disk shelves to switch
with software-based disk ownership 173
cf forcegiveback command
using to force giveback 117
cf.giveback.auto.cifs.terminate.minutes options 119
cf.giveback.check.partner option 115
cf.takeover.use_mcrc_file 93
Index | 253
change_fsid option 79
Channel A
cabling 47, 55
defined 33
Channel B
cabling 49, 57
chassis configurations, single or dual 37
CIFS clients and giveback delay 119
CIFS sessions terminated on takeover 29, 101
cluster interconnect, cabling 171, 175
command exceptions
emulated node 112
commands
accessing takeover node with partner 111
cf (enables and disables takeover) 106
cf forcesgiveback (forces giveback) 117
cf forcetakeover -d (forces takeover) 235
cf forcetakeover (forces takeover 104
cf giveback (enables giveback) 93
cf partner (displays partner's name) 99
cf status (displays status) 109
cf takeover (initiates takeover 104
cf takeover (initiates takeover) 93
disabling HA mode 65
enabling non-HA mode 65
exceptions for emulated node 112
for monitoring HA pair status 95
ha-config 80
halt (halts system without takeover) 105, 106
options cf.mode ha | non-ha 75
partner (accesses emulated node) 110
storage show disk -p (displays paths) 125
sysconfig 99
takeover (description of all takeover commands) 104
comparison of types of HA pairs 17
Config Advisor
downloading and running 93
configuration speeds
changing stretch MetroCluster configuration default
149
changing stretch MetroCluster default 214
configuration variations
fabric-attached MetroCluster configurations 137
mirrored HA pairs 19
standard HA pairs 36
stretch MetroCluster configurations 136
configurations
HA differences between supported system 38
reestablishing MetroCluster configuration 238
testing takeover and giveback 93
configuring
automatic giveback 118
interfaces 87
shared interfaces 86
connections
between a V-Series system and a FibreBridge 6500N
bridge 139, 155
considerations
for running setup on HA pairs 72
controller failover
benefits of 13
monitoring status of 95
controller failovers
events that trigger 21
controller-to-switch cabling, fabric-attached MetroCluster
configurations 168, 172
D
Data ONTAP
in MetroCluster configurations 135
upgrading nondisruptively, documentation for 12,
133
dedicated interfaces
configuring using setup 74
described 83
diagram 85
default configuration speed
managing 149
delay, specifying before takeover 107
disabling takeover (cf) 106
disaster recovery
using MetroCluster configurations 232
when not to perform 232
disasters
determining whether one occurred 232
recognizing 232
recovering from, with MetroCluster configuration
233
recovery from
forcing takeover 235
manually fencing off the disaster site node 234
reestablishing MetroCluster configuration 238
restricting access to the failed node 234
disk firmware update 30
disk information, displaying 99
disk paths, verifying in a fabric-attached MetroCluster
configuration
with software-based disk ownership 177
254 | High Availability and MetroCluster Configuration Guide
disk shelf pool assignments, fabric-attached MetroCluster
configurations 176
disk shelves
about modules for 124
adding to an HA pair with multipath HA 122
hot swapping modules in 127
managing in a MetroCluster configuration 122
managing in an HA pair 122
disk-shelf-to-switch cabling, fabric-attached MetroCluster
configurations 169, 173
disks
MetroCluster configuration with V-Series systems
219
distances between nodes in the HA configuration 17
documentation, required 42, 142, 160
dual-chassis HA configurations
diagram of 37
interconnect 37
Duplicate Address Detection (DAD) 86
E
e0M management interface 85
eliminating single point of failure 13
EMS message, takeover impossible 31
emulated node
accessing from the takeover node 110
backing up 114
description of 110
dumps and restores 114
managing 110
emulated node, accessing with Remote Shell 112
emulated nodes
command exceptions for 112
enabling HA mode capability and controller failover 75
enabling takeover (cf) 106
equipment racks
installation in 41
preparation of 45
events
table of failover triggering 21
exceptions
emulated node command 112
