3Com 800ix-SMO Dishwasher User Manual

NSR-ORA V3.2
Database Adapter Module for NetWorker
Edition October 2001
Copyright and Trademarks
Copyright © Fujitsu Siemens Computers GmbH 2001.
All rights reserved.
Delivery subject to availability; right of technical modifications reserved.
All hardware and software names used are trademarks of their respective manufacturers.
This manual was produced by
cognitas. Gesellschaft für Technik-Dokumentation mbH
www.cognitas.de
1
Preface
This guide provides information about how to install, configure and operate the
“NetWorker Save and Restore for ORAcle (NSR-ORA)” V3.2 NetWorker component.
1.1
Audience
The information in this guide is intended for system and database administrators.
1.2
Conventions
This manual uses the following typographical conventions and symbols to make
information easier to access and understand.
boldface
Indicates emphasized words, e.g.:
Using Oracle 8 be always careful for every instance
being offline or online!
fixed-width
Examples and information displayed on the screen, window titles, names of buttons, menues, fields for input or
output, scroll lists. For example:
Click on the Cancel button to close the Help window.
fixed width,
boldface
Commands, text and other input you type exactly as
shown, e.g.:
sh> start_amon.sh start ora
italic
Directory pathnames, names of files, parameters, commands and machines, or new terms defined in the
glossary or within the chapter, e.g.:
The SQLDBA parameter contains the pathname of sqldba,
svrmgrl or sqlplus.
!
This Symbol is used to indicate and cautionary notes that
e.g. prevent you form making a mistake
i
This Symbol indicates important information.
System Administrator’s Guide
U42252-J-Z915-1-76
Preface
1.3
What is New in Release 3.2
What is New in Release 3.2
With release V3.2 a number of changes have been made to NSR-ORA, influencing this guide:
●
NSR-ORA V3.2 does now support Oracle 8.x or 9i
●
Version 9 of the Oracle database does not include the SVRMGRL command
any more. So NSR-ORA now uses the command stored in the SQLDBA variable to communicate with the database. However, functionality of NSR-ORA
has not been changed.
●
NSR-ORA V3.2 now supports the backup on Network Attached Storage
filers (NAS filers) by NetworkAppliance® (NetApp). Furthermore NetWorker
supports backups performed directly onto a NetApp filer by using snapshots
(for further information see chapter “Using NSR-ORA with NetApp Filers”).
●
The configure_nsrora configuration skript provides some new parameters,
described in section “Configuration” .
●
The NetWorker V6.0 software provided by Fujitsu Siemens Computers now
comes with two new algorithms for compression. Using these new compression modes causes in part a considerable improvement of compression
rates and simultaneously reduces CPU load. This feature is only supported
bei Fujitsu Siemens Computers NetWorker for Solaris and LINUX.
1.4
Update from Previous Releases
If you want to update NSR-ORA from a previous release you will find the necessary information in section “How to Proceed for an Update from an Earlier
Version” .
System Administrator’s Guide
U42252-J-Z915-1-76
2
Introduction
This guide to Backing up and Restoring a Database describes the functionality,
installation, configuration and adminstration of the NetWorker component “NetWorker Save and Restore for ORAcle (NSR-ORA)” V3.2. This software component allows to integrate ORACLE databases into the Networker data backup
concept.
NetWorker is intended for automatic data backup in heterogeneous networks.
Among its advantages are the large number of supported backup media and
simple user prompting in spite of the wide range of functions. However, the
backup of database files with NetWorker is only possible offline. Where a database needs to be recovered, NetWorker can restore the faulty files and return
the database to a consistent, updated status, although additional steps need to
be performed with ORACLE tools.
NSR-ORA closes the gap between NetWorker and ORACLE as it can communicate with both of them. It can be used to back up ORACLE databases online
or offline. Backup is possible in normal and compressed modes. Whether the
database is located in a Unix file system or on raw devices is immaterial. When
using NetApp NAS filers the database can be backed up on tapes and also as
a snapshot on the NAS filer directly. The database (thus also NSR-ORA) and
the NetWorker server need not be installed on the same computer; database
backup is also possible from NetWorker clients.
NSR-ORA enables you to initiate the recovery of an inconsistent database to a
specific second automatically. NSR-ORA automatically checks the status of the
database and commences the required restore and recovery measures to
restore the current status. This means that you can perform a database recovery at any time even without any knowledge of ORACLE (you will find further
information about recovery in the chapter “Recovery”).
For a better understanding of the database interface, the fundamental ORACLE
database terms are described briefly below:
●
The physical database structure
●
The logical database structure
●
The database backup
●
Archive mode of the database.
System Administrator’s Guide
U42252-J-Z915-1-76
Introduction
2.1
The Physical Database Structure
The Physical Database Structure
The physical database (see figure 1) consists of at least one control file, one
parameter file, at least two redolog files and at least one database file.
control file
redolog files
configuration file
database file
Figure 1: The physical database structure
●
The database file(s) contain the data dictionary as well as the actual user
data (tables, indices, views etc.).
●
The redolog files contain information on database recovery. They instantly
record any change made regarding the database. A database always
possesses at least two redolog files which are written cyclically.
●
The control file(s) contain structural and management information.
●
In the configuration file specifies the configuration of a database.
A database also possesses suitable programs (executables), which ensure its
functionality and form the Database Management System (DBMS).
2.2
The Logical Database Structure
At the logical level (see figure 2), ORACLE only recognizes tablespaces (database-specific, user-defined and temporary) consisting of at least one physical
database file. This tablespace structure also constitutes a backup unit for NSRORA. The allocation between tablespaces and the corresponding files is performed by NSR-ORA.
System Administrator’s Guide
U42252-J-Z915-1-76
Introduction
The Database Backup
database files
control file
redolog files
USER
SYSTEM
TEMP
Tablespaces
Figure 2: Logical database structure
2.3
The Database Backup
Every recovery procedure depends on a correctly performed database backup,
which consists of copies of the database files being made at operating system
level and transferred to an external storage medium or created as a NAS
snapshot. These copies depict a state that is not identical to the current one.
In the case of one or more database files being lost, the backup copies serve
as the starting time for a required recovery. During any recovery, all redolog files
accruing since the backup are reapplied, thus updating from the backup to the
current status.
2.4
Archive Mode of the Database
To ensure that all the redolog files accruing since the backup are available, the
database must be operated in ARCHIVELOG mode, which makes sure that
redolog files are not overwritten unless they have first been archived on a
separate disk area (archive directory) by ORACLE’s archiving function (see
figure 3).
System Administrator’s Guide
U42252-J-Z915-1-76
Introduction
Archive Mode of the Database
NSR-ORA checks that the database is in ARCHIVELOG mode, and will abort
the backup of the database if this is not the case if an online backup is to be carried out. An offline backup can be performed even if the database is not in
ARCHIVELOG mode.
archive
directory
database files
Oracles
archiving
process
control file
redolog files
SYSTEM
USER
TEMP
Tablespaces
Figure 3: Database running in ARCHIVELOG mode
The database system deals only with archiving, it does not provide for the
subsequent management of the redolog files. The database will come to a halt
as soon as the archive directory is full. NSR-ORA therefore contains a module
which monitors the archiving process, the archiving monitor amon. At regular
time intervals, the daemon of the archiving monitor transfers the redolog files
archived by ORACLE to external backup media.
To make a full database backup, all file types must be included in the saving
procedure.
Both the control file and the database programs (executables) can be backed
up using standard NetWorker functions (i.e. without NSR-ORA).
For securing the database files, NSR-ORA supports full backup as well as
restoration and recovery of the database by reapplying the redolog files.
For a backup, a correlation is first made between tablespaces and database
files. In this manner, the logical tablespace concept of the database is mapped
onto a physical structure, and the database files are then backed up. In the case
of an online backup, the ORACLE-specific commands begin backup and end
backup are included.
When a recovery is required, shutdown of the database is followed by a crash
analysis. If a database file has been lost, the required backup files are first
restored from the backup medium. The recovery of the database is subsequently commenced; this involves reapplying the appropriate redolog files. If
System Administrator’s Guide
U42252-J-Z915-1-76
Introduction
Function Scope
these redolog files had already been saved on the backup medium by the
archiving monitor and then automatically deleted, they will be read in again and
reapplied until the database is up to date. The database is then opened and is
again ready for normal operation. The recovery is triggered by the ora_recut and
nsrora_rec commands and proceeds automatically; a knowledge of ORACLE is
not necessarily required for this (see also chapter “Recovery”).
2.5
Function Scope
Physical faults that bring normal database operation to a halt basically fall into
the following four categories:
●
loss of a (all) control file(s)
●
loss of a current online redolog file
●
loss of files containing system data
●
loss of files containing user data.
As a rule, the loss of a control file does not pose major problems because the
control file can be mirrored at the software level using the control_files parameter
in the ORACLE parameter file init<ORACLE_SID>.ora. ORACLE strongly
recommends saving at least two copies of the control file on different disks. We
agree with this recommendation.
The loss of a current online redolog file is also precluded almost completely by
mirroring (at the software level by ORACLE or at the hardware level by mirror
disks).
!
For this reason, neither of these fault categories (loss of the control file
and loss of current online redolog files) is taken into account during an
automatic recovery of the database. Only in recovery until time with
ora_recut is the relevant control file also recovered. Nor is NSR-ORA in a
position to recognize or correct logical database errors (for instance, if
incorrect files are loaded into the database).
The ORACLE Administrator Guide explains what you can do if you should lose
the control file and/or a number of redolog files. You can also refer to the chapter
“Recovery” of this manual, which describes the restrictions that apply to
automatic recovery for ORACLE Parallel Server (OPS) systems.
In the other two cases (loss of files containing system or user data), NSR-ORA
offers various ways of recovering the data: fully automatic or manual recovery
or restore of a specific database level (recovery until time).
System Administrator’s Guide
U42252-J-Z915-1-76
3
The individual Components
The individual components for backing up and recovering an ORACLE
database in a NetWorker environment are described in the following section.
Four different areas are dealt with:
●
The archiving monitor
●
The backup component
●
The recovery component
●
The overall configuration
Each section starts with a functional description. This is followed by an explanation of the required configuration steps, after which the procedure and operations are illustrated with the help of practical examples. Two configuration tools
are provided for your convenience: configure_nsrora and configure_networker.
The installation and configuration of NSR-ORA and the two configuration tools
are dealt with in chapter “Installation and Configuration”. We recommend you to
read through chapter “Introduction” before you start carrying out the installation
and configuration described in chapter “Installation and Configuration”.
3.1
The Archiving Monitor
During the operation of an ORACLE database, the online redolog files are
archived by the database system in an archive directory to protect the database
from data losses. For this, the database must be in ARCHIVELOG mode (“Automatic archival ENABLED”). However, the database system is not responsible for
the subsequent management of the offline redolog files created in this manner.
If the archive directory becomes full, your database will come to a halt. This
administrative gap is bridged by the archiving monitor (amon) (see figure 4).
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
The Archiving Monitor
DBMS
database files
Oracle’s
archiving
process
control file
SYSTEM
USER
TEMP
redolog files
Treshold
value
Tablespaces
archive
directory
Database interface NSR-ORA
Archiving
monitor
Volume management
Volume pool management
Multiplexing
Jukebox support
Tape library
Figure 4: The archiving monitor
The archiving monitor amon has the task of permanently and regularly
monitoring and backing up the offline redolog files, which accrue in the archive
directory. Once the configurable number of archived redolog files has been
reached (the threshold value is specified by AMON_MAX_LOGS), the archiving
monitor daemon initiates a backup and then deletes the accrued files. It retains
a number of the newest redolog files, specifically the number given by the value
of AMON_MIN_LOGS. The daemon checks at predefined intervals (specified by
the parameter AMON_INTERVALL) whether the specified threshold value has
been exceeded, and checks this again at the end of each backup.
i
On OPS systems the archiving monitor amon is used on every OPS node,
so we recommend to use the same pathnames to the redolog files on
every node.
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
3.1.1
The Archiving Monitor
Starting and Ending the Archiving Monitor
A control program start_amon.sh is provided in the directory ${NSR_INST}/oracle/bin for coordinating the activities of the archiving monitor. This program can
be invoked at operating system level with the relevant parameters (start, status,
stop).
start_amon.sh [-t {nsr|emc}] <command> [<ORACLE_SID> ...]
The -t option is required if NSR-ORA (nsr) and NSR-ORA-EMC (emc) are
installed on one machine, so that it is possible to distinguish which archiving
monitor is to be started.
If ORACLE_SID is not specified, the environment variable $ORACLE_SID is evaluated. If this variable is not set, amon is started for all configured ORACLE
instances; otherwise only for the specified ORACLE_SIDs.
The following commands are available:
start
stop
status
Starts the archiving monitor
Stops the archiving monitor
Status inquiry
The start command is used to start the archiving monitor. This requires the
database to be in ARCHIVELOG mode.
The configuration file dbo${ORACLE_SID}.init is then read out and the archiving
monitor (amon daemon) is started. All important events are recorded in a
protocol file located in the directory ${DBO_CONFIG_DIRECTORY}/prot.
The stop command deactivates the archiving monitor.
The status check status shows, among other things, the following information:
●
the time the archiving monitor is activated
●
the database to be monitored
●
the archive directory
●
the maximum number of offline redolog files
●
the current number of offline redolog files
Example:
The following entries are assumed in the configuration file in this example:
NSR_SERVER=psi
DBO_CONFIG_DIRECTORY=/nsr/oracle/ora
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
The Archiving Monitor
AMON_GROUP=amon_ora_psi
AMON_MAX_LOGS=4
AMON_INTERVALL=1800
AMON_MIRROR_GROUP=amon2_ora_psi
AMON_MIRROR_SERVER=psi
With these prerequisites fulfilled, the user can start the archiving monitor:
sh> start_amon.sh start ora
ORACLE_SIDs : ora
start amon for oracle instance ora ...
