Chelsio Communications Chelsio T5 User`s guide

Add to my manuals
452 Pages

advertisement

Chelsio Communications Chelsio T5 User`s guide | Manualzz

Ue mo

This document and related products are distributed under licenses restricting their use, copying, distribution, and reverse-engineering.

No part of this document may be reproduced in any form or by any means without prior written permission by Chelsio Communications.

All third party trademarks are copyright of their respective owners.

THIS DOCUMENTATION IS PROVIDED “AS IS” AND WITHOUT ANY EXPRESS OR IMPLIED

WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF

MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.

THE USE OF THE SOFTWARE AND ANY ASSOCIATED MATERIALS (COLLECTIVELY THE

“SOFTWARE”) IS SUBJECT TO THE SOFTWARE LICENSE TERMS OF CHELSIO

COMMUNICATIONS, INC.

Chelsio Communications (Headquarters)

370 San Aleso Ave.

Suite 100

Sunnyvale, CA 94085

U.S.A www.chelsio.com

Tel: 408.962.3600

Fax: 408.962.3661

Chelsio KK (Japan)

SHIMA Akasaka Bldg.

Minato-ku, Tokyo

Japan 107-0052

Tel: 03-6234-4353

Chelsio (India) Private Limited

Subramanya Arcade, Floor 3, Tower B

No. 12, Bannerghatta Road,

Bangalore-560029

Karnataka,

India

Tel: +91-80-4039-6800

Fax: +91-80-4039-6807

Sales

For all sales inquiries please send email to [email protected]

Support

For all support related questions please send email to [email protected]

Copyright © 2013.Chelsio Communications. All Rights Reserved.

Chelsio ® is a registered trademark of Chelsio Communications.

All other marks and names mentioned herein may be trademarks of their respective companies.

Chelsio T5/T4 Unified Wire For Linux ii

Version

1.0.0

1.0.1

1.0.2

1.0.3

1.0.4

1.0.5

1.0.6

1.0.7

1.0.8

1.0.9

1.1.0

1.1.1

Document History

Revision Date

12/08/2011

01/09/2013

01/27/2013

03/26/2013

04/12/2013

06/20/2013

08/17/2013

10/22/2013

03/08/2013

05/15/2013

07/26/2013

08/01/2013

Chelsio T5/T4 Unified Wire For Linux iii

TABLE OF CONTENTS

I.

CHELSIO UNIFIED WIRE

1.

Introduction

1.1. Features

1.2. Hardware Requirements

1.3. Software Requirements

1.4. Package Contents

2.

Hardware Installation

3.

Software/Driver Installation

3.1. Pre-requisites

3.2. Installing Chelsio Unified Wire from source

3.3. Installing Chelsio Unified Wire from RPM

3.4. Firmware update

4.

Software/Driver Uninstallation

4.1. Uninstalling Chelsio Unified Wire from source

4.2. Uninstalling Chelsio Unified Wire from RPM

5.

Configuring Chelsio Network Interfaces

5.1. Configuring network-scripts

5.2. Creating network-scripts

5.3. Checking Link

6.

Software/Driver Update

II.

NETWORK (NIC/TOE)

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

2.1. Loading in NIC mode (without full offload support)

2.2. Loading in TOE mode (with full offload support)

3.

Software/Driver Unloading

3.1. Unloading the NIC driver

3.2. Unloading the TOE driver

4.

Software/Driver Configuration and Fine-tuning

4.1. Instantiate Virtual Functions (SR-IOV)

4.2. Performance tuning

III.

VIRTUAL FUNCTION NETWORK (VNIC)

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

Chelsio T5/T4 Unified Wire For Linux

47

48

51

51

52

52

52

48

49

50

50

50

51

58

59

59

60

11

12

12

13

13

13

16

19

19

20

32

35

36

36

41

43

43

44

45

46

iv

2.

Software/Driver Loading

2.1. Loading the driver

3.

Software/Driver Unloading

3.1. Unloading the driver

4.