F
fabric-attached MetroCluster configuration
assigning disk pools 176
cabling cluster interconnect for
cabling FC-VI adapter for
with software-based disk ownership 175
with software-based disk ownership 175
cabling HA interconnect for
cabling FC-VI adapter for 171
for disks 157
implementing, with disks 153
local node
cabling controller to switch
with software-based disk ownership 168
cabling disk shelves to switch
with software-based disk ownership 169
planning 153
remote node
cabling controller to switch
with software-based disk ownership 172
verifying disk paths
with software-based disk ownership 177
fabric-attached MetroCluster configurations
advantages of 136
array LUNs
recommended configuration 188
requirements 186
Brocade switch configuration 168
cabling 166, 168, 169, 172, 173
illustration of 166
limitations 155
planning worksheet 159
restrictions 153
setup requirements for 153
storage arrays
recommended configuration 221
variations 137
failover
benefits of controller 13
determining status (cf status) 109
monitoring status of 95
failovers
events that trigger 21
failures
table of failover triggering 21
fault tolerance 11
FC-VI adapter, cabling 171, 175
fencing, manual 234
Fibre Channel ports
identifying for HA pair 46, 53
Fibre Channel switches 144, 162
forcing
giveback 117
takeover 104
FRU replacement, nondisruptive
Index | 255
documentation for 12, 133
giveback
cf.giveback.check.partner option and 115
configuring automatic 118
definition of 20
delay time for CIFS clients 119
disabling automatic, after takeover on panic 120
enabling automatic, after takeover on panic 120
enhancing speed of automatic 119
forcing 117
initiating normal 115
interrupted 117
managing 115
partial-giveback 117
performing a 115
setting to terminate long-running processes 119
testing 93
troubleshooting 120
what happens during 29, 101
guidelines
MetroCluster configuration with disks and array
LUNs 219
setup requirements 32
setup restrictions 32
status messages 95
types of
compared 17
installed in equipment racks 41
installed in system cabinets 41
mirrored 19
HA state 38, 80
ha-config modify command 38, 80
ha-config show command 38, 80
halting system without takeover 105, 106
hardware
components described 16
HA components described 16
single point of failure 13
hardware assisted takeover 28, 36, 81
hardware replacement, nondisruptive
documentation for 12, 133
hardware-assisted takeover
checking statistics 98
checking status of 97
hardware-assisted takeovers
setting partner IP address for 82
history of changes to this guide 243
H
I
HA configurations
benefits of 11
converting to MetroCluster configuration 164
definition of 11
differences between supported system 38
single- and dual-chassis 37
HA interconnect
cabling 51, 61
cabling, 32xx dual-chassis HA configurations 52, 61
single-chassis and dual-chassis HA configurations
ifconfig command 86
installation
equipment rack 41
system cabinet 41
installing
HA pairs 41
interface configurations
dedicated 83
shared 83
standby 83
interfaces
configuration for takeover 85
configuring 87
configuring dedicated 74
configuring shared 73
configuring standby 74
dedicated, diagram 85
IPv6 considerations 86
shared, diagram 84
standby, diagram 85
types and configurations 85
internode distance 17
G
37
HA mode
disabling 65
enabling 75
HA pairs
cabling 41, 46
cabling mirrored 52
changing nodes to stand-alone 63, 64, 67, 69, 70
events that trigger failover in 21
installing 41
managing disk shelves in 122
required connections for using UPSs with 62
256 | High Availability and MetroCluster Configuration Guide
IP address
partner IP, specifying 88
IPv6 considerations 86
L
licenses
cf 75
not required 75
LIF configuration, best practice 31
local node
cabling 168
long-running processes
setting giveback to terminate 119
lun commands, lun online 236
LUNs (array)
MetroCluster configuration implementation
overview 185
MetroCluster configuration with disks and array
LUNs 219
LUNs, bringing online 236
M
mailbox disks in the HA pair 11
manual fencing 234
messages
active/active configuration status 95
HA pair status 95
MetroCluster