A subsequent status enquiry generates the following output:
# start_amon.sh status ora
ORACLE_SIDs : ora
Configuration and Status summary of amon at 15.09.00 (11:30:41):
System
: psi
Database
: ora
Networker Server
: psi
Save Group
: amon_ora_psi
Mirror Server
: psi
Mirror Group
: amon2_ora_psi
User
: root
Protocol File
: /nsr/oracle/ora/prot/amon.prot
Networker Directory : /opt/nsr
Archive Directory
: /home/oracle/dbs
Archive Format
: arch
Interval (seconds) : 1800
Maximum # of Logs
: 4
Actual # of Logs
: 2
daemon is up since : 13.09.00 (15:15:00)
If the archiving monitor is inactive, the following display appears:
sh> start_amon.sh status ora
amon is down at 14:46:54 on 13.09.00
If the functions of the archiving monitor are temporarily not required (e.g. if the
database is shut down, a recovery is carried out, or a change is made to the
archive directory), it can be deactivated:
sh> start_amon.sh stop ora
ORACLE_SIDs : ora
stop amon for oracle instance ora ...
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
3.1.2
The Archiving Monitor
Logging
While the archiving monitor is active, the occupancy level of the archive directory and other important events are logged at regular intervals
(AMON_INTERVALL) in the protocol file
${DBO_CONFIG_DIRECTORY}/prot/redologs/amon.prot.
Protocol file
There are four generations of the protocol file. The current protocol file is
appended to the file amon<sid>.old with each restart. If this file exceeds a size
of 5 Mbytes, it is renamed from amon<sid>.002 to amon<sid>.003 and
amon<sid>.old is changed to amon<sid>.002.
Archiving log file
All of the files backed up by the archiving monitor are logged in the /nsr/oracle/<ORACLE_SID>/prot/save_logfiles file. There are four generations of this file.
With each restart, the file is appended to save_logfiles.old, if it does not exceed
1 Mbyte as a result. Were this size to be exceeded, save_logfiles.old is renamed
save_logfiles.002 and the current file is renamed save_logfiles.old.
The following is a representative sample from the log file:
00-05-08:16:51:59: Amon Instanz /opt/networker/oracle/bin/amon
(26293) startet ============================
00-05-08:16:51:59: INFO: NSR_SERVER <psi>
00-05-08:16:51:59: INFO: Variable CONTR_SERVER set to <psi>
00-05-08:16:51:59: INFO: Amon Group name set to <amonorapsi>
00-05-08:16:51:59: INFO: AMON_MAX_LOGS <5>
00-05-08:16:51:59: INFO: AMON_MIN_LOGS <1>
00-05-08:16:51:59: INFO: AMON_INTERVAL <1800>
00-05-08:16:51:59: INFO: mail user <root>
00-05-08:16:51:59: INFO: amon_watch_dog_cmd </usr/bin/mailx -s
’Redologs not saved in 900 seconds’ root>
00-05-08:16:51:59: INFO: AMON_WATCH_DELAY <900>
00-05-08:16:51:59: INFO: AMON_SAVE_ARG_NR <50>
00-05-08:16:51:59: INFO: AMON_PS_FILE </nsr/oracle/ora/tmp/amon.ps>
00-05-08:16:51:59: INFO: AMON_RM_WITHOUT_MIRROR_DELAY <0>
00-05-08:16:52:00: INFO: forked expect proc <26301>
00-05-08:16:52:00:
00-05-08:16:52:01:
00-05-08:16:52:01:
00-05-08:16:52:01:
ARCHIVE_DIR=/oradb/db734/ora/saparch
ARCHIVE_FORMAT=oraarch%t_%s.dbf
AMON_EXP_TIME=
forked got process nr 26309
00-05-08:16:52:01:
00-05-08:16:52:01:
00-05-08:16:52:01:
00-05-08:16:52:01:
INFO: PS_FILE </nsr/oracle/ora/tmp/amon.ps>
INFO: checking Nr of Redologs
/oradb/db734/ora/saparch
DEBUG:oraarch1_42.dbf 01/11/00 14:55:59 s: 0 m: 0
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
The Backup Component
......
00-05-08:16:52:01: DEBUG:oraarch1_54.dbf 01/21/00 13:13:15 s: 0 m: 0
00-05-08:16:52:01: DEBUG: Print of a save_cmd:
/opt/networker/bin/save -i -l full -g amonorapsi -s psi
00-05-08:16:52:01:
oraarch1_42.dbf oraarch1_43.dbf
oraarch1_45.dbf oraarch1_44.dbf oraarch1_46.dbf oraarch1_47.dbf
oraarch1_48.dbf oraarch1_49.dbf oraarch1_50.dbf oraarch1_51.dbf
oraarch1_52.dbf oraarch1_53.dbf oraarch1_54.dbf.
3.2
The Backup Component
The backup component of NSR-ORA allows the online as well as offline full or
partial backup of an ORACLE database within an archiving process. (see
figure 5). In case of an online backup, the database must be in the
ARCHIVELOG mode. The extent of an archiving process is defined by means
of a configurable file structure in the directory $DBO_CONFIG_DIRECTORY of
the NSR-ORA administrator (refer to the section “Configuration and management”).
i
In the case of Oracle 8.0 backups, the Oracle shared libraries ibclm.so,
libclntsh.so and libmipc.so must be linked to /usr/lib. Where Oracle is used
with 64-bit support, these libraries must be linked accordingly to
/usr/lib64s.
When backing up Oracle 8.1 under Linux, the Oracle shared libraries
libjox8.so, libskgxp8.so and libclntsh.so.8.0 must be linked to /usr/lib. This is
the only way that data backup is possible with NSR-ORA.
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
The Backup Component
DBMS
database files
Oracle’s
archiving
process
control file
SYSTEM
USER
TEMP
redolog files
Treshold
value
Tablespaces
archive
directory
Database interface NSR-ORA
Backup
Archiving
component
monitor
NetWorker user interface
Volume management
Volume pool management
Multiplexing
Jukebox support
Tape library
Figure 5: The backup component
3.2.1
Offline Backup of the Database
configure_nsrora and configure_networker can be used, if required, to create a
new client and a new group called OFFLINE<oracle_sid><client>. If this client is
backed up, an offline backup is performed, i.e. the database is shut down before
the backup and then rebooted after the backup.
Furthermore, an offline backup of the database is performed whenever the
database is closed at the time of the backup. This procedure consists of the
following phases:
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
The Backup Component
1. The database is started up briefly to determine the database files allocated
to a tablespace. Normally, several tablespaces are backed up during an
archiving process.
2. The database is closed.
3. All the database files of a tablespace are backed up, irrespective of whether
system files or raw devices are involved.
!
The database must not be started up during an offline backup! You will
receive an error message at the end of the backup if the database was
started up in the meantime.
!
If you want to back up a database that is currently online under OPS, and
the instance that is to be used for the backup is offline, that instance can
only be started automatically under Oracle 7 and an online backup will
be carried out instead. You may get an inconsistent backup under Oracle
8. You should therefore always note under Oracle 8 before the backup
that all instances are either offline or online!
If you use the default configuration described in Chapter 4, a copy of the control
file is also made every day, and - when you carry out an offline backup - the
online redolog files are also backed up.
Please note that the database is started up briefly once in the course of an
offline backup, in order to request the paths to the database files and the raw
devices. This alters the time-stamps on the database files and the control file. It
follows that the state of the database after an offline backup will not be identical
to its state before the backup.
If you want to make a snapshot of your database so that you can exactly
recreate its current state at some later time, we recommend you use one of the
following two procedures, which are carried out while the database is stopped:
1. Back up all the tablespaces, the online redolog files and the control file
directly, using NetWorker (i.e. without using NSR-ORA). You will need to
know all the necessary pathnames in order to do this.
2. Then carry out the backup using NSR-ORA and the save set
/nsr/oracle/$ORACLE_SID/${ORACLE_SID}_full_save. This ensures that the
control file and redolog files are only backed up after the tablespaces, so that
the time-stamps of the control file will be as expected by ORACLE.
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
3.2.2
The Backup Component
Online Backup of the Database
An online backup of the database is performed whenever the database is open
at the time of the backup. In contrast to offline backups, the following phases are
carried out:
1. All database files allocated to a tablespace are determined.
2. Backup is commenced with the ORACLE-specific command
alter tablespace <tablespacename> begin backup.
3. All database files of this tablespace are backed up, irrespective of whether
they are system files or raw devices.
4. Completion of the backup is indicated to the database system with the
command alter tablespace <tablespacename> end backup.
3.2.3
Configuration and management
The configuration file dbo${ORACLE_SID}.init, which must be located in the
directory ${DBO_CONFIG_DIRECTORY}/config, can be used to determine certain characteristics of the backup component. The following representation
shows the parameters which are relevant to the backup component and located
in the dbo${ORACLE_SID}.init configuration file (a model file named dbo.init is
stored in the directory $NSR_INST/oracle/config). This configuration file is created automatically using the program configure_nsrora (see also the chapter
“Installation and Configuration”).
The backup_TS_on_different_clients configuration file allows to split a backup
among different NetWorker servers; when using ORACLE Parallel Server
(OPS) also among different NetWorker clients.
To enable this feature, the file backup_TS_on_different_clients must be created in
the /nsr/oracle/$ORACLE_SID/config direcory. If the file does not exist, NSR-ORA
behaves as usual. You can find an example of such a configuration file in the
/opt/nsr/oracle/config directory (for HP-UX: /opt/networker/oracle/config). The
entries also are described there in detail.
!
If you split a backup, Recover until Time is not possible any more.
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
3.2.4
The Backup Component
Checking backups (troubleshooting)
If an error occurs when backing up the database or if one of the backup processes does not end cleanly, the entire database backup is aborted and all
tablespaces are reset to the status NoSave.
You can check whether a backup has been completed successfully by
examining the protocol files (*.prot) which are located in the directory
/nrs/oracle/${ORACLE_SID}/prot and its subdirectories.
Log files from local backups (not for NAS filers)
This file directory contains two protocol files:
●
In the file save_history you will find an overview of the backups that have been
carried out. For each backup of a tablespace or redolog file, it contains a
record of the date and time, the name of the backed up tablespace or
redolog, whether the backup was online or offline and the result of the
backup. You should examine this file regularly to check that backing up the
redolog files is being performed correctly.
●
The file actual_saves_of_tablespaces contains the output of the shell script
saveinfo.sh from the most recent tablespace backup (see below). When you
evaluate this file, note the date of the backup. It always shows the most
recently completed backup, even if this was some time ago.
!
If there are error messages in this file (or in the output from saveinfo.sh)
you should first check the individual protocol files, which are described
below; this will often give you more information about the encountered
error.
In the NetWorker windows you will find information about any problem
which NetWorker encountered while backing up the tablespaces or
redolog files, e.g. problem regarding access to tapes or devices.
If you encounter an error that is related to ORACLE you should also
check your database.
The subdirectory tablespaces contains a protocol file for each backed up
tablespace. These files have names of the form
save_${ORACLE_SID}_${TS}.prot, where ${TS} stands for the name of the
tablespace. There is also the file save_${ORACLE_SID}_overview.prot, which
contains any information regarding the backups that does not relate to a
particular tablespace. The protocol files in this subdirectory are overwritten by
each new backup.
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
The Backup Component
The shell script saveinfo.sh in the directory $NSR_INST/oracle/bin evaluates the
protocol files save*.prot in the subdirectory tablespaces and tells you the status,
the start and end of a backup, the type of backup (online|offline) and the names
and sizes of the backed up tablespaces.
You can call saveinfo.sh at any time, even while there is a backup in progress. It
is called automatically when a tablespace backup has been completed and
writes its output to the file actual_saves_of_tablespaces. This information is also
sent to you by mail.
!
Please note that only the above-mentioned protocol files and the
saveinfo.sh tool provide you with complete information on the status and
success of a backup process.
AMON Recordings
The file /nrs/oracle/${ORACLE_SID}/prot/save_logfiles contains a log of all redolog
files backed up by the archiving monitor (see the section “Logging”). The subdirectory /nrs/oracle/${ORACLE_SID}/prot/dbg contains debug information, which
can be evaluated by Fujitsu Siemens Computers if errors are encountered.
If you want to observe a backup in progress on the screen, you can do so using
the command nsrwatch or from the main NetWorker window.
Please note that only the above-mentioned protocol files and the saveinfo.sh tool
provide you with complete information on the status and success of a backup
process.
If you call ora_recut, additional debugging information, which can be evaluated
by Fujitsu Siemens Computers specialists if an error occurs, is entered in the
directory /nsr/${ORACLE_SID}/tmp/RUT.
!
The global consistency of the backup schedules must always be checked
to ensure that all tablespaces occur in at least one backup cycle. If a
tablespace does not occur in at least one partial backup interval, it is
never saved! Consequently, a configuration tool has been supplied (see
the section “Establishing tablespace names”) for supporting this task.
At least one backup process is started for each backup if the attribute
Backup Command was set to nsrora_save_cmd for the relevant client in the
Clients window. Otherwise, at least two backup processes are started
for each backup - one by NetWorker and one by NSR-ORA. Please keep
this in mind when setting the concurrency of the NetWorker server. The
System Administrator’s Guide
U42252-J-Z915-1-76
The individual Components
The Backup Component
concurrency used by NSR-ORA when backing up tablespaces is one
less than the concurrency of NetWorker when nsrora_save_cmd is not
used, because NSR-ORA always uses one backup for internal purposes.
!
The database must not be started up while an offline backup is in
progress. Under ORACLE Parallel Server (OPS), the instance which is
carrying out the backup is given exclusive access to the database; this
also prevents the database from being started up by any other instance.
In other cases you will receive an error message at the end of the backup
if the database was started up in the meantime.
If you want to back up a database which is currently online under OPS,
while the instance that is to be used for the backup is offline, that instance
will be started and an online backup will be carried out instead.
System Administrator’s Guide
U42252-J-Z915-1-76
4
Installation and Configuration
The following chapter describes the installation and configuration of NSR-ORA.
4.1
Quick Start
This section gives experienced users an short overview of all steps required to
install and configure NSR-ORA. It is a synopsis of the information presented in
this chapter and the rest of the manual and is intended to help you to get NSRORA “up and running” quickly and easily.
In order to install NSR-ORA and prepare it for use you need to carry out the
following steps:
1. Install NSR-ORA on the database computer (Enabler and NSR-ORA package) using the relevant command for the respective platform. Refer to the
sections “Licensing NSR-ORA” and “Installing NSR-ORA from CD”.
2. Now start the database (with startup or, in the case of OPS, with startup
parallel) so that configure_nsrora can determine the tablespace names.
3. Call up the configure_networker tool and fill in the informations described in
section “Parameters for configure_nsrora”.