Software/Driver Configuration and Fine-tuning

4.1. Instantiate Virtual Functions

IV.

IWARP (RDMA)

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

2.1. Compiling and Loading iWARP driver

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Testing connectivity with ping and rping

4.2. Enabling various MPIs

4.3. Setting up NFS-RDMA

V.

WD-UDP

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Compiling and Loading

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Accelerating UDP Socket communications

VI.

WD-TOE

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Running the application

VII.

ISCSI PDU OFFLOAD TARGET

1.

Introduction

1.1. Features

Chelsio T5/T4 Unified Wire For Linux

81

82

82

82

84

85

86

86

92

93

93

93

95

96

97

97

64

67

68

69

69

70

79

65

65

65

67

61

61

62

62

63

63

98

99

99

v

1.2. Hardware Requirements

1.3. Software Requirements

2.

Software/Driver Loading

2.1. Latest iSCSI Software Stack Driver Software

2.2. Obtaining the iSCSI Software License

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Command Line Tools

4.2. iSCSI Configuration File

4.3. A Quick Start Guide for Target

4.4. The iSCSI Configuration File

4.5. Challenge-Handshake Authenticate Protocol (CHAP)

4.6. Target Access Control List (ACL) Configuration

4.7. Target Storage Device Configuration

4.8. Target Redirection Support

4.9. The command line interface tools “iscsictl” & “chisns”

4.10. Rules of Target Reload (i.e. “on the fly” changes)

4.11. System Wide Parameters

VIII.

ISCSI PDU OFFLOAD INITIATOR

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Accelerating open-iSCSI Initiator

4.2. Auto login from cxgb4i initiator at OS bootup

IX.

FCOE FULL OFFLOAD INITIATOR

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Configuring Cisco Nexus 5010 switch

4.2. FCoE fabric discovery verification

4.3. Formatting the LUNs and Mounting the Filesystem

4.4. Creating Filesystem

4.5. Mounting the formatted LUN

Chelsio T5/T4 Unified Wire For Linux

137

138

138

139

140

142

143

143

146

100

102

104

104

106

108

109

109

109

110

113

122

124

125

128

129

134

135

147

148

148

148

149

150

151

151

156

162

164

164

vi

X.

OFFLOAD BONDING DRIVER

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Offloading TCP traffic over a bonded interface

XI.

UDP SEGMENTATION OFFLOAD AND PACING

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Modifying the application

4.2. Configuring UDP Pacing

XII.

OFFLOAD IPV6 DRIVER

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

3.

Software/Driver Unloading

3.1. Unloading the NIC driver

3.2. Unloading the TOE driver

XIII.

BYPASS DRIVER

1.

Introduction

1.1. Features

1.2. Hardware Requirements

1.3. Software Requirements

2.

Software/Driver Loading

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Starting ba server

4.2. Bypass API (CLI)

Chelsio T5/T4 Unified Wire For Linux

172

173

174

174

175

176

177

177

179

166

167

167

167

169

170

171

171

180

181

181

181

183

184

184

184

185

186

186

187

188

189

190

191

191

191

vii

XIV.

WD SNIFFING AND TRACING

1.

Theory of Operation

1.1. Hardware Requirements

1.2. Software Requirements

2.

Installation and Usage

2.1. Installing basic support

2.2. Using Sniffer (wd_sniffer)

2.3. Using Tracer (wd_tcpdump_trace)

XV.

CLASSIFICATION AND FILTERING

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Usage

2.1. Configuration

2.2. Creating Filter Rules

2.3. Listing Filter Rules

2.4. Removing Filter Rules

XVI.

TRAFFIC MANAGEMENT

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Software/Driver Loading

3.

Software/Driver Unloading

4.

Software/Driver Configuration and Fine-tuning

4.1. Traffic Management Rules

4.2. Configuring Traffic Management

5.

Usage

5.1. Non-Offloaded Connections

5.2. Offloaded Connections

5.3. Offloaded Connections with Modified Application

XVII.