managing disk shelves in 122
MetroCluster configuration
about 134
performing NDSR 128, 227
performing nondisruptive shelf replacement 128, 227
preparing for NDSR 129, 227
replacing the disk shelf nondisruptively 130, 228
resetting the default speed of stretch MetroCluster
configuration 151
types 134
MetroCluster configurations
array LUNs
cabling guidelines 195
connecting devices in 198
connecting local systems in 199
connecting remote systems in 203
connecting the fabric and storage array 207
connecting the switch fabric 205
FC-VI ports, testing zoning of 215
implementation overview 185
planning zoning with 196
requirements and restrictions 186
tasks after connecting devices 215
testing setup 216
testing zoning of FC-VI ports 215
testing, simulating disaster recovery 217
zoning 213
changing the default speed of stretch 149, 214
converting to, from standard or mirrored HA pair
164
events that trigger failover in 21
How Data ONTAP works with 135
how mirroring works in 134
LUNs and 236
reestablishing configuration after disaster 238
shared-switches configurations
cabling with storage arrays 209
requirements with array LUNs 188
storage arrays
recommended fabric-attached configuration 221
recommended stretch MetroCluster
configuration 192
supported stretch configuration 223
third-party storage
recommended fabric-attached configuration 188
See also array LUNs
V-Series systems
FibreBridge 6500N bridge with 139, 155
guidelines for mixed disks and LUNs 219
hardware requirements for disks 139, 155
mirrored aggregates
re-creating after disaster 240
mirrored HA pairs
about 19
advantages of 19
cabling 52
cabling Channel A 55
cabling Channel B 57
restrictions 33
setup requirements for 33
variations 19
mirroring, NVMEM or NVRAM log 11
modules, disk shelf
about 124
best practices for changing types 124
hot-swapping 127
restrictions for changing types 124
testing 125
monitoring in normal mode 95
multipath HA
Index | 257
advantages of 35
connection types used by 35
description of 34
multipath HA loop
adding disk shelves to 122
N
NDSR
performing 128, 227
preparing for 129, 227
verifying the disks 131, 230
negotiated failover 108
network interface
automatic takeover 108
nfo 108
network interfaces
configuration for takeover 85
IPv6 considerations 86
types and configurations 85
Node A
cabling 168
Node B
cabling 172
nodes
accessing takeover, with partner command 111
command exceptions for emulated 112
Non-HA mode
enabling 65
nondisruptive hardware replacement
documentation for 12, 133
nondisruptive operations 11
nondisruptive shelf replacement
performing 128, 227
See also NDSR
nondisruptive upgrades
Data ONTAP, documentation for 12, 133
normal giveback
initiating 115
normal mode
monitoring in 95
NVMEM log mirroring 11
NVRAM adapter 43, 144, 162
NVRAM log mirroring 11
O
options, matching 76
options, setting 75
P
parameters
change_fsid 79
required to be identical between nodes 77
setting 75
partner command 110
partner commands
accessing takeover node with 111
partner IP addresses
setting for hardware-assisted takeover 82
partner name, displaying (cf partner) 99
planning worksheet for fabric-attached MetroCluster
configurations 159
plexes, requirements for in the HA pair 33
pool assignments, fabric-attached MetroCluster
configurations 176
port list
creating for mirrored HA pairs 54
ports
identifying which ones to use 46, 53
power supply best practice 31
preferred primary port
removing the settings 183, 213
preparing equipment racks 45
primary connections, in multipath HA 35
R
racking the HA pair
in a system cabinet 41
in telco-style racks 41
reasons for automatic takeover 102
redundant connections, in multipath HA 35
reestablishing MetroCluster configuration 238
remote node
cabling 172
removing an HA pair 63
requirement, multipath HA 34
requirements
adapters 144, 162
documentation 42, 142, 160
equipment 43, 144, 162
Fibre Channel switches 144, 162
HA pair setup 32
hot-swapping a disk shelf module 127
NVRAM adapter 144, 162
SFP modules 144, 162
tools 43, 144, 161
restrictions
258 | High Availability and MetroCluster Configuration Guide