4. Make sure that you have NSR administrator rights.
5. Call up the configure_networker tool (refer also to the section “Configuring
NetWorker”).
When using a NetApp filer in order to perform NDMP-backups, you may
have to define some NetWorker ressources manually (see “Configuring NetWorker”).
6. Specify the Start time for the database group in NetWorker and set the
attribute Autostart to Enabled (for further information, please refer to your
NetWorker documentation).
7. Highlight the required data carriers for the database and the archiving
monitor amon (see the manual NetWorker - Installing, Configuring and
Operating).
8. Make sure that the database is in the archive mode using the command
archive log list.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Installation
9. Start the archiving monitor amon using the following command:
start_amon.sh start $ORACLE_SID
(refer also to section “Starting and Ending the Archiving Monitor“)
10. Check the results of your first backup (see “Checking backups (troubleshooting)” for this).
4.2
Installation
NSR-ORA is supplied together with NetWorker on CD-ROM. The installation is
carried out (analogous to the installation of NetWorker) on the NetWorker client
on which the ORACLE database is located.
In addition to NSR-ORA, you also have to install an enabler for NSR-ORA itself
on the NetWorker server.
4.2.1
Licensing NSR-ORA
Before you can use NSR-ORA, you first have to install the enabler for NSR-ORA
on the NetWorker server. This is how you install the enabler:
For NetWorker servers from Fujitsu Siemens Computers from NetWorker
Version V5.5A21
1. Insert the license diskette into the drive of your NetWorker server.
2. Log in as root.
3. Enter the following command:
# keylic
The installation then continues interactively.
For different NetWorker servers (prior to Version V5.5A21 or third party)
1. Insert the license diskette into the drive of your Database server.
2. Log in as root.
3. Enter the following command:
●
For NetWorker clients from Fujitsu Siemens Computers (Solaris or
LINUX):
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Installation
# /opt/nsr/keylic -s <server>
<server> is the name of the NetWorker server.
●
For NetWorker servers from Fujitsu Siemens Computers prior to
NetWorker Version V5.5A21 and Reliant UNIX clients or third-party clients for Solaris:
# /opt/nsr/oracle/bin/keylic -s <server>
<server> is the name of the NetWorker server.
●
For clients under HP-UX:
# doscp /dev/floppy/<devicename>:/NSR-ORA <file>
# /opt/networker/oracle/bin/keylic -f <file> -s <server>
<file> is any file name and <server> is the name of the NetWorker
server.
Example (for a NetWorker server backup01):
# doscp /dev/floppy/c0t1d0:/NSR-ORA nsrorakey
# keylic -f nsrorakey -s backup01
The installation then continues interactively.
i
In order to backup a database on a local volume using NSR-ORA the key
diskette NSR-ORA V3.2 (NetWorker license NetWorker Module for Oracle,
UNIX Client) is required.
For using the NDMP functionality additionally the new license NetWorker
Oracle Module for NDMP is required. For this the key diskette NSRONDM
is to be installed.
Both of the licenses may be installed on the NetWorker server by using
the keylic command.
NetApp Licences
On the NetApp filer the snaprestore license is to be installed. To check this you
can start the NetApp command license using a telnet connection. If this license
is not available, please contact your NetApp filer administrator.
4.2.2
Installing NSR-ORA from CD
This section describes how to install NSR-ORA from CD for the different platforms.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Installation
Proceed as follows for installations under Solaris:
Ê If you are performing the NetWorker installation on the console, the installation program opens automatically when you insert the CD. If you are not
working on the console, work through the following steps to start the installation:
Ê Insert the NetWorker CD into the CD-ROM drive.
Solaris then automatically executes the mount operation.
Ê Log in as root.
Ê Set the DISPLAY variable:
# DISPLAY=<myhost>:0
Ê Enter the following command to change to the NetWorker directory on
the CD:
# cd /cdrom/<CD name>
Ê Rufen Sie folgendes Kommando auf:
# ./installer
Ê Invoke the following command:
# ./installer
Ê Choose the SMAWnwora package.
Use the following command to uninstall NSR-ORA under Solaris:
# pkgrm SMAWnwora
LINUX
Proceed as follows for installations under LINUX:
Ê If you are performing the NetWorker installation on the console, the installation program opens automatically when you insert the CD.
Ê If you are not working on the console, or if your desktop does not support
automatic installation, work through the following steps to start the installation:
Ê Insert the NetWorker CD into the CD-ROM drive.
Linux then automatically executes the mount operation.
Ê Log in as root.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Installation
Ê Set the DISPLAY variable:
# DISPLAY=<myhost>:0
Ê Enter the following command to change to the product directory on the
CD:
# cd /mnt/cdrom
Ê Invoke the following command:
# ./autorun
Ê Choose the nsrora package.
The installation is performed with rpm (see rpm(8)). The software to be
installed is packaged in RPMs. These are installed automatically with the following command:
# rpm -i nsrora
The installation directory cannot be selected. All programs are installed
under /opt/nsr.
Ê The NSR-ORA daemon amon can be started following the installation by
invoking the following script:
RedHat LINUX
# /etc/rc.d/init.d/nsrora_amon start
SuSE LINUX
# /etc/rc.d/nsrora_amon start
The NSR-ORA configuration and the associated protocol files are stored in the
/home/nsr/oracle directory.
Use the following command to uninstall NSR-ORA under LINUX:
# rpm -e nsrora
HP-UX
Proceed as follows for installations under HP-UX:
Ê Mount the CD
# mount -o cdcase <device> /cdrom
Ê Change to the /tmp directory
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
# cd /tmp/nsr-ora
Ê Unpack the NSR-ORA.tar file from the CD:
# tar xvf /cdrom/NSR-ORA.tar
Ê The installation is then continued with the command:
# swinstall -s /tmp/NSRORA_HP_PKG \NSR-ORA
4.3
Configuration
4.3.1
How to Proceed for an Update from an Earlier
Version
If you had already installed Version 2.0 or higher of NSR-ORA before you install
V3.2, then the following sections on the configuration do not necessarily apply
in your case. You can take over the configuration of the older version for the
greater part and only need to make a few minor adjustments:
●
There are a few new variables in the dbo${ORACLE_SID}.init file that need to
be allocated. Simply call the configure_nsrora tool once for each database (for
each ORACLE_SID). The tool proposes the values already set in the case of
each variable that has already been defined and the default value for each
new one.
●
Change all clients containing the save set ${ORACLE_SID}_env in such a way
that the save set is no longer contained. If a client consists solely of this save
set, then you can delete it completely.
●
If you are using NSR-ORA V3.2, you have to enter the following backup
command in NetWorker’s Clients window:
nsrora_save_cmd
!
The save set ${ORACLE_SID}_recover must be removed from all existing
configurations.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
4.3.2
Configuration
Preparations
First make sure of the following:
If you have several database instances on one computer, the redolog
files of the various instances must have been created each in their
own separate directory.
!
As the next step you should draw up a backup schedule for your database. This
is a task for the database administrator. You will have to decide whether you
want the database being backed up as a whole or the backup being devided up
to several days of the week.
!
Please note that its is not possible to perform a recover until time if the
backup is divided up to several days.
However if you make a distributed backup, you must decide which
tablespaces should be saved on which day of the week. For example,
tablespaces in which extensive changes take place daily should be
backed up more frequently than tablespaces, which are almost always
accessed in read mode. In this way a faster recovery is possible in an
emergency. Tablespaces that are almost always accessed in read mode
can be backed up less frequently. A suitable backup schedule can allow
you to avoid full backups, which require lots of time and equipment, without creating excessively long recovery times.
Establishing tablespace names
To provide an overview of the existing tablespaces and their sizes, a tool named
config.sh has been supplied as a basis for creating backup schedules. It provides
an overview of existing tablespaces, the associated database files and their
sizes. config.sh is located in the directory $NSR_INST/oracle/config.
Example:
The configuration tool is invoked with the command
sh> config.sh <ORACLE_SID>
The ORACLE_SID can be omitted if it has been set as a shell variable and
exported.
The following information is output, for example:
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Tablespace
---------SYSTEM
File
-------------------------/oracle/7_0_16/dbs/system1.dbf
/oracle/7_0_16/dbs/system2.dbf
**********
sum
Configuration
Size
in [K]
in [M]
8,192
2,048
------10,240
8
2
----10
TEMP_TS
**********
sum
/oracle/7_0_16/dbs/temp1.dbf
4,096
------4.096
4
----4
ROLLBACK_TS
**********
sum
/oracle/7_0_16/dbs/rb1.dbf
8,192
------8,192
8
----8
USER1_TS
/oracle/7_0_16/dbs/system11.dbf
/oracle/7_0_16/dbs/system12.dbf
4,096
4.096
------8,192
4
4
----8
/oracle/7_0_16/dbs/system21.dbf
8,192
------8,192
8
----8
**********
sum
USER2_TS
**********
sum
----38
sum
This information can serve as a basis for creating a backup schedule.
i
If no regular full backups are carried out (${ORACLE_SID}_full_save) then
it is mandatory that the global consistency of the backup schedules is
checked. This is the only way to ensure that all tablespaces occur in at
least one backup cycle during the subsequent setup of the backup directories and their invocation within NetWorker (see section “Configuration
and management“). If a tablespace does not occur in at least one partial
backup interval, it is never saved!
Checking the ORACLE Administration Program
NSR-ORA communicates with the ORACLE database via the ORACLE administrator program sqldba, svrmgrl or sqlplus depending on your version of ORACLE: You will not be able to operate NSR-ORA unless sqldba, svrmgrl or sqlplus
as appropriate, is present and is executable. Please also check the file
${ORACLE_HOME}/rdbms/admin/sqldba.sql (where ORACLE_HOME is the instal-
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
lation directory). This file contains SQL commands, which are executed each
time sqldba or svrmgrl is called. There should be no SQL commands here which
could lead to error messages of the form (ORA-xxxx) when sqldba or svrmgrl is
called: NSR-ORA would notice these error messages when sqldba or svrmgrl is
called and might draw false conclusions from them about the state of the database.
Please note that with ORACLE Version 9.x the svrmgrl command is no longer
supported. Instead of this NSR-ORA uses the sqldba command by default.
4.3.3
Configuring NSR-ORA
The following steps refer to the NetWorker client on which the ORACLE
database and NSR-ORA are installed. For the first-time configuration of NSRORA or when changing to a more recent version, you can use the interactive
tool configure_nsrora, which is located in the directory $NSR_INST/oracle/config. If
you have several databases (several ORACLE instances) on your computer,
you will have to call configure_nsrora once for each database (for each
ORACLE_SID).
If you call configure_nsrora after changing to a different version, it will adopt the
settings from the corresponding old files dbo${ORACLE_SID}.init as defaults.
The tablespaces recorded in the directories monday to sunday will also be taken
over and displayed, and the parameters given in the file
shell_variables_for_${ORACLE_SID} will be used for the new version.
The tool first asks you for the name of the ORACLE instance (ORACLE_SID), the
owner of the ORACLE software (ORACLE_OWNER), the owner’s group
(ORACLE_GROUP), the path to the ORACLE software (ORACLE_HOME). If
these variables are already defined as shell variables and exported, they will be
offered to the user as the default setting. Then configure_nsrora creates the
configuration directory /nsr/oracle/${ORACLE_SID} as well as the directory
/nsr/oracle/etc. If the directory /nsr does not exist on a NetWorker client (this
directory always exists on NetWorker servers) it will be created during installation. But you can also create /nsr yourself before the installation (e.g. if you
want to use links).
The file all_ORACLE_SIDs is created in the directory /nsr/oracle/etc and the
ORACLE_SID is saved there. This file will be accessed if the archiving monitor
daemons are automatically started up again following a boot of your NetWorker
client for all ORACLE instances (i.e. for all ORACLE_SIDs).
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
In the configuration directory /nsr/oracle/${ORACLE_SID} on request a directory
is created for each weekday (monday, tuesday,...), and the directory
${ORACLE_SID}_full_save is created for full backups. If the database is running,
all tablespaces that are currently active are entered in the
${ORACLE_SID}_full_save file. In addition, directories are created for temporary
files (./tmp), protocol files (./prot) and for the NSR-ORA configuration file
(./config).
In the directories named after weekdays, the names of the tablespaces that you
want to store on these days are stored interactively in the form of empty files.
You can change or add to your backup schedule at any time after configuration
with the tool configure_nsrora, e.g. by deleting tablespaces (i.e. removing the
empty tablespace files with the shell command rm) or adding them (i.e. creating
new tablespace files with the touch command). If you add new tablespace files,
note that the owner of these files must be $ORACLE_OWNER and their group
must be $ORACLE_GROUP.
Finally the NSR-ORA configuration file dbo${ORACLE_SID}.init is created interactively and entered in the directory /nsr/oracle/${ORACLE_SID}/config. You will
be asked for several parameters which were described in section “Parameters
for configure_nsrora” (NetWorker server, backup compression, minimum and
maximum number of offline redolog files, time interval for the archiving monitor
daemon, mirror server for the archiving monitor, and so on). Afterwards you can
still edit the file dbo${ORACLE_SID}.init, if for example you do not like the
suggested names for the NetWorker groups.
Furthermore the parameters NSR_INST, ORACLE_HOME and ORACLE_GROUP
are stored in the file shell_variables_for_${ORACLE_SID} in the config directory.
4.3.4
Parameters for configure_nsrora
When you call the configure_nsrora configuration tool NSR-ORA will ask you to
set up the following parameters (depending on the existing configuration the
order may vary).
Oracle Parameters
In a first instance the parameters required for the ORACLE database are to be
set:
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
ORACLE_SID
Name of the ORACLE entity. If several ORACLE databases are installed on
the same computer, NSR-ORA can distinguish them clearly by means of the
ORACLE_SID.
ORACLE_OWNER
ORACLE owner.
ORACLE_GROUP
ORACLE owner group.
ORACLE_HOME
Directory where ORACLE is installed in.
NetWorker Parameters
NSR_INST=/opt/nsr
NetWorker installation directory. It is recommended to include the paths
${NSR_INST}/oracle/bin and ${NSR_INST}/oracle/config in the shell variable
PATH; these paths contain the NSR-ORA software.
Subsequently you will be asked, whether you want the backups to be splitted
onto several days of the week. If you enter y, you kann define a splitting as
described in section “Defining backup sets (save sets)”.
!
If a backup session is splitted, a Recover until Time is no more possible.
Now the tool will ask you, whether you want to perform offline backups, too. Answering y a corresponding client will be creted as described in section “Offline
Backup of the Database”.
NSR_SERVER
Here enter the NetWorker server.
Thereafter configure_nsrora asks you whether you are using a NetWorker cluster
server. When you enter y you will be asked to define an additional parameter.
CONTR_SERVER
If the backups under NetWorker are to be carried out via a NetWorker server
cluster you use this parameter to specify the control server of the cluster.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
The next query depends on the usage of an NAS filer. When you are using such
an NAS filer and so answer y to this question you have to input the information
described in section “Configuring NSR-ORA”.
BACKUP_COMPRESS=true | false | gzip | bzip2
Compression of the database files is controlled via the BACKUP_COMPRESS
parameter. It may take the following values:
true
Selecting true for BACKUP_COMPRESS causes the default compression
mode (Lempel-Ziv) to be used.
false
No compression is used.
gzip
The GZIP mode is used for compression. In addition a compression level
may be selected by setting the (hidden) COMPRESS_LEVEL parameter
(see below).
bzip2
The BZIP2 mode is used for compression. In addition a compression
level may be selected by setting the (hidden) COMPRESS_LEVEL parameter (see below).
Now configure_nsrora asks you whether you want to perofrm parallel backups of
the tablespaces. Answering y an additional parameter has to be set.
AMON Parameters
Now the amon parameters are queried:
AMON_MAX_LOGS
This parameter defines a threshold value for the maximum number of offline
redolog files in the archive directory. When this value is reached, the next
poll (see AMON_INTERVALL parameter below) will initiate a backup of these
redolog files. When the offline redolog files have been backed up successfully on external media they are deleted from the archive directory.
AMON_MIN_LOGS
You use this parameter to specify how many offline redolog files should not
be backed up but should remain in the archive directory when the threshold
value AMON_MAX_LOGS is reached. It is always the newest redolog files
that remain in the archive directory without being backed up.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
This makes it possible to carry out a recovery more rapidly, or to carry out a
point-in-time recovery without having to read in the redolog files, since the
existing redolog files can be used.
AMON_INTERVALL
The parameter AMON_INTERVALL determines the time interval (in seconds)
at which the archiving monitor is induced to compare the current number of
accrued offline redolog files with the value of AMON_MAX_LOGS.
MIRROR
With this parameter you define whether the offline redolog files will be
backed up twice by using mirroring. The following parameter is to be set only
if your answer is y(es).
AMON_MIRROR_SERVER
Here enter the name of the NetWorker server that should control the backup
of the optional second NetWorker backup group. Additionally with this parameter a name is defined for the corresponding backup group. You may
change this name later on, if necessary (see “Parameters with Expert
Mode”).
If there is no entry for AMON_MIRROR_GROUP and
AMON_MIRROR_SERVER in the dbo${ORACLE_SID}.init configuration file no
mirrored backup will be performed.
For the mirrored backup being activated by using AMON_MIRROR_GROUP
und AMON_MIRROR_SERVER the backed up redolog files will only be deleted if both redolog backups have benn completed successfully.
SEND_MAIL_TO
You can use this parameter to specify a mail address to which the list of
backed up redologs/tablespaces should be sent.
AMON_WATCH_DELAY
Period of time (in seconds) after which AMON_WATCH_DOG_CMD (see next
parameter) should be called if the backup or mirror backup of the redolog
files has not been performed.
AMON_WATCH_DOG_CMD
Command to be executed if the backup or mirror backup of redolog files fails
or if the backup was not ended after AMON_WATCH_DELAY seconds.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
OPS Parameters
OPS=yes|no
You use this parameter to specify whether or not NSR-ORA will be used
under Oracle Parallel Server (OPS).
OPS_INSTANCES
If you are using NSR-ORA under OPS, this parameter contains a list of
the names of all computers on which an OPS instance is running (e.g.:
psi_1,psi_2).
Recovery Parameters
Finally, two parameters concerning the recovery are to be set:
NSR_MAX_RECOVER
This parameter specifies the maximum number of parallel recoveries of
tablespace files. The default value is 10, i.e. a maximum of 10 recovery
processes is started in parallel. For devices where the data block
between two file marks is larger (e.g. 3590 devices), the value should be
larger to ensure an optimum recovery time. If the tape is spooled back
and forth several times during a recovery this parameter should be
increased. This parameter is only of interest for the Fujitsu Siemens
Computers NetWorker Version up to V5.5 (including).
NSR_MAX_REC_DEV
This specified the number of backup devices to be used for a recovery.
The value set here should correspond to the number of devices used in
a recovery.
Parameters with Expert Mode
Some parameters are not asked for by configure_nsrora but may be changed
manually in the dbo${ORACLE_SID}.init configuration file after the skript’s run
has completed. In General this is not necessary, but to be complete these parameters are described below. They serve, first of all, to change default values
(e.g. names of ressources created by configure_nsrora) if this is explicitly intended.
AMON_GROUP
Name of the NetWorker backup group for the archiving monitor.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
AMON_MIRROR_GROUP
Name of an optional second NetWorker backup group for the archiving
monitor. If this mirror backup is to take place on a different NetWorker server,
the name of that server must also be specified under
AMON_MIRROR_SERVER. This database server must also be known to the
NetWorker server as a client.
If no entry is made for AMON_MIRROR_GROUP and
AMON_MIRROR_SERVER in the configuration file dbo${ORACLE_SID}.init, no
mirror backup is performed.
If mirror backups have been activated, by specifying
AMON_MIRROR_GROUP and AMON_MIRROR_SERVER, the backed up
redolog files will be deleted only when both of the redolog backups have
been successful.
AMON_SAVE_ARG_NR
This variable indicates how many redolog files should be combined in a save
set. It was introduced to speed up the cycle between backup and mirror
backup and to enable the associated backed up redolog files to be deleted.
The default value is 50.
AMON_RM_WITHOUT_MIRROR_DELAY
If this variable has been set to a value, redolog files that have been
backed up but not mirrored, or vice versa, are still deleted when this time
(in seconds) expires. This operation is logged in the Amon log file. This
prevents a situation where log files have been backed up but the database waits because there is no more space in the redolog area. The
default value is 3600.
BACKUP_GROUP
The BACKUP_GROUP parameter allows the allocation of a backup to a
NetWorker backup group (and thus to defined backup media, time
schedules etc.). You only have to set up this parameter if you do not want
the default values to be used.
COMPRESS_LEVEL
Using this parameter you can configure the compression factor of the
respective algorithm. You get the possible values by viewing the manpage uasm(1).
For this parameter not being set NSR-ORA uses the following values:
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
●
GZIP is used with the default value 1.
●
BZIP2 no level is given to, so the default value of the algorithm will be
used (see the uasm(1) manpage for further information).
4.3.5
Defining backup sets (save sets)
The NSR-ORA administrator is able to specify save sets (offline backups, full
backups, partial backups) by defining corresponding sub-directories in the
directory $DBO_CONFIG_DIRECTORY. The path to these sub-directories is
entered as Save Set in NetWorker’s client windows. Furthermore, empty files
must be created in the sub-directories (e.g. with touch) to represent the names
of tablespaces requiring backup.
The configuration tool config.sh can be used to get an overview of the available
tablespaces and their space requirements (see section “Establishing tablespace names”).
Example
Suppose the configuration directory (DBO_CONFIG_DIRECTORY) is
/nsr/oracle/ora. The database consists of the tablespaces SYSTEM,
ROLLBACK_TS, TEMP_TS, USER1_TS and USER2_TS. The objective is to perform
a full backup of the database once a week every Saturday and partial backups
of it daily (except Sunday). Furthermore, the frequently modified tablespace
USER1_TS is to be backed up daily to ensure quick restarting in case of a failure.
Automatic offline backups should also be performed occasionally.
i
The backup concept described herein (backup of tablespaces in a
weekly cycle) is not mandatory. You may also develop different backup
concepts.
The NSR-ORA administrator creates the directories monday to friday and
${ORACLE_SID}_full_save for full and ${ORACLE_SID}_offline for offline backups,
under /nsr/oracle/ora, which contain the following empty files:
/nsr/oracle/ora/monday:
SYSTEM
USER1_TS
/nsr/oracle/ora/tuesday:
TEMP_TS
USER1_TS
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
/nsr/oracle/ora/wednesday:
USER1_TS
USER2_TS
/nsr/oracle/ora/thursday:
ROLLBACK_TS
USER1_TS
/nsr/oracle/ora/friday:
USER1_TS
/nsr/oracle/ora/${ORACLE_SID}_full_save:
SYSTEM
/nsr/oracle/ora/${ORACLE_SID}_offline:
SYSTEM
i
Note that only one file (in our example: SYSTEM) is needed in the directories ${ORACLE_SID}_full_save and ${ORACLE_SID}_offline for a
tablespace that is to be backed up. If at least one of the existing
tablespaces is specified here, that tablespace along with all other existing tablespaces and the control file will be backed up automatically.
i
It is also recommended generally to carry out a full backup
(${ORACLE_SID}_full_save or ${ORACLE_SID}_offline) at least once a
week. This ensures that all tablespaces are backed up at least once a
week and at the same time all control files (and in an offline backup
additionally all redolog files) will be backed up as well.
!
For the automatic recovery of the database to the state of a previous
point in time (see section “Automatic Recovery for an Optional Point in
Time”) backups with „full“ level and redolog files are explicitely required.
Individually saved tablespaces are not used therefore.
!
The <ORACLE_SID>_recover save set needed with earlier versions must
not be used any more.
If you are using NSR-ORA for the first time, it is advisable and convenient to use
the configuration tool configure_nsrora described in section “Configuring NSRORA”. This tool creates a directory for every weekday (monday, tuesday, ..) and
prompts you for the names of the tablespaces to be backed up.
It is important that the owner of the (empty) tablespace files be
$ORACLE_OWNER and for them to belong to the same group as
$ORACLE_OWNER (usually dba).
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
This configuration helps fulfill the requirements for the NSR-ORA administrator
to define a backup schedule within NetWorker. The steps necessary for this are
explained in the following section.
4.3.6
Configuring NetWorker
To simplify the NetWorker configuration process, the tool configure_networker is
provided for the first-time configuration. It is located in the directory
$NSR_INST/oracle/config. Note that this tool requires the file
dbo${ORACLE_SID}.init. The settings made by configure_networker correspond
to those in section “Defining backup sets (save sets)” (tablespace backup at
weekly intervals). If you want to split up the backup of the tablespaces the
default is to create one client for each day of the week and an additional client
for the full backup. However, the client for the full backup is then not active by
default.
If you simply wish to carry out full backups then only the client for full backups
is created (${ORACLE_SID}_full_save). Alternatively, a client for offline backups
can also be added here (${ORACLE_SID}_offline).
The tool is called with
# configure_networker <create|delete> <ORACLE_SID>
You can erase a first-time configuration with the delete option (changes made
after the first-time configuration with configure_networker are not erased).
While the tool is working you are prompted to enter the following information:
●
the name of the NetWorker client on which NSR-ORA and the ORACLE
database are installed
●
the name of the NetWorker server by means of which the database is to be
backed up
●
the way in which the NetWorker server is to be configured. It depends on this
which resources have to be defined on the NetWorker server:
– for backup only (backup; normal use)
– for mirroring only (mirror; configuration of the second NetWorker server
for a dual backup of the redolog files)
– for both (both; normal use and mirroring)
Please check the settings that configure_networker made and compare them with
the entries described in section “Parameters for configure_nsrora”.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
Configuration
You will find an example of the progress of configure_networker below:
root 68 # configure_networker create ORA2
*-----------------------------------------------------------------------------*
|
|
| This script helps you to configure your Networker server(s) so you are
|
| able
|
| - to save and restore your database and to run the archive monitor,
|
| - to run a mirror archive monitor on another Networker server.
|
|
|
| This script will use the NSR-ORA configuration file, which was created
|
| by "configure_nsrora". If your database and NSR-ORA are not installed
|
| on this machine, you will be asked to copy the configuration file to /tmp. |
|
|
| The configuration of the Networker server(s) is also described in detail
|
| in the manual.
|
|
|
*-----------------------------------------------------------------------------*
ORACLE_SID=ORA2
Please enter the name of the Networker client, where NSR-ORA and
the ORACLE database are installed: [nero]
nero
nero is alive
Please enter the name of the Networker SERVER which should save the database:
nero
nero is alive
The NetWorker Server nero will be configured as:
- a database backup server with archive monitor
(please type "backup"),
- only as a mirror server of your archive monitor (no database backup)
(please type "mirror"),
- as a database backup server and as a (mirror) server of the archive monitor
(please type "both").
both
NetWorker server nero will be used for all NetWorker/NSR-ORA activities.
You are working on the Networker client nero.
The configuration file /nsr/oracle/ORA2/config/dboORA2.init
will be used as an input for the configuration of
the Networker Server nero.
The configuration file still exists.
The following parameters have been read from
/nsr/oracle/ORA2/config/dboORA2.init :
ORACLE_SID
DBO_CONFIG_DIRECTORY
NSR_SERVER
BACKUP_GROUP
OBSERVE_NAME
:
:
:
:
:
ORA2
/nsr/oracle/ORA2
nero
backup_ORA2_nero
AMON_GROUP
AMON_MIRROR_GROUP
AMON_MIRROR_SERVER
: amon_ORA2_nero
: amon2_ORA2_nero
: nero
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
NSR-ORA in OPS clusters
They will be used to configure your NetWorker Server.
Please press ENTER to continue
created
created
created
created
created
created
created
created
created
resource
resource
resource
resource
resource
resource
resource
resource
resource
id
id
id
id
id
id
id
id
id
0.20.3.237.51.225.160.62.77.0.0.4(1)
0.21.3.237.51.225.160.62.77.0.0.4(1)
0.22.3.237.51.225.160.62.77.0.0.4(1)
0.23.3.237.51.225.160.62.77.0.0.4(1)
0.24.3.237.51.225.160.62.77.0.0.4(1)
0.25.3.237.51.225.160.62.77.0.0.4(1)
0.26.3.237.51.225.160.62.77.0.0.4(1)
0.27.3.237.51.225.160.62.77.0.0.4(1)
0.28.3.237.51.225.160.62.77.0.0.4(1)
Please NOTE:
- The Group backup_ORA2_nero is disabled now. Please enable it and
choose the start time.
created resource id 0.29.3.237.51.225.160.62.77.0.0.4(1)
created resource id 0.30.3.237.51.225.160.62.77.0.0.4(1)
created resource id 0.31.3.237.51.225.160.62.77.0.0.4(1)
4.4
NSR-ORA in OPS clusters
This section describes points to observe during installation and configuration of
NSR-ORA in an OPS cluster (ORACLE Parallel Server cluster).
4.4.1
Installation
Installing NSR-ORA on all the OPS nodes as described in the
section “Installation“.
For OPS systems the archiving monitor amon is used on every OPS
node, so we recommend to use the same pathname for the redolog files
on every node.
i
4.4.2
Configuration
Special attention must be paid to the following factors when installing NSR-ORA
in OPS systems:
●
NSR-ORA is configured from just one of the OPS nodes. The following
commands have to be executed one after the other on that node
# configure_nsrora
...
# configure_networker
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
NSR-ORA in OPS clusters
...
Please also refer to the descriptions in the sections “Configuring NSR-ORA”
und “Configuring NetWorker”.
●
All OPS nodes have to be entered as clients on the OPS node on which
NSR-ORA is being configured. The clients do not have to belong to any
group, they just need to have been configured.
●
The archiving monitor (amon) has to be present on each OPS node. Use the
tool configure_nsrora for the configuration, however, the parameters applying
to tablespace backup are irrelevant here. It is best to configure the monitors
manually in NetWorker. Do not forget to start the monitors using the following
command:
/opt/nsr/oracle/bin/start_amon.sh
(on HP-UX: /opt/networker/oracle/bin/start_amon.sh).
●
If the tablespaces are to be backed up for several NetWorker servers using
the file backup_TS_on_different_clients there has to be a save group (and also
a pool) on each of the NetWorker servers. The name of the save group has
to be identical in all of the pools (the default name is OPS_TS_BACKUP).
●
The OPS_INSTANCES variable contains the names of all the OPS instances
that were active when the configure_nsrora command was called. If further
instances are added later the variable definition in the
dbo${ORACLE_SID}.init file has to be adjusted.
●
OPS does not support backups with the save set OFFLINE.
●
Different OPS instances are supported on the various OPS nodes with the
environment variable OPS_DIFFERENT_SID_NAMES. This variable is not
queried in the configure_nsrora script and may have to be set directly in the
NSR-ORA configuration file dbo{$ORACLE_SID}.init.
System Administrator’s Guide
U42252-J-Z915-1-76
Installation and Configuration
4.5
Troubleshooting
Troubleshooting
Backing Up an Oracle Database Version 8.0 or higher
There is an Oracle Error in Oracle 8.0 to 8.0.5 (Generic BUGID 673508 and
677804) which may lead to an inconsistent backup when performing an online
backup (the redolog files are corrupted).
NSR-ORA on LINUX
You need ksh to use NSR-ORA. This means that the pdksh package must be
installed on the LINUX computer. If the NSR-ORA backup is to be analyzed with
set -x in the scripts, we recommend you use the ksh supplied with NSR-ORA.
pdksh-5.2.14-1 contains an error in this regard, which can cause ksh to crash.
However, this error has already been reported to pdksh development and should
be rectified in later versions. The fix is already integrated in the ksh that has been
supplied.
System Administrator’s Guide
U42252-J-Z915-1-76
5
Recovery
This chapter describes the various ways in which you can carry out a recovery
for an ORACLE database.
We recommend you try out a recovery of your database on a test system.
This will enable you to become familiar with the possible procedures, so
that in a genuine emergency you will be in a position to decide which type
of recovery is appropriate.
i
There are basically three ways of recreating a damaged ORACLE database:
●
firstly, performing a complete fully-automatic recovery of the entire database
with the nsrora_rec command
●
secondly, performing a recovery until time with the ora_recut command for an
optional point in time between the first full_save backup and the point in time
the recent changes within the database have occured.
●
thirdly, carrying out a manual recovery, in which the corrupted data is
restored by hand.
5.1
Selecting the Type of Recovery
Which type of recovery you choose (recover until time, automatic or manual) will
depend entirely on the actual situation. There are advantages and disadvantages to all methods; you will have to weigh these up as appropriate for the
specific circumstances. This section is intended to help you decide whether to
perform an specific type of recovery.
5.1.1
The Recovery Component
The NSR-ORA recovery component allows the automatic recovery of an
ORACLE database saved with NSR-ORA if one or more database file(s) are lost
or an logical error has occurred (see figure 6 following).
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
Selecting the Type of Recovery
Naturally, you can also restore the database “manually”, i.e. without using the
automatic facilities of NSR-ORA.
DBMS
Oracle’s
archiving
process
Database
files
Control
file
SYST.
USER
TEMP
Tablespaces
Threshold
value
Redolog
files
Archive
directory
Database interface NSR-ORA
Recovery
component
Archiving
monitor
NetWorker user interface
Volume management
Volume pool management
Multiplexing
Jukebox support
Tape library
Figure 6: The recovery component
5.1.2
Recovery Until Time with Logical Errors
The recovery until time (with the command ora_recut) allows the database to be
recovered for an optional point in time. Thus it is also possible to automatically
recover logical errors in the database.
Changes made to the structure of the database can also be reversed, since
ora_recut also recovers the saved control file for the relevant point in time.
i
The database must be physically sound (operable) because ora_recut
needs to query the position of the control files (not for NetApp filers).
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
5.1.3
Selecting the Type of Recovery
Automatic Recovery (Crash Recovery)
In the case of a fully-automatic recovery, NSR-ORA deals with the entire task of
restoring the damaged tablespaces and redolog files without any further
operator intervention. It communicates with ORACLE, enquires which
tablespaces are damaged and recreates them. Then it asks for the necessary
redolog files and recreates these, if they have already been backed up. It then
tells ORACLE to perform a roll forward using the redolog files.
Within a recovey process three phases are to be distinguished:
●
Analyzing the Database (Crash Analysis)
During the database is examined an analysis with regard to a loss of database files is performed. Therefore, first of all, NSR-ORA tries to close the
database. If this action fails the database is shut down hard using the
shutdown abort command. Then several informations required for a following
restore and the recovery of the database are generated by analyzing database views.
If this examination shows that no database files were lost the database will
be reset to normal operation mode again already within this phase.
●
Restoring the database files
The restore of the database files will be done using the information generated in the analyzation phase. This phase is unnecessary, if the previous examination has not found any database files corrupted.
●
Recovering the database
The recovery on the database level is performed offline, i.e. the database is
not usable. During this phase all database files are switched to offline mode
to assure, that all the database files are recovered. In addition the archiving
monitor is stopped to prevent it from obstructing the recovery process.
Thereby the tablespaces are recovered parallel whereby the parameters
NSR_MAX_RECOVER and NSR_MAX_DEVICES rule the parallelism and/or
the number of backup devices to be used.
Subsequently all offline redolog files (after restoring) will be redrawn. After a
final check for the recovery activities having been completed successfully
the database is reset to normal operation mode.
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
!
Performing a recovery
An automatic recovery is quite unsuitable if the fault in the database files
is not a physical media failure but a logical error in the data of the
database. The automatic recovery procedure has no way of detecting
logical errors.
An automatic recovery always reproduces the last backed up state prior
to a physical fault. In order to reproduce an earlier condition, e.g. before
the occurrence of a logical error, you can use the command ora_recut
(recovery until time).
!
If the data was distributed over a number of NetWorker servers by the
backup then it cannot be recreated automatically.
Under ORACLE Parallel Server (OPS), the automatic recovery terminates
directly after the tablespaces have been restored. The redolog files that are distributed over the various nodes of an OPS system cannot be recreated automatically, instead the user must ensure that they can be accessed from a single
OPS node and must then recreate them manually. You will find a description of
this in the section “Recovery with ORACLE Parallel Server (OPS)”.
5.1.4
Manual recovery
In all cases in which an automatic recovery is not appropriate, you should carry
out a manual recovery. For manual recovery you will need to know in which
directories your redologs and the files or raw devices of your tablespaces are
located.
5.2
Performing a recovery
The following sections describe how you can perform a recovery of an ORACLE
database. The first section describes automatic recovery of an optional
database level using ora_recut, the second section describes an fully-automatic
recovery of the last saved level of the database, the third section describes a
manual recovery and the fourth one a recovery with OPS.
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
5.2.1
Performing a recovery
Automatic Recovery for an Optional Point in Time
This section documents the use of the ora_recut command with the aid of which
you can automatically recover a specific database level (recovery until time).
In order to recover a specific database level you can use the following
command:
ora_recut [<ORA-SID>] [-t <time>] [-v]
The ORACLE ID <ORA-SID> only needs to be specified if there is more than one
database. If the -t option is used to define a point in time (format yyyy-mmdd:hh:mm:ss,i.e., for instance: 1997-11-27:22:00:00) the ora_recut command
will attempt to roll back the database to precisely that moment. If you do not
enter a time here, then ora_recut will prompt you for a time presently. The -v
option enables the output of additional information.
The following prerequisites must be met so that ora_recut can be used:
●
The database must be physically sound (operable) because ora_recut needs
to query the position of the control files.
●
All the servers and device servers used for the backups have to be active.
●
The last full recovery made before the required point in time using
${ORACLE_SID}_full_save has to have run under Version 3.0 of NSR-ORA or
higher.
If OPS is being used, you must pay attention to some further prerequisites (see
section “Recovery with ORACLE Parallel Server (OPS)” for further information).
!
With the automatic recover until time you cannot use backups which are
distributed on several days of the week.
In order to restore a specific database level with these prerequisites you must
carry out the following steps:
1. Specify the time.
2. Start ora_recut:
# ora_recut [<ORA-ID>] [-t <time>]
If ora_recut requires further information, you will be prompted to enter it.
Basically ora_recut runs fully automatically. Only if errors occur that ora_recut
cannot solve on its own, does an error message appear and you are
informed that you can attempt to solve the problem yourself, for instance:
The database hadn’t gone down in 5 minutes
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
Performing a recovery
Please look for a solution or give up!!
Sqldba>
or
ERROR: couldn’t get log_file <....dbf>
Please try to find it yourself
$
You can enter optional commands after the prompts Sqldba> and $ in order
to recover errors (e.g. to stop a database, rename a file, etc.). When you
have finished, just enter %? to return to ora_recut. You can then decide
whether you want to continue the recovery using ora_recut or whether you
want to abort it.
!
If the recovery is meanwhile aborted, it is not possible to restart it from
that point any more. The database has to be restored first before a
new recovery attempt with ora_recut can be made.
After the recovery is completed ora_recut opens the database using the
following command:
database reset logfiles
!
5.2.2
The database should be backed up as soon as possible after a
successful recovery.
Fully-automatic Recovery
In the event of a physical fault in your database you can carry out a fully-automatic recovery. You can carry out a fully-automatic recovery with the assistance
of the nsrora_rec command.
nsrora_rec [<ORA-SID>]
Under normal circumstances, this file does not require any options. If only one
Oracle database is configured, the recovery is also performed immediately. If
several databases are configured on this machine, the user is asked to choose
a database using the ORACLE_SID. This can also be specified in the call.
# nsrora_rec
INFO:
INFO:
(0).
INFO:
INFO:
*** Invoking Crash Analyzer ***
Trying to shutdown database within 30 seconds ... done
Getting required database files ... done.
Generating miscellenous informations ... done.
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
Performing a recovery
INFO: Following database files are required to perform a
recovery:
/oracle/7_0_16/dbs/benutzer1NSR.dbf
INFO: *** Crash Analyzer finished ***
INFO: *** Invoking Database Restauration ***
INFO: Trying to restore /oracle/7_0_16/dbs/benutzer1NSR.dbf
recover: Using ufo as server for ufo
Recovering 1 file into its original location
Total estimated disk space needed for recovery is 4.1 MB
Requesting 1 file, this may take a while ...
./benutzer1NSR.dbf
Received 1 file(s) from NSR server ’ufo’
INFO: *** Database Restauration finished ***
INFO:
INFO:
INFO:
INFO:
INFO:
INFO:
INFO:
*** Invoking Database Recovery ***
Mounting the database ... was previously done.
Ensure that all database files are online ... done.
Recovering the database.
Database accessible.
Recovery finished successfully!
*** Database Recovery finished ***
The second example deals with a situation in which all database files are intact.
In this case, an appropriate message is displayed after the database has been
analyzed:
# nsrora_rec
INFO: *** Invoking Crash Analyzer ***
INFO: Trying to shutdown database within 30 seconds ... done
(20).
INFO: Getting required database files ... done.
INFO: No damaged files found !
INFO: *** Crash Analyzer finished ***
INFO: *** No Database recovery required (RC: >5<) ***
Finally, the third example demonstrates the behavior of the crash analyzer in a
situation in which an automatic recovery is not possible (e.g. when a control file
is lost). You will have to carry out a manual recovery in such cases:
# nsrora_rec
INFO: *** Invoking Crash Analyzer ***
INFO: Trying to shutdown database within 30 seconds ... done
(0).
INFO: Getting required database files ... failed.
ERROR-310: Unable to mount the database !
ERROR-350: *** Crash Analyzer failed (RC: >20<) ***
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
5.2.3
Performing a recovery
Manual Recovery
This section describes how to completely restore a database (that is, to restore
all the tablespaces) manually (i.e. without using NSR-ORA’s automatic facilities)
and - if necessary - to a state which it had at some earlier time. We shall assume
that you are familiar with both NetWorker and ORACLE.
ORACLE gives you the option of carrying out a recovery either offline or online.
When you perform an offline recovery the entire database has the status Mount;
this means that the database cannot be used while the recovery is in progress.
In the case of an online recovery, only those tablespaces and data files that are
currently being restored are switched offline and you can continue working with
the rest of the data. An offline recovery is generally recommended if the
tablespaces are strongly interdependent (as is the case for e.g. R/3 applications) - in this case the database will have the status Mount throughout the
recovery process. A manual offline recovery is described in the following
section.
Offline recovery of an offline or full backup
A manual offline recovery is only possible if the backup was also created offline
or using the save set ${ORACLE_SID}_full_save (full backup). When you perform
an offline recovery you can proceed as follows:
1. Stop the NSR-ORA archiving monitor:
# /opt/nsr/oracle/bin/start_amon.sh stop $ORACLE_SID
(on HP-UX: /opt/networker/oracle/bin/start_amon.sh stop $ORACLE_SID).
2. Shut down the database:
shutdown
3. Make sure that NetWorker does not start any other backup jobs while the
recovery is in progress, otherwise the recovery jobs and save jobs might
compete for the backup devices and so delay the restoration of the
tablespaces and redologs. For this reason you should set disabled for any
save groups that are about due to start, until the recovery has been
completed.
4. Determine all the save sets belonging to the desired full offline backup (must
be a full backup), for instance as follows:
# mminfo -v -q ´pool=<ora_pool>, savetime>="mm/dd/yy hh:mm:ss",\
savetime<="mm/dd/yy hh:mm:ss"´ -r "name,ssid"
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
Performing a recovery
This command gives you the names and the save set IDs of all the files
saved in <ora_pool> in the period specified.
5. Remove the save set IDs of the following files from this output:
– recover_info_last
– bootstrap
– .../nsr/index/<client>/db
6. Recover the remaining save sets again:
# recover -a -f -S<ssid1> -S<ssid2> ...
7. For the recovery of an online backup only:
In an online backup the control file is not saved with its correct pathname.
So, if you need a copy of the saved control file you will have to copy the
control file to the correct position.
The file name of the saved file is:
/nsr/oracle/${ORACLE_SID}/tmp/cntrl${ORACLE_SID].dbf.back). The name of
the control file in the database can be copied with the SQL command
select name from v$controlfile. In order to copy this file, use the UNIX
command cp for databases in the file system and the UNIX command dd bs=
8k if=... of=...for databases on raw devices.
You must then use the using backup controlfile option with the ORACLE
command recover database ....
8. Reapply the redolog files since the last full backup to the ORACLE archiving
directory, if they have already been saved with NetWorker.
9. Start the ORACLE database using the startup mount command in the mount
state.
10. Recover the database by calling either the recover database, recover database
until time or recover database until cancel command (see ORACLE documentation).
11. Open the database using
alter database open NORESETLOGS;
or
alter database open RESETLOGS;
12. If you set save groups to disabled in step 3., then you can now set them back
to enabled.
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
5.2.4
Performing a recovery
Recovery with ORACLE Parallel Server (OPS)
If OPS is being used, you must pay attention to some further prerequisites:
●
All OPS instances have to have been specified in the configuration of NSRORA. Instances that were added at a later date have to be contained in the
file /nsr/oracle/${ORACLE_SID}/config/dbo${ORACLE_SID}.init and also have
to have been added to the variable OPS_INSTANCES.
●
On all OPS instances, the .rhost file has to contain entries for all other
instances for the ORACLE user.
●
All OPS instances have to have had recover access assigned to them in
the NetWorker client resource for all other OPS instances.
●
The redolog files of the individual instances have to be accessible under a
single directory name. This may be a symbolic reference, which is created
on all of the instances.
●
All OPS instances have to be accessible during the recovery.
In an OPS system, each OPS node has a separate directory for its archived
redolog files; these directories are not sharable.
You have the following options for the recovery:
●
If the database was backed up online, use the automatic recovery (by calling
nsrora_rec). The recovery of the tablespaces then runs as described in the
section “Fully-automatic Recovery”.
!
The redolog files have to be restored manually when using nsrora_rec,
because it is not possible to automate access to the distributed
archived redolog files. In the case of an offline recovery on an OPS
system, the ORACLE administrator must therefore ensure that all the
redolog files can be accessed from the OPS node from which the
database recovery was started.There are various ways of doing this,
for instance the ORACLE administrator can
– read the redolog files that were backed up by NetWorker back into
the archive directories of the individual instances and mount these
directories via NFS,
– read or copy all the necessary redolog files (both those that have
been backed up by NetWorker and those that are still located in
the archive directories of the various instances) into a directory
which is located on the instance from which the recovery will be
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
Troubleshooting
started. You are recommended to use the archive directory of the
”recovery instance“ for this purpose. You can determine which
directory this is by issuing the following command:
show parameter log_archive_dest
In this case you can use the autorecovery option; after you call
recover database ... ORACLE will then automatically process the
redolog files. If you use one or more other directories you will have
to specify the pathname for each redolog file interactively with
ORACLE.
Backups are much simpler if the archive directory has the same path
on all the instances. In this case NetWorker will copy the redolog files
to the correct location straightaway.
●
If the database was backed up offline, then proceed as described in the
section “Manual Recovery”.
●
A recovery until time (using ora_recut) can be carried out both for offline and
online backups and can only be run on the OPS node, which also starts
the database backup.
i
If you note when recovering the tablespaces that you also have to
recreate directories that are in the pathnames of the tablespaces, please
check the owner and group of these directories. NetWorker normally
creates new directories with the owner root, which will cause access
problems in the case of a database directory.
i
Please note that when the recovery has finished, the archiving monitor
(amon) will have to be restarted on each individual (!) node of the OPS
system.
5.3
Troubleshooting
No space left on device xxx
You may possibly receive this error message during a recovery. Apart from the
fact that there might really be insufficient space on the disk, there is also the
possibility that an sqldba-, svrmgrl-, sqlplus- or some other database process is
blocking the required disk space. In this case you only need to terminate the
database processes.
System Administrator’s Guide
U42252-J-Z915-1-76
Recovery
Troubleshooting
log belongs to wrong database
archived log ends at change nnnnn, need later change mmmmm
If you get this or a similar error message, then the reason is generally that you
executed an open database resetlogs command after a recovery until time and then
copied in another database level again. This means that the online redolog files
are no longer numbered consistently. Only a manual recovery can correct this
error.
System Administrator’s Guide
U42252-J-Z915-1-76
6
Using NSR-ORA with
NetApp Filers
This chapter describes how to configure and use NSR-ORA with NetApp Filers
6.1
Running NSR-ORA with NAS devices
NSR-ORA allows database files stored on a Network Attached Storage (NAS)
you to be backed up or recover or recovered using NDMP (Network Data
Management Protocol). The data is written and read directly from the NAS
device to the backup media, so the performance of the database computer is
not degraded during the backup process. The NetWorker server controls the
tape drives, labeling and initiates tape changers to provide tape media. The
database itself is managed (start, stop, set to “Backup Mode”) by NSR-ORA on
the database computer.
Currently, NSR-ORA only works together with the NAS devices from
NetworkAppliance®(NetApp), called NetApp Filers. Fujitsu Siemens Computers
will provide support for other systems in future.
As a special feature of NSR-ORA using NetApp Filers the software supports the
NetApp Snapshot feature. A NetApp Snapshot is a snapshot of the database at
the moment the NetApp Snapshot is created. This snapshot file will not be
changed even though the database is still operating and its data is being continuously. Due to NetApp’s specific working method the snapshots are created in
a very short time and generally request only a small amount of storage.
Therefore several snapshots can be held on a NetApp Filer.
The integration of NetApp snapshots provides the following benefits:
●
Using a Snap-Restore (Restore of a snapshot) can considerably reduce the
time needed for a recovery operation.
●
Using NetApp snapshots allows the administrator to back up the database
more often. This reduces the time needed for recovering a database after a
logical error has occurred, because the number of redo logs to be readjusted will be smaller.
●
An online backup of the database takes only a few minutes (from the
database server’s point of view). Since the database in a consistent state is
already stored on a snapshot a following tape backup of the data is
completely independent of the backup itself.
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Prerequisites
●
The administrator may perform an offline backup within a few minutes. This
is done by making NSR-ORA shut down the database automatically and
restarting it after the backup process is completed (i. e. after the snapshot is
created).
●
Depending on the backup cycle you may reduce the number of tape media
units required may be reduced (having additional snapshot backups means
that less backups on tape media are required).
6.2
Prerequisites
To run NSR-ORA with NDMP support you have to meet the following system
requirements:
●
Solaris 7 or higher
●
NetWorker 6.x
●
NetApp:
– Operating System OnTap 6.x
– Snap-Restore license
●
Oracle 8.x or 9i
●
NetWorker is licensed for usage with NDMP
●
The NetApp Filer is equipped with a local volume, or Three Party Copy (with
NetWorker V6.1 or higher) is installed (NDMP volume).
The database computer must support NDMP backups and satisfy the following
requirements:
●
One NDMP-Client
i
●
To perform a test backup, the volume /vol/vol0 of the NetAppFilers is
recommended (see “Defining the NDMP Client”).
One NetApp volume
When the test backup has successfully completed you can start configuring
NSR-ORA using the command scripts configure_nsrora and configure_networker.
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
6.3
Recommendations
Recommendations
This section provides recommendations for using NSR-ORA to backup data on
high performance NAS devices.
1. For the Archivelog files backup (Amon backups) and to save the backup information a second, non-NDMP device is required. This device should be
connected to the database server. However it is also possible to use another
remote volume, that does not work with the NDMP protocol and in this case
the data is backed up over the network.
2. It is strongly recommended that the NetApp volumes be arranged as follows:
– For every database to backup you should use exactly one volume
– On the relevant volume the database should be exclusively stored (this
rule may be broken when backing up the control and the online redolog
files which can be located on another remote and/or local volume).
Otherwise performing a snap restore operation may fail.
3. Do not store all of the control files on the NetApp volume. At least one of
them should be located on a local drive.
4. The online redo log files may be mirrored by ORACLE. It is recommended
that the original online redolog files should be stored on the database’s
NetApp volume and the mirrored files on the database’s local drives.
6.4
Configuring NSR-ORA with NetApp Filers
This section describes the configuration of NSR-ORA when used with a NetApp
Filer.
6.4.1
Configuring NSR-ORA
When running the configure_nsrora script the administrator will be asked if the
database is located on an NDMP Filer. If “yes” the administrator will be prompted
to setup the following parameters:
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
NETAPP_SNAP_NR=[0-30]:[0-30]:[0-30]
This parameter defines the type of and number of snapshot backups to be
kept for a restore process. NetWorker uses three types of snapshots:
snapshot-by-week (first value), snapshots-by-day (second value) and
snapshot-by-hour (third value). The sum of all three values must not exceed
the value 30. NetWorker names the snapshot files automatically using the
following scheme:
nsrora_<ORACLE-SID>_<mode>.<n>
The type names of the snapshots do not describe the retention time the
relevant snapshot files are kept by NetWorker not the cycle when the backup
is performed.
– The last “n” snapshots are called “snapshots-by-hour”. They may have
been created on several different days. If, for example, the value for this
type of snapshots is set to “10” and NetWorker generates a snapshot
every six hours, then the ten snapshots-by-hour kept by NetWorker date
from three different days. With every new snapshot-backup of the
database all the existing snapshots-by-hour are renamed. Therefore the
oldest snapshot will either be deleted (if the number of existing snapshot
files exceeds the defined maximum number of snapshots to be kept) or
will be renamed to a snapshot-by-day (see example).
– The snapshots-by-day are “older” snapshots-by-hour, which have been
set and renamed to this type of snapshot. A snapshot-by-hour is
renamed to a snapshot-by-day if it is the last backup on a day. Whenever
NetWorker generates a new snapshot-by-day the existing snapshots-byday are renamed using a higher counter. Thus the oldest snapshot will
either be deleted (if the number of existing snapshot files exceeds the
defined maximum number of snapshots to be kept) or will be renamed to
a snapshot-by-week (see example).
– The snapshots-by-week are former snapshots-by-day which were
generated on the last day of a week. If a snapshot-by-day is to be
deleted, NSR-ORA checks, wether it has been created on the last day of
a week as defined by the LAST_WEEK_DAY parameter (see below). Thus
this snapshot becomes a snapshot-by-week. The oldest snapshot will be
deleted (if the number of existing snapshot files exceeds the defined
maximum number of snapshots to be kept) and all the other snapshotsby-week to be kept by NetWorker are renamed using a higher counter
(see example).
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
i
Configuring NSR-ORA
Using the snap list command the administrator can obtain information
about the number and the dates of creation of the existing snapshots.
Example:
NETAPP_SNAP_NR=2:5:5
With this command, NetWorker will perform a level-1-backup every four
hours and will additionally run a backup with the level full at 23.00 h every
day. If the parameter NETAPP_SNAP_NR is set as above NetWorker will keep
two snapshots-by-week, five snapshots-by-day and also five snapshots-byhour. The following figure shows the backup cycle resulting from these
settings.
Fr.,
07/13
Fr.,
13.07.01
Fr.,
07/13
Fr.,
13.07.01
Fr.,
07/13
Fr.,
13.07.01
12.00Uhr
h
12.00
16.00Uhr
h
16.00
20.00Uhr
h
20.00
hour.1
07/13 12:00
12:00
07/13
hour.1
07/13 16:00
16:00
07/13
hour.1
07/13 20:00
20:00
07/13
hour.2
07/13 08:00
08:00
07/13
hour.2
07/13 12:00
12:00
07/13
hour.2
07/13 16:00
16:00
07/13
hour.3
07/13 04:00
04:00
07/13
hour.3
07/13 08:00
08:00
07/13
hour.3
07/13 12:00
12:00
07/13
hour.4
07/13 00:00
00:00
07/13
hour.4
07/13 04:00
04:00
07/13
hour.4
07/13 08:00
08:00
07/13
hour.5
07/12 23:03
23:03
07/12
hour.5
07/13 00:00
00:00
07/13
hour.5
07/13 04:00
04:00
07/13
day.1
07/11 23:03
23:03
07/11
day.1
07/12 23:03
23:03
07/12
day.1
07/12 23:03
23:03
07/12
day.2
07/10 23:03
23:03
07/10
day.2
07/11 23:03
23:03
07/11
day.2
07/11 23:03
23:03
07/11
day.3
07/09 23:03
23:03
07/09
day.3
07/10 23:03
23:03
07/10
day.3
07/10 23:03
23:03
07/10
day.4
07/08 23:03
23:03
07/08
day.4
07/09 23:03
23:03
07/09
day.4
07/09 23:03
23:03
07/09
day.5
07/07 08:00
08:00
07/07
day.5
07/08 23:03
23:03
07/08
day.5
07/08 23:03
23:03
07/08
week.1
06/30 23:00
23:00
06/30
week.1
07/07 08:00
08:00
07/07
week.1
07/07 08:00
08:00
07/07
week.2
06/23 23:00
23:00
06/23
week.2
06/30 23:00
23:00
06/30
week.2
06/30 23:00
23:00
06/30
deleted
gelöscht
deleted
gelöscht
The snapshots marked with black symbols are newly created at the
respective point in time (e.g. 16.00 h). The snapshots with icons filled with
dark-gray are files from the previous backup run (e.g. for the 16.00 h backup
this is the backup started at 12.00 h), which have been renamed. The
snapshots marked with light-gray filled symbols are not only renamed but
also changed their type (e.g. from “hour” to “day”).
i
Depending on the rate of change of the database to be backed up the
NETAPP_SNAP_NR parameter provides an optimized adjustment of
the database snapshots to be saved. If the data and state of your
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
database rarely change, it may be usefull to keep the snapshots for a
few weeks. If the rate of change is high it may be suffient to keep only
a few snapshots-by-hour. For a moderate rate of changes it may be
useful to keep a number of snapshots-by-hour and, additionally, some
snapshots-by-day. The NetWorker adminstrator and database(s)
administrator need to work together to tune the parame-ters for a
particular database.
NETAPP_SNAP_PERCENTAGE
Using this parameter the administrator can define the percentage of storage
on a volume that will be reserved to store the snapshot files (15% by default).
As a NetWorker Administrator you will be informed, if the amount of storage
available for snapshot backups is running out. To use snapshot backups the
NAS filer needs an amount of free space on the database volume. With the
snap_reserve command (NetApp) you can reserve this amount of storage.
Before starting a backup NSR-ORA checks whether the amount of reserved
free disk space is sufficient or not: If the percentage of usage reaches more
than 90% of the amount reserved by snap_reserve (e.g. 90% used of 25%
reserved) then NSR-ORA gives a warning delivered by the savegroup
completion notification of NetWorker.
i
With the reserved amount of storage being entirely used for the
snapshot files NetWorker will use additional free space on the
database volume for storage (if it is available). This causes no
reduction of functionality.
LAST_WEEK_DAY
The LAST_WEEK_DAY parameter has to be changed, if you want to specify
a day other than Saturday as the last day of the week to generate the
snapshots-by-week. By default the value for the LAST_WEEK_DAY
parameter is set to Saturday but you may set it to any other day of the week.
This parameter is not queried during configuration with configure_nsrora but
must be changed manually in the dbo${ORACLE_SID}.init configuration file.
NFS_SAVE
The NFS_SAVE parameter has to be set, if the archivelog files are stored on
a NAS filer volume, too. The allowed values are 0 (inactive) and 1 (active).
Choosing NFS_SAVE=1 you activate the backup from NFS-mounted file
systems for the backup of the archivelog files.
NDMP_LOGIN_METHOD
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
The configure_nsrora script does not ask you to set up this parameter, so you
may have to set it up manually by changing the dbo${ORACLE_SID}.init
configuration file. With this parameter you choose the protocol NSR-ORA
uses to communicate with the database. Currently NetWorker supports the
two following protocols:
RSH
The RSH communication protocol is highly recommended. The administrator must simply enregister the database computer in the files /etc/hosts
and /etc/hosts.equiv on the NAS Filer.
TELNET
Using Telnet imposes considerable restrictions: Only one user may
communicate with the NAS Filer at a time. There are two other
parameters (User and Password) to be set up using the TELNET value for
NDMP_LOGIN_METHOD. Both values are set up automatically by NSRORA using the values of the respective NetWorker client’s resources.
6.4.2
Configuring NetWorker
The configure_networker command creates a number of NetWorker resources by
providing a menu with three possible scenarios. Selecting one of these
scenarios creates the following NetWorker resources accordingly:
●
the two or three groups, needed for the scenario
●
the NSR-ORA clients for database backups (additonally, one required
NDMP client must be created afterwards manually)
●
the required schedules
●
the required pools
●
the required label templates
You may modify these resources later to adapt them to your local installation
(see also section “NetWorker Resources”).
!
The following required resource is not created automatically and you
must therefore be created manually before starting the first backup:
●
the NDMP client necessary to define the communication between
NetWorker and the NetApp filer (see “Defining the NDMP Client”).
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
The configure_networker script provides the choice between the following three
scenarios for the backup of NetApp filers:
Scenario 1: Three Backups per Day
In this scenario NetWorker performs three backups per day: At 12:00 h and
16:00 h a snapshot backup is performed. At 23:50 h another snapshot with an
additional tape backup is started.
Three NetWorker groups are generated automatically:
●
one group for tape backups:
–
–
–
–
●
Interval: 24 h
Level: full
Start: 23:50
Name: <ORACLE_SID>_NDMP
two groups for pure snapshot backups:
–
–
–
–
Interval: 24 h
Level: 1
Start: 12:00 and 16:00 respectivly
Name: <ORACLE_SID>_SNAP1 and <ORACLE_SID>_SNAP2 respectivly
Scenario 2: Two Backups per Day
In this scenario NetWorker performs two backups per day: At 12:00 h a
snapshot backup is performed. At 23:50 h another snapshot with an additional
tape backup is started.
Two NetWorker groups are generated automatically:
●
one group for tape backups:
–
–
–
–
●
Interval: 24 h
Level: full
Start: 23:50
Name: <ORACLE_SID>_NDMP
one group for pure snapshot backups:
–
–
–
–
Interval: 24 h
Level: 1
Start: 12:00
Name: <ORACLE_SID>_SNAP
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
Scenario 3: Seven Backups per Day
In this scenario NetWorker performs seven backups per day: Every four hours
a snapshot backup is performed. Additionally another snapshot with an
additional tape backup is started at 23:50 h.
Two NetWorker groups are generated automatically:
●
one group for tape backups:
–
–
–
–
●
Interval: 24 h
Level: full
Start: 23:50
Name: <ORACLE_SID>_NDMP
one group for pure snapshot backups:
–
–
–
–
6.4.3
Interval: 4 h
Level: 1
Start: 10:00
Name: <ORACLE_SID>_SNAP_INTERVAL
Defining the NDMP Client
After performing the configuration with configure_networker the administrator
must create an additional client resource manually with NetWorker to define the
communication with the NetApp Filer.
This client resource must be created with following attributes:
●
Save set attribute
You should use /vol/vol0 as the save set.
●
Remote access attribute
To allow the database server to start an NDMP backup, the database server
must be entered into the attribute Remote access.
This means, the database host must be entered as a client with Remote
access rights.
●
User attribute
For an NDMP backup the NDMP client needs the name of the NDMP user
for the NetApp Filer (usually root). This name must be entered into the
attribute Remote user.
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
Password attribute
●
For an NDMP backup the NDMP client needs the password of the NDMP
user for the NetApp Filer. This password must be entered into the attribute
Password.
Backup command attribute
●
Here the administrator should enter the following command:
nsrndmp_save -T dump
Application information attribute
●
Here enter the following information:
HIST=y
UPDATE=y
REMOTE=y
NDMP attribute
●
For the NDMP client this attribute must be set to “Yes”.
6.4.4
NetWorker Resources
This section describes the NetWorker resources, that must be created for
NetApp backups. The configure_networker script creates these according to the
scenarios described above.If these scenarios do not completely fit your needs
you may change the use of the NetWorker resources as detailed below.
The resources needed for the amon archive monitor are not described in
this section. The backup of redolog files is not done using NDMP so the
rules for these resources are the same as for NSR-ORA without NDMP.
i
Groups
For both of the backup types “pure snapshot backup” and “snapshot plus the
backup” there should be at least two different groups:
●
One or more groups for the snapshot backups.
One group may be sufficient if you want to perform “pure snapshot
backups” indepentent of tape backups on a regular interval (e.g. every
four hours).
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
However if you require irregular snapshot backups (e.g. only during the
working hours) then you must create different groups depending the
different snapshot backups.
One group for the tape backup (daily of every n days).
●
There are some basic rules to follow for the groups controlling the backup
cycle:
Be careful to provide a sufficient delay between the tape backup and
the next pure snapshot backup. If the snapshot backups is stated
while the tape backup is still running, the snapshot backup is skipped
(i.e. the snapshot backup will not be performed).
i
On the other hand if a tape backup is started while NSR-ORA is still
running a snapshot backup then the tape backup will be paused and
resumed again after the snapshot backup is finished.
If a tape backup cannot be finished because of missing tapes, no
further snapshot backup will be performed. So take care to provide a
sufficient amount of tapes for the tape backup. Please take note of all
messages from NetWorker informing you that a new tape is needed
for (the completion of) the backup and react as soon as possible
allowing NetWorker to perform all scheduled backups (tape and
snapshot backups) as planned.
!
For the groups the following additional settings are necessary:
●
Interval attribute
The Interval attribute should be set to a value between 1 and 24 hours.
Thus you can perform a snapshot backup regularily, e.g. every 4 hours.
If NetWorker should only perform a few backups per day, then for every
start time a separate group can be created. In this case the Interval
parameter is not used. Scheduling more backups during a day the
Interval parameter saves configuration effort since in this case you may
use only two groups.
i
●
The Interval and Level (see below) attributes are so called hidden
attributes. For additional information see the Fujitsu Siemens
Computers manual NetWorker V6.0: Installing, Configuring and
Operating.
Level attribute
This attribute allows setting the backup level of a group to a fixed value.
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
This parameter overwrites the backup level of the respective NetWorker
client which is preset in the clients schedule. Thus you can reduce the
configuration to the definition of a client.
“Pure snapshot backups” always use level “1” in this attribute.
!
●
If the groups Interval value is set to less than 24 hours, NetWorker
will overwrite the Level setting with “incr”, as long as the Force
incremental attribute is not set to “No”.
Force incremental attribute
This parameter should always be set to “No”. In the other case all
backups would be performed with level “incr”, if the Interval attribute is set
to a value less than 24.
Clients
To backup an Oracle database on a NAS Filer, at least two NetWorker client
resources are required:
●
One client must contain the configuration of the Filer connection. This
client defines a NetWorker NDMP client which is not used for backup by
NSR-ORA, but is used by NetWorker whenever an NDMP backup is
issued by NSR-ORA (for this NetWorker needs the NDMP user name
and password). It is recommended to use save set /vol/vol0 for that client
and to use this client also to perform a test backup of the NetWorker
NDMP configuration,before starting backups of the database. The
attributes of that client are described in detail in section “Defining the
NDMP Client”.
●
For the database backup at least one NSR-ORA client is required. The
clients Name attribute must contain the name of the database host.
Those NSR-ORA clients need the following additional settings:
●
Save set attribute
for NDMP backups NSR-ORA provides two new save sets
– NDMP
– NDMP_OFFLINE
Both anchors (directories below /nsr/oracle/${ORACLE_SID}/) are created
by the configure_nsrora script and contain files with the names of all
tablespaces (or at least a file named System).
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Configuring NSR-ORA
If the save set NDMP is backed up, NSR-ORA saves the database in the
current state (online or offline). This corresponds to save set
${ORACLE_SID}_full_save).
Save set NDMP_OFFLINE causes NSR-ORA always to switch the
database offline, before it is saved. After the backup NSR-ORA restarts
the database, if it has been online during the start of the backup.
Otherwise the database stays offline.
i
If OFFLINE backup has been selected during configure_nsrora, you
may select any daily backup as OFFLINE backup. To achieve this
it is necessary to modify the configuration as follows:
Example:
To perform an OFFLINE backup on Sunday, the parameters
should be set as follows:
– While configuring the NDMP client select the
NDMP_without_sunday schedule instead of the daily schedule.
Using this schedule there will be a full level backup every day
except Sunday. Additional the sunday schedule has to be
created and assigned to the NDMP_OFFLINE client.
– The NDMP_OFFLINE client has to be assigned to the
<ORACLE_SID>_NDMP group.
If an OFFLINE backup was not selected during configure_nsrora
NetWorker assigns the daily schedule to the NDMP client and the
client with save set NDMP_OFFLINE is assigned to the
<ORACLE_SID>_NDMP_OFFLINE group. Nevertheless, this
group is not set to enabled.
●
NDMP attribute
All NetWorker clients that perform database backups (save set NDMP...)
must not modify this value (i.e. leave this attribute set to “No”!).
●
Groups attribute
To perform snapshot and tape backups on a given save set, the client’s
membership to the relevant groups must be assigned to this attribute.
The number of groups depends on your own requirements.
●
Schedule attribute
As the type of snapshot backup is determined by the Level group attribute
the “daily” schedule can be used (one level “full” backup every day) for
tape backups.
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Backup
Pools
NSR-ORA needs two or three NetWorker pools for NDMP backups:
●
One pool is required for the backup of database tablespace files. This
pool must be usable on the NDMP drive.
These pools must also be available for normal (non-NDMP)
drives. When starting a database backup and during the backup
of log- and Oracle files the relevant pool is also required by the
NetWorker server or storage node.
!
●
One or two pools are required for the backup of archive redo log files,
depending on whether the redolog files are backed up only once or twice
(mirror backup).
Amon Client
To backup redolog files a separate client with the name of the database host
is neccessary. However you do not need to create one if a client already
exists for this host.
Label Template
For each of the pools mentioned above, NSR-ORA needs a separate label
template.
6.5
Backup with NSR-ORA
The following section gives information about controlling the backup process
and the errors which may occur.
6.5.1
Backup methods
There are two parameters which determine the method NSR-ORA uses for a
backup:
●
Backup level:
With this parameter set to “full” a snapshot is created and afterwards NSRORA starts a backup on a NetWorker medium using the nsrndmp_save
command.
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Backup
If the level is set to a value between 1 and 9 then NSR-ORA simply creates
a snapshot.
Save Set:
When the save set NDMP is used the current state of the database will be
backed up.
●
With the save set specified NSR-ORA performs an offline backup of the
database. Normally, this means that NSR-ORA shuts down the database
before starting a backup and starts it again when the backup process has
completed.
i
At the end of a backup process (regardless of whether there has only
been a snapshot created or additionally a tape backup) NSR-ORA will
additionally store the following information on tape media:
●
For online backups: the binary control file
/nsr/oracle/<ORACLE_SID>/tmp/cntl_<ORACLE_SID>_bin
●
The control file as an ASCII trace file
/nsr/oracle/<ORACLE_SID>/tmp/cntl_<ORACLE_SID>_trace
●
The init file /nsr/oracle/<ORACLE_SID>/tmp/init<ORACLE_SID>.ora
●
The information and log files of the current backup process in the
●
The information and log files of the current backup process in the
directory /nsr/oracle/<ORACLE_SID>/prot.
These files may be used for debugging or a manually controlled recovery.
6.5.2
The savegrp completion notification
The savegrp completion notification contains all error messages and warnings
which occur while running a NDMP backup or creating a snapshot.
i
Some messages are not designed for a direct delivery (e.g. license
missing or NetWorker configuration failure). In that case the savegrp
completion notification contains the name of the log file to which the error
message has been written.
Performing a NDMP backup the savegrp state meets exactly the real state of the
backup process. If, for example, savegrp completion contains the message
“succeeded” (all clients have been backed up successfully) then the complete
backup of the Oracle database has been successfully. When an error occurs
during the backup savegrp completion shows the “failed” message.
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
i
Recovery
The saveinfo.sh command is not relevant to the NDMP backup method,
and it will not be started after the backup has completed. If the backup
process has been performed successfully you will be informed by the
savegrp completion notification.
6.5.3
Debugging
To enable debugging functions NSR-ORA creates a number of files within the
/nsr/oracle/{ORACLE_SID}/prot/dbg directory for every backup performed. In this
directory you can find records of the communication between NSR-ORA and
the NetApp filer, between NSR-ORA and Networker and, finally, between NSRORA and Oracle. If any problems occur during the communication it is recommended to send a copy of the relevant record file to your IT service.
6.6
Recovery with NSR-ORA
In general there are no syntax changes to the commands of the recovery
process from NSR-ORA V3.1B to the newer version 3.2. This section describes
the processes and procedures involved in performing a recovery of NetApp
backups.
6.6.1
Recover until time (ora_recut)
Performing a Recover until time NSR-ORA checks, whether there is a snapshot
preceding the specified time. Later you will be asked to decide to use this
snapshot or an respective tape backup for the recovery.
Example:
ora_recut -t 2001-07-27:16:55:00 NET816
Which saved Version do you want to use
1) the Snapshot from <Jul 27 16:46 nsrora_NET816_actual>
2) the tape version from <Jul 27 16:38
nsrora_NET816_NDMP_ONLINE>
(default 1:) >
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Recovery
Selecting a tape backup the database recovery process is the same as the one
for a usual database recover performed by NSR-ORA (see also section
“Recovery Until Time with Logical Errors”).
If you specify a recovery using a snapshot NSR-ORA first tries to perform a
switchlog of the most recently created log files and then stops the database and
archivelog daemon. Subsequently NSR-ORA allows the administrator to
perform the recovery using the snap restore command. For every required filer
volume the following dialog appears:
Do you want to do a snap restore for the volume
/vol/mangan1 on filer netapp
The DB Files are on top of the volume but if there are other
files aside, they will also be resetted
(n|y) (default y) >
i
If the top level database directory is not equal to the mounting point the
warning created by NSR-ORA will be more urgent and in this case the
default answer is “n”.
With a snap restore selected NSR-ORA checks the relevant volume to see if the
archivelog files are stored on it. If so the following warning message is
displayed:
ATTENTION !!! ATTENTION !!! ATTENTION !!!
================================================================
The Archivelog directory ’/netapp_oracle/saparch’ is on the same
volume as the DB files you want to restore!
If you go on all current archivelog files will be overwritten.
Should I move the files or should I overwrite them?
1) copy the files
2) overwrite the files
3) abort the recover session
> 2
Are you really sure that you want to overwrite
the existent logfiles???
Be aware that the DB could be irrecoverable if you overwrite
needed logfiles!
Overwrite? (y | n) > y
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Recovery
If you want NSR-ORA to copy the files (selection “1”) it will search for the
directory with the most free space and will ask you whether the existing
archivelog files should be copied to this filesystem. Additionally you may also
specify a different directory. Later on, if there are redolog files required for the
database recovery, NSR-ORA searches this directory for these redolog files
before it tries to perform a backup from the NetWorker tape media.
However if overwriting of the archivelog files is selected (as shown in the
example above, selection “2”), the dialog shown above appears.
If you do not select the snap restore function, (because for example another
database is located on the same volume, NSR-ORA checks the relevant directories, if it is possible to copy the files using the ndmp_copy command or not. If
there are any files not related to the database to be restored to one of the
relevant directories, NSR-ORA asks, if you really want ndmp_copy to be used.
This is because this command also restores these other files. If the answer is
“no”, NSR-ORA performs the default copy command.
6.6.2
Crash Recovery (nsrora_rec)
When a database file is corrupted NSR-ORA first tries to recover it using the last
(most recent) snapshot (see “Example 1: Recovery based on the last
snapshot”). However the recovery can also be based on a snapshot dating from
an earlier point in time (see “Example 2: Recovery based on a specified
snapshot”).
Performing a recover based on a snapshot NSR-ORA restores the files on the
database computer using the standard copy command or, if possible, it restores
them on the NetApp filer using the ndmp_copy command. In the second case the
corrupted files must be the only ones stored in the relevant directory, because
the ndmp_copy command copies complete directories only but not single files.
Example 1: Recovery based on the last snapshot
Should I try to use a copy from the LAST SNAPSHOT or
do you want to select a specific snapshot?
1) last SNAPSHOT
2) select SNAPSHOT
(default 1:) > 1
INFO: doing ndmpcopy for /netapp_oracle/sapdata4/proti_1
System Administrator’s Guide
U42252-J-Z915-1-76
NSR-ORA and NetApp
Recovery
Example 2: Recovery based on a specified snapshot
Should I try to use a copy from the LAST SNAPSHOT or
do you want to select a specific snapshot
1) last SNAPSHOT
2) select SNAPSHOT
(default 1:) > 2
Please select one of the following SNAPSHOTS
1) Jul 31 04:03 nsrora_NET816_hour.1
2) Jul 31 00:03 nsrora_NET816_hour.2
3) Jul 30 23:03 nsrora_NET816_hour.3
4) Jul 30 20:03 nsrora_NET816_hour.4
5) Jul 30 16:03 nsrora_NET816_hour.5
6) Jul 29 23:03 nsrora_NET816_day.1
7) Jul 28 23:03 nsrora_NET816_day.2
8) Jul 27 23:03 nsrora_NET816_day.3
Please select a SNAPSHOT from the list
> 1
INFO: doing ndmpcopy for /netapp_oracle/sapdata4/proti_1
i
The redolog files needed for the recovery of the Oracle database are
restored directly from NSR-ORA (as in version 3.1) and made available
to the database.
System Administrator’s Guide
U42252-J-Z915-1-76
Index
${ORACLE_SID}_full_save
(directory) 37
A
actual_saves_of_tablespaces
(file) 18
amon 6, 9
daemon 10
display status of 11
figure 10
log file 13
on OPS systems 10
start 11
task of 10
threshold value of 10
AMON_GROUP
shell parameter 34
AMON_INTERVALL
shell parameter 33
AMON_MAX_LOGS
shell parameter 32
AMON_MIN_LOGS
shell parameter 32
AMON_MIRROR_GROUP
shell parameter 35
AMON_MIRROR_SERVER
shell parameter 33
AMON_RM_WITHOUT_MIRROR_D
ELAY
shell parameter 35
AMON_SAVE_ARG_NR
shell parameter 35
AMON_WATCH_DELAY
shell parameter 33
AMON_WATCH_DOG_CMD
shell parameter 33
archive mode (ARCHIVELOG) 5
ARCHIVELOG (mode) 5
ARCHIVELOG mode 5, 9
archiving monitor 6, 9
System Administrator’s Guide
daemon 10
display status of 11
figure 10
log file 13
start the 11
task of 10
threshold value of 10
audience 1
B
backup
backup command (attribute) 19
backup schedules 19
backup sets 36
checking 18
complete 6
complete backup 36, 37
database 6
database backup 5
level "full" 6, 36, 37
observing at the screen 19
offline 15
offline backup 36
online 17
partial backup 36
with NetApp 70
backup command (attribute) 19
backup component 14
configuration 17
backup level
for NetApp backups 70
backup schedules
checking global consistency
of 19
backup sets
defining 36
BACKUP_COMPRESS
shell parameter 32
BACKUP_GROUP
shell parameter 35
backup_TS_on_different_clients
U42252-J-Z915-1-76
Index
(configuration file)
17
C
command
nsrndmp_save 70
nsrora_rec 45, 50
ora_recut (for NetApp) 72
components
backup component 14
of NSR-ORA 9
recovery component 45
COMPRESS_LEVEL
shell parameter 35
config.sh
output 27
config.sh (configuration tool) 27, 36
configuration
backup component 17
configuration tool (config.sh) 27
configure_networker 38
configure_nsrora (script) 29
for NetApp 57
NetWorker 38
NSR-ORA 26, 29
on OPS clusters 40
preparations 27
quick start 21
configuration file 4
backup_TS_on_different_clients
17
configure_networker 38
example 39
for NetApp 63
configure_nsrora 29
for NetApp 59
CONTR_SERVER
shell parameter 31
control file
loss of 7
control program
start_amon.sh 11
conventions 1
crash analysis 47
crash analyzer 6
System Administrator’s Guide
crash recovery 47
for NetApp 74
D
data dictionary 4
database
archive mode (ARCHIVELOG) 5
backup 5, 6
examination 47
offline backup 15
online backup 17
recovery 6, 47
restore 6, 47
database backup 5
database file 4
loss of 6
database structure
logical 4, 6
physical 4, 6
directories
for weekdays 36
directory
${ORACLE_SID}_full_save 37
/nsr/oracle/${ORACLE_SID}/
prot 18
tablespaces 18
E
enabler 22
error message
No space left on device
55
F
fault categories 7
file
actual_saves_of_tablespaces 18
save_history 18
full backup 36
recommendations on
37
I
installation 22
NSR-ORA from CD 24
U42252-J-Z915-1-76
Index
on OPS clusters
40
L
LAST_WEEK_DAY
shell parameter 62
Licensing
snaprestore (NetApp) 23
licensing
NetApp licenses required 23
logical database structure 4, 6
M
MIRROR
shell parameter
33
N
NDMP 57
NDMP_LOGIN_METHOD
shell parameter 62, 63
NetApp
backups 70
configuration of NSR-ORA 57
configure_nsrora 59
crash recovery 74
licenses required 23
recover until time 72
recovery 72
snaprestore (License) 23
NetApp backups
backup level 70
NetWorker ressources 66
predefined scenarios 63
recommendations 59
save set 71
NETAPP_SNAP_NR
shell parameter 60, 61
NETAPP_SNAP_PERCENTAGE
shell parameter 62
Network Attached Storage (NAS) 57
NetworkAppliance 2, 57
NetWorker
configuration 38
configure_networker 38
System Administrator’s Guide
ressources for NetApp
backups 66
NFS_SAVE
shell parameter 62
No space left on device (error
message) 55
NSR_INST
shell parameter 31
NSR_MAX_REC_DEV
shell parameter 34
NSR_MAX_RECOVER
shell parameter 34
NSR_SERVER
shell parameter 31
nsrndmp_save (command) 70
NSR-ORA
backup component 14
components 9
configuration 26, 29
in OPS clusters 40
installation 22
installing from CD 24
recovery component 45
tasks and objectives of 3
update from an earlier
version 26
NSR-ORA enabler 22
nsrora_rec (command) 45, 50
nsrora_save_cmd 19
O
offline backup 3, 36
of the database 15
offline recovery
manual 52
online backup 3
of the database 17
OPS
amon 10
recovery 54
recovery until time 55
restore redolog files 54
shell parameter 34
OPS clusters
U42252-J-Z915-1-76
Index
configuration 40
installation on 40
remarkable things 40
OPS_INSTANCES
shell parameter 34
ora_recut (command)
for NetApp 72
ORACLE_GROUP
shell parameter 31
ORACLE_HOME
shell parameter 31
ORACLE_OWNER
shell parameter 31
ORACLE_SID
shell parameter 31
P
parameter file 4
partial backup 36
physical database structure
protocol files 18
for backup 18
under OPS 55
redolog files 4
loss of 7
restore (on OPS) 54
restore 6
redolog files (OPS) 54
RSH
shell parameter 63
4, 6
R
recover
until time 72
recover until time
for NetApp 72
recovery 6
automatic 7, 47
crash recovery 47
database 47
for an optional point in time
for NetApp 72
fully-automatic 50
manual 48, 52
offline 52
performing a 45, 48
types of 45
until time 46, 55
with OPS 54
recovery component 45
recovery types 45
recovery until time
System Administrator’s Guide
49
S
save set
for NetApp backups 71
save sets
defining 36
save_history (file) 18
saveinfo.sh
shell script 18, 19
scenarios
for NetApp backups 63
SEND_MAIL_TO
shell parameter 33
shell parameter
AMON_GROUP 34
AMON_INTERVALL 33
AMON_MAX_LOGS 32
AMON_MIN_LOGS 32
AMON_MIRROR_GROUP 35
AMON_MIRROR_SERVER 33
AMON_RM_WITHOUT_MIRROR
_DELAY 35
AMON_SAVE_ARG_NR 35
AMON_WATCH_DOG_CMD 33
BACKUP_COMPRESS 32
BACKUP_GROUP 35
COMPRESS_LEVEL 35
CONTR_SERVER 31
LAST_WEEK_DAY 62
MIRROR 33
NDMP_LOGIN_METHOD 62,
63
NETAPP_SNAP_NR 60, 61
NETAPP_SNAP_PERCENTAGE
62
NFS_SAVE 62
U42252-J-Z915-1-76
Index
NSR_INST 31
NSR_MAX_REC_DEV 34
NSR_MAX_RECOVER 34
NSR_SERVER 31
OPS 34
OPS_INSTANCES 34
ORACLE_GROUP 31
ORACLE_HOME 31
ORACLE_OWNER 31
ORACLE_SID 31
RSH 63
SEND_MAIL_TO 33
TELNET 63
shell script
saveinfo.sh 18, 19
shell skript
configure_networker (for
NetApp) 63
configure_nsrora (for
NetApp) 59
snaprestore (NetApp-Lizenz) 23
sshell parameter
AMON_WATCH_DELAY 33
T
tablespaces 4, 27
directory 18
tasks
of NSR-ORA 3
TELNET
shell parameter 63
U
update
NSR-ORA
26
System Administrator’s Guide
U42252-J-Z915-1-76
Download PDF

advertising