UNIFIED WIRE MANAGER (UM)

1.

Introduction

1.1. Features

1.2. Reference Architecture

1.3. Unified Wire Manager Components

1.4. Authentication and encryption

2.

Hardware and Software

2.1. Supported Adapters

Chelsio T5/T4 Unified Wire For Linux

209

210

210

211

212

213

214

214

216

220

220

220

221

222

223

223

224

224

225

226

226

197

198

199

200

201

201

201

201

203

204

204

204

206

206

206

208

208

viii

2.2. Platform/Component Matrix

2.3. Platform/Driver Matrix

3.

Installing Unified Wire Manager

4.

Verifying UM components status

4.1. Verifying Management Agent

4.2. Verifying Management Client

4.3. Verifying Management Station

5.

Management Agent

5.1. Communication

5.2. Configuration

5.3. Service configuration

5.4. Firewall

6.

CLI client

6.1. CLI Help system

6.2. Client conflict resolution

7.

Web GUI client

7.1. Management Station

7.2. Accessing Web Management Interface

7.3. Layout and Navigation

7.4. Home page

7.5. System page

7.6. Network page

7.7. Storage Page

7.8. Hardware Features

8.

Uninstalling Unified Wire Manager

8.1. Uninstalling Management Agent

8.2. Uninstalling Management Client

8.3. Uninstalling Management Station

XVIII.UNIFIED BOOT

1.

Introduction

1.1. Hardware Requirements

1.2. Software Requirements

2.

Flashing configuration file, firmware & option ROM

3.

Configuring PXE Server

4.

PXE boot process

4.1. For Legacy PXE boot

4.2. For uEFI PXE Boot

5.

FCoE boot process

5.1. For Legacy FCoE boot

5.2. For uEFI FCoE Boot

Chelsio T5/T4 Unified Wire For Linux

339

348

348

353

364

364

373

340

340

341

342

347

235

235

236

238

239

252

263

290

320

337

337

337

338

232

232

232

233

234

234

234

227

227

229

230

230

231

231

232

ix

6.

iSCSI boot process

6.1. For Legacy iSCSI boot

6.2. For uEFI iSCSI Boot

7.

Creating Driver Update Disk (DUD)

7.1. Creating DUD for RedHat Enterprise Linux

7.2. Creating DUD for Suse Enterprise Linux

8.

OS Installation

8.1. Installation using Chelsio NIC DUD (PXE only)

8.2. Installation on FCoE LUN

8.3. Installation on iSCSI LUN

XIX.

APPENDIX A

1.

Troubleshooting

2.

Chelsio End-User License Agreement (EULA)

387

387

400

420

420

420

422

422

432

437

447

448

450

Chelsio T5/T4 Unified Wire For Linux x

Chapter I. Chelsio Unified Wire

I. Chelsio Unified Wire

Chelsio T5/T4 Unified Wire For Linux Page 11

Chapter I. Chelsio Unified Wire

1. Introduction

Thank you for choosing Chelsio T5/T4 Unified Wire adapters. These high speed, single chip, single firmware cards provide enterprises and data centers with high performance solutions for various Network and Storage related requirements.

The Terminator 5 (T5) is Chelsio’s next generation of highly integrated, hyper-virtualized

40/10GbE controllers. The T5 is built around a programmable protocol-processing engine, with full offload of a complete Unified Wire solution comprising NIC, TOE, iWARP RDMA, ISCSI,

FCoE and NAT support. It scales true 40Gb line rate operation from a single TCP connection to thousands of connections, and allows simultaneous low latency and high bandwidth operation thanks to multiple physical channels through the ASIC.

The T4 adapters can fully offload TCP, UDP, iSCSI, iWARP and FCoE over a single Unified

Wire. The adapters also fully support SR-IOV, EVB/VNTag, DCB, Traffic Management and

Filtering.

Ideal for all data, storage and high performance clustering applications, the T5/T4 Adapters enable a unified fabric over a single wire by simultaneously running all unmodified IP sockets,

Fibre Channel and InfiniBand applications over Ethernet at line rate.

Designed for deployment in virtualized data centers, cloud service installations and high performance computing environments, Chelsio T5/T4 adapters bring a new level of performance metrics and functional capabilities to the computer networking industry.

Chelsio Unified Wire software comes in two formats: Source code and RPM package forms.

Installing from source requires compiling the package to generate the necessary binaries. You can choose this method when you are using a custom-built kernel. You can also install the package using the interactive GUI installer. In other cases, download the RPM package specific to your operating system and follow the steps mentioned to install the package. Please note that the OFED software required to install Chelsio iWARP driver comes bundled in both source as well as RPM packages.

This document describes the installation, use and maintenance of the software and its various components.

1.1.

Features

The Chelsio Unified Wire Package uses a single command to install various drivers and utilities.

It consists of the following software:

Network (NIC/TOE)

Virtual Function Network (vNIC)

iWARP (RDMA)

Chelsio T5/T4 Unified Wire For Linux Page 12

Chapter I. Chelsio Unified Wire

WD-UDP

WD-TOE

iSCSI PDU Offload Target

iSCSI PDU Offload Initiator

FCoE full offload Initiator

Offload Bonding driver

UDP Segmentation Offload and Pacing

Offload IPv6 driver

Bypass driver

Classification and Filtering feature

Traffic Management feature (TM)

Unified Wire Manager (UM)

Unified Boot Software

Utility Tools(cop,cxgbtool,t4_perftune,benchmark tools, sniffer & tracer)

libs (iWARP and WD-UDP libraries)

For detailed instructions on loading, unloading and configuring the drivers/tools please refer to their respective sections.

1.2.

Hardware Requirements

The Chelsio Unified Wire software supports Chelsio T5 and T4 Series of Unified Wire Adapters.

To know more about the list of adapters supported by each driver, please refer to their respective sections.

1.3.

Software Requirements

The Chelsio Unified Wire software has been developed to run on 64-bit Linux based platforms and therefore it is a base requirement for running the driver. To know more about the complete list of operating systems supported by each driver, please refer to their respective sections.

1.4.

Package Contents

1.4.1. Source Package

The Chelsio Unified Wire source package consists of the following files/directories:

install.py,dialog.py: Python scripts needed for the GUI installer.

EULA: Chelsio’s End User License Agreement

install.log: File containing installation summary.

Chelsio T5/T4 Unified Wire For Linux Page 13

Chapter I. Chelsio Unified Wire

docs: The docs directory contains support documents - README, Release Notes and

User’s Guide (this document) for the software.

libs: This directory is for libraries required to install the WD-UDP and iWARP drivers. The libibverbs library has implementation of RDMA verbs which will be used by iWARP applications for data transfers. The librdmacm library works as an RDMA connection manager. The libcxgb4 library works as an interface between the above mentioned generic libraries and Chelsio iWARP driver. The libcxgb4_sock library is a LD_PRELOAD-able library that accelerates UDP Socket communications transparently and without recompilation of the user application.

Makefile: The Makefile for building and installing from the source.

sample_machinefile: Sample file used during iWARP installation on cluster nodes.

scripts: Support scripts used by the Unified Wire Installer.

specs: The packaging specification files required for building RPM packages.

src: Source code for different drivers.

support: This directory contains source files for the dialog utility.

tools:

ba_tools: Management and configuration tools for bypass adapters.

benchmarks: This directory contains various benchmarking tools to measure throughput and latency of various networks.

cop: The cop tool compiles offload policies into a simple program form that can be loaded into the kernel and interpreted. These offload policies are used to determine the settings to be used for various connections. The connections to which the settings are applied are based on matching filter specifications. Please find more details on this tool in its manual page (run man cop command).

cxgbtool: The cxgbtool queries or sets various aspects of Chelsio network interface cards. It complements standard tools used to configure network settings and provides functionality not available through such tools. Please find more details on this tool in its manual page (run man cxgbtool command).

Note To use cxbtool for FCoE Initiator driver, use [root@host]# cxgbtool stor –h

rdma_tools: This directory contains iWARP benchmarking tools.

sniffer: This directory contains sniffer tracing and filtering libraries. See WD Sniffing and Tracing chapter for more information.

um: This directory contains Unified Wire Manager RPMs for different distributions and Management Station configuration files.

chsetup: The chsetup tool loads NIC,TOE and iWARP drivers, and creates WD-

UDP configuration file.

chstatus: This utility provides status information on any Chelsio NIC in the system.

t4_perftune.sh: This shell script is to tune the system for higher performance. It achieves it through modifying the IRQ-CPU binding. This script can also be used to change Tx coalescing settings.

t4-forward.sh: RFC2544 Forward test tuning script.

Chelsio T5/T4 Unified Wire For Linux Page 14

Chapter I. Chelsio Unified Wire

uname_r: This file is used by chstatus script to verify if the Linux platform is supported or not.

wdload: UDP acceleration tool.

wdunload: Used to unload all the loaded Chelsio drivers.

Uboot: There are two sub-directories in the Uboot directory: optionROM and LinuxDUD.

The Uboot directory contains Unified Boot Option ROM images and a Legacy Flash Utility, which can be used to flash Unified Boot Option ROM onto Chelsio's T5 and T4 Converged

Network Adapters (CNAs). It also contains T5 and T4 Firmware and configuration files.

The LinuxDUD directory contains image (.img) files required to update drivers for Linux

(RHEL and SLES) distributions.

1.4.2. RPM package

The Chelsio Unified Wire RPM package consists of the following:

docs: The docs directory contains support documents i.e. README, Release Notes and

User’s Guide (this document) for the software.

scripts: Support scripts used by the Unified Wire Installer.

DRIVER-RPMS: RPM packages of Chelsio drivers.

OFED-RPMS: OFED RPM packages required to install iWARP driver.

install.py: Python script that installs the RPM package. See Software/Driver Installation

section for more information.

uninstall.py: Python script that uninstalls the RPM package. See

Software/Driver

Uninstallation section for more information.

EULA:

Chelsio’s End User License Agreement.

Chelsio T5/T4 Unified Wire For Linux Page 15

Chapter I. Chelsio Unified Wire

2. Hardware Installation

1. Shutdown/power off your system.

2. Power off all remaining peripherals attached to your system.

3. Unpack the Chelsio adapter and place it on an anti-static surface.

4. Remove the system case cover according to the system manufacturer’s instructions.

5. Remove the PCI filler plate from the slot where you will install the Ethernet adapter.

6. For maximum performance, it is highly recommended to install the adapter into a PCIE x8 slot.

7. Holding the Chelsio adapter by the edges, align the edge connector with the PCI connector on the motherboard. Apply even pressure on both edges until the card is firmly seated. It may be necessary to remove the SFP (transceiver) modules prior to inserting the adapter.

8. Secure the Chelsio adapter with a screw, or other securing mechanism, as described by the system manufacturer’s instructions. Replace the case cover.

9. After securing the card, ensure that the card is still fully seated in the PCIE x8 slot as sometimes the process of securing the card causes the card to become unseated.

10. Connect a fiber cable, multi-mode for short range (SR) optics or single-mode for long range

(LR) optics, to the 40/10Gb Ethernet adapter or regular Ethernet cable for the 1Gb Ethernet adapter.

11. Power on your system.

12. Verify if the adapter was installed successfully by using the lspci command a. For T5 adapters :

[root@host]# lspci |grep -i Chelsio

07:00.0 Ethernet controller: Chelsio Communications Inc T520-LL-CR Unified

Wire Ethernet Controller

07:00.1 Ethernet controller: Chelsio Communications Inc T520-LL-CR Unified

Wire Ethernet Controller

07:00.2 Ethernet controller: Chelsio Communications Inc T520-LL-CR Unified

Wire Ethernet Controller

07:00.3 Ethernet controller: Chelsio Communications Inc T520-LL-CR Unified

Wire Ethernet Controller

07:00.4 Ethernet controller: Chelsio Communications Inc T520-LL-CR Unified

Wire Ethernet Controller

07:00.5 SCSI storage controller: Chelsio Communications Inc T520-LL-CR

Unified Wire Storage Controller

07:00.6 Fibre Channel: Chelsio Communications Inc T520-LL-CR Unified Wire

Storage Controller

Chelsio T5/T4 Unified Wire For Linux Page 16

Chapter I. Chelsio Unified Wire b. And for T4 adapters :

[root@host]# lspci | grep –i Chelsio

03:00.0 Ethernet controller: Chelsio Communications Inc T420-CR Unified Wire

Ethernet Controller

03:00.1 Ethernet controller: Chelsio Communications Inc T420-CR Unified Wire

Ethernet Controller

03:00.2 Ethernet controller: Chelsio Communications Inc T420-CR Unified Wire

Ethernet Controller

03:00.3 Ethernet controller: Chelsio Communications Inc T420-CR Unified Wire

Ethernet Controller

03:00.4 Ethernet controller: Chelsio Communications Inc T420-CR Unified Wire

Ethernet Controller

03:00.5 SCSI storage controller: Chelsio Communications Inc T420-CR Unified

Wire Storage Controller

03:00.6 Fibre Channel: Chelsio Communications Inc T420-CR Unified Wire

Storage Controller

03:00.7 Ethernet controller: Chelsio Communications Inc Device 0000

For Chelsio T5/T4 adapters, the physical functions are currently assigned as:

Physical functions 0 - 3: for the SR-IOV functions of the adapter

Physical function 4: for all NIC functions of the adapter

Physical function 5: for iSCSI

Physical function 6: for FCoE

Physical function 7: Currently not assigned

Once Unified Wire package is installed and loaded, examine the output of dmesg to see if the card is discovered.

For T5 adapters: eth2: Chelsio T520-LL rev 0 10GBASE-SFP RNIC PCIe 8 GT/s x8 MSI-X, Offload capable

0000:07:00.4: S/N: RE12130097, P/N: 11011675004

And, for T4 adapters: eth0: Chelsio T420-CR rev 2 10GBASE-SFP RNIC PCIe x8 MSI-X

0000:04:00.4: S/N: PT18111226, P/N: 110112140D0

Chelsio T5/T4 Unified Wire For Linux Page 17

Chapter I. Chelsio Unified Wire

The above outputs indicate the hardware configuration of the adapters as well as the Serial numbers. As observed by the x8, the card is properly installed in an x8 slot on the machine and using MSI-X interrupts.

Note

Network device names for Chelsio’s physical ports are assigned using the following convention: the port farthest from the motherboard will appear as the first ethernet network interface. However, for T580-LP-CR and T420-BT adapters, the association of physical ethernet ports and their corresponding network device names is opposite. For these adapters, the port nearest to the motherboard will appear as the first network interface.

Chelsio T5/T4 Unified Wire For Linux Page 18

Chapter I. Chelsio Unified Wire

3. Software/Driver Installation

There are two main methods to install the Chelsio Unified Wire package: from source and RPM.

If you decide to use source, you can install the package using CLI or GUI mode.

If you decide to use RPM, you can install the package using Menu or CLI mode.

Irrespective of the method chosen for installation, the machine needs to be rebooted for changes to take effect.

The following table describes the various configuration tuning options available during installation and drivers/software installed with each option by default:

T5/T4

Configuration

Tuning Option

Description Driver/Software installed

Unified Wire

Low latency

Networking

High capacity

RDMA

High capacity

TOE

UDP Seg.Offload

& Pacing

T5 Wire Direct

Latency

Configures adapters to run multiple protocols like NIC/TOE, iWARP, iSCSI and FCoE

Initiator simultaneously

Configures adapters to run NIC/TOE and iWARP traffic with low latency specially needed for financial applications.

Configures adapters to establish a large number of RDMA connections

Configures adapters to establish a large number of TOE connections

NIC/TOE, vNIC, iWARP, WD-UDP, iSCSITarget, iSCSI Initiator, Bonding, IPv6,

Sniffer & Tracer, FCoE Initiator,

UM(Agent,Client), Filtering, TM

NIC/TOE, vNIC, iWARP, WD-UDP, WD-

TOE, IPv6, Sniffer & Tracer,

UM(Agent,Client), Filtering, TM

NIC/TOE, vNIC, iWARP, WD-UDP, Bonding,

IPv6, Sniffer & Tracer, UM(Agent,Client),

Filtering, TM

NIC/TOE, vNIC, Bonding, IPv6,

UM(Agent,Client), Filtering,

TM

NIC/TOE, IPv6, USO, UM(Agent,Client),

Filtering, TM

Configures adapters to establish a large number of UDP Segmentation Offload connections

Configures T5 adapters to provide low Wire

Direct latency. Supported only on T5 adapters.

NIC/TOE,iWARP,WD-UDP,WD-TOE,

UM(Agent,Client)

Note

High capacity RDMA and High capacity TOE options are currently not supported with T5 Adapters.

3.1.

Pre-requisites

Depending on the component you choose to install, please ensure that the following requirements are met, before proceeding with the installation.

If you want to install OFED with NFSRDMA support, please refer “Setting up NFS-RDMA”

in iWARP (RDMA) ( Click here ).

 If you’re planning to install iSCSI PDU Offload Initiator, please install openssl-devel package.

IPv6 should be enabled in the machine to use the RPM Packages.

Chelsio T5/T4 Unified Wire For Linux Page 19

Chapter I. Chelsio Unified Wire

3.2.

Installing Chelsio Unified Wire from source

3.2.1. GUI mode (with Dialog utility)

i. Download the tarball ChelsioUwire-x.x.x.x.tar.gz from Chelsio Download Center, http://service.chelsio.com/ ii. Untar the tarball using the following command:

[root@host]# tar -zxvfm ChelsioUwire-x.x.x.x.tar.gz iii. Change your current working directory to Chelsio Unified Wire package directory and run the following script to start the GUI installer:

[root@host]# ./install.py iv. If Dialog utility is present, you can skip to step (v). If not, press ‘y’ to install it when the installer prompts for input. v. Select “View EULA” to read the Chelsio End User License Agreement:

Chelsio T5/T4 Unified Wire For Linux Page 20

Chapter I. Chelsio Unified Wire vi. Selec t “Agree” to continue: vii. Select “install” under “Choose an action”

Chelsio T5/T4 Unified Wire For Linux Page 21

Chapter I. Chelsio Unified Wire viii. Select Enable IPv6-Offload to install drivers with IPv6 Offload support or Disable IPv6-

offload to continue installation without IPv6 offload support. ix. Select the required T5/T4 configuration tuning option:

Note

The tuning options may vary depending on the Linux distribution.

Note

High capacity RDMA and High capacity TOE options are currently not supported with T5 Adapters.

Chelsio T5/T4 Unified Wire For Linux Page 22

Chapter I. Chelsio Unified Wire x. Under “Choose install components”, select “all” to install all the related components for the option chosen in step (ix) or select “custom” to install specific components.

Important

To install Bypass driver, please select Unified Wire with FCoE Initiator in step

(ix). Then select “custom” option. xi. Select the required performance tuning option. a. Enable Binding IRQs to CPUs: Bind MSI-X interrupts to different CPUs and disable

IRQ balance daemon. b. Retain IRQ balance daemon: Bind MSI-X interrupts to different CPUs and do not disable IRQ balance daemon. c. TX-Coalasce: Write tx_coal=2 to modprobe.d/conf.

Note For more information on the Performance tuning options, please refer to the

Software/Driver Configuration and Fine-tuning section of the Network

(NIC/TOE) chapter.

Chelsio T5/T4 Unified Wire For Linux Page 23

Chapter I. Chelsio Unified Wire xii. If you already have the required version of OFED software installed, you can skip this step.

To install OFED-3.5 choose the Install-OFED option. To install a different version, select

Choose-OFED-Version and then select the appropriate version. To skip this step, select

Skip-OFED.

Note OFED is currently not supported on RHEL 6.4 and SLES11SP3. xiii. The selected components will now be installed:

Chelsio T5/T4 Unified Wire For Linux Page 24

Chapter I. Chelsio Unified Wire xiv. After successful installation, summary of installed components will be displayed. xv. S elect “View log” to view the installation log or “Exit” to continue. xvi. Select “Yes” to exit the installer or “No” to go back.

Chelsio T5/T4 Unified Wire For Linux Page 25

Chapter I. Chelsio Unified Wire xvii. Reboot your machine for changes to take effect.

Note Press Esc or Ctrl+C to exit the installer at any point of time.

3.2.1.1. Installation on updated kernels

If the kernel version on your Linux distribution is updated, follow the steps mentioned below to install the Unified Wire package: i. Change your current working directory to Chelsio Unified Wire package directory and run the following script to start the GUI installer:

[root@host]# ./install.py ii. Select “Yes” to continue with the installation on the updated kernel or “No” to exit. iii. Select the nearest supported kernel version from the list and select “OK”. iv. Follow steps (xv) to (xvii) mentioned in the previous section.

Chelsio T5/T4 Unified Wire For Linux Page 26

Chapter I. Chelsio Unified Wire

3.2.2. CLI mode (without Dialog utility)

If your system does not have Dialog or you choose not to install it, follow the steps mentioned below to install the Unified Wire package: i. Download the tarball ChelsioUwire-x.x.x.x.tar.gz from Chelsio Download Center, http://service.chelsio.com/ ii. Untar the tarball using the following command:

[root@host]# tar -zxvfm ChelsioUwire-x.x.x.x.tar.gz iii. Change your current working directory to Chelsio Unified Wire package directory and run the following script to start the installer:

[root@host]# ./install.py iv. When the installer prompts you for your input, press ‘n’ to continue installation without the

Dialog utility. v.

Press ‘1’ to read the Chelsio End User License Agreement; Press ‘q’ to go back to installation menu. Press ‘2’ to Accept the EULA and start the installation or ‘3’ to Disagree and and exit from the installer. vi. Enter the number corresponding to the Configuration tuning option in the Input field and press Enter.

Note

High capacity RDMA and High capacity TOE options are currently not supported with T5 Adapters. vii. If you already have the required version of OFED software installed, you can skip this step.

To install OFED-3.5 choose the Install-OFED option. To install a different version, select

Choose-OFED-Version and then select the appropriate version. To skip this step, select

Skip-OFED.

Note OFED is currently not supported on RHEL 6.4 and SLES11SP3.

Chelsio T5/T4 Unified Wire For Linux Page 27

Chapter I. Chelsio Unified Wire viii. The selected components will now be installed.

After successful installation you can press 1 to view the installation log. Press any other key to exit from the installer.

Important

To install Bypass or WD-TOE driver, run

[root@host]# ./install.py -c <target> and follow steps (iv) - (vi) mentioned above.

To customize the installation, view the help by typing

[root@host]#./install.py –h ix. Reboot your machine for changes to take effect.

3.2.2.1. iWARP driver installation on Cluster nodes

Important

Please make sure that you have enabled password less authentication with ssh on the peer nodes for this feature to work.

Chelsio’s Unified Wire package allows installing iWARP drivers on multiple Cluster nodes with a single command. Follow the procedure mentioned below: i. Create a file (machinefilename