fabric-attached MetroCluster configuration 153
HA pair setup 32
in mirrored HA pairs 33
in stretch MetroCluster configurations 138
rsh, using to access node after takeover 101
S
setting options and parameters 75
setup
considerations for running on active/active
configurations 72
SFP modules 43, 144, 162
shared interfaces
configuring using ifconfig 86
configuring using setup 73
described 83
diagram 84
shared-switch 178, 209
shared-switch configuration
requirements 157
shared-switches
setting preferred primary port 182, 212
shared-switches configuration
cabling FC-VI adapter and ISL 178
sharing storage loops or stacks 36
shelves
managing in a MetroCluster configuration 122
managing in an HA pair 122
shorting giveback time 115
single point of failure
analysis 13
definition of 13
eliminating 13
single-chassis HA configurations
diagram of 37
interconnect 37
SNMP protocol and takeover mode 109
software-based disk management 176
software-based disk ownership 168, 169, 172, 173
spare disks in the HA pair 11, 33
stand-alone operation
changing an HA pair node to 63, 64, 67, 69, 70
standard HA pair
cabling Channel A 47
cabling Channel B 49
cabling HA interconnect for 51, 61
cabling HA interconnect for, 32xx systems 52, 61
variations 36
standby connections, in multipath HA 35
standby interfaces
configuring using setup 74
described 83
diagram 85
status
active/active configuration message 95
HA pair message 95
monitoring HA pair 95
of hardware-assisted takeover 97
storage arrays
recommended fabric-attached MetroCluster
configuration 188, 221
recommended MetroCluster configuration 192
recommended stretch MetroCluster configuration
223
stretch MetroCluster configuration
disk shelf pool assignments
fabric-attached MetroCluster configurations
147
implementing 138
planning 138
pool assignments 147
resetting default speed 151
stretch MetroCluster configurations
advantages of 135
array LUNs
requirements 186
cabling 146
changing the default speed of 149, 214
connections required 141
illustration of 141
on dual-controller systems 147
restrictions 138
setup requirements 138
storage arrays
recommended configuration 192, 223
variations 136
switch configuration, for fabric-attached MetroCluster
configurations 168
switch zoning
planning for MetroCluster configuration with array
LUNs 196
switches
V-Series MetroCluster requirements 186
SyncMirror
requirements for a MetroCluster configuration 186
system cabinets
installation in 41
preparing for cabling 45
system configurations
Index | 259
HA differences between supported 38
T
takeover
CIFS sessions and 29, 101
configuring when it occurs 102
configuring with dedicated and hot standby
interfaces 85
definition of 20
determining why one occurred 109
disabling 106
disabling automatic 107
enabling 106
enabling automatic 107
forcing 104
forcing for disaster recovery 235
hardware assisted 28, 36, 81
reasons for 102
rsh access after 101
SNMP settings and 109
specifying delay before 107
statistics 109
Telnet access after 101
testing 93
troubleshooting 120
using /etc/mcrc file at takeover 93
what happens after 101
what happens during 29, 101
takeover impossible EMS message 31
takeover mode
managing in 109
statistics in 109
takeover nodes
accessing with partner command 111
takeovers
setting partner IP address for hardware-assisted 82
when they occur 20, 100
Telnet, using to access node after takeover 101
testing
hardware-assisted takeover 98
takeover and giveback 93
third-party storage
MetroCluster configurations with
requirements and restrictions 186
tools, required 43, 144, 161
triggers for automatic takeover 102
U
unconfiguring an HA pair 63
uninterruptible power supplies
See UPSs
update history for this guide 243
UPS
using with MetroCluster configurations 140, 156
UPSs
required connections with HA pairs 62
utilities
downloading and running Config Advisor 93
V
V-Series MetroCluster configurations
See MetroCluster configurations
verifying
takeover and giveback 93
VIF configuration, best practice in an HA configuration
31
Z
zoning
MetroCluster configurations, array LUNs 213
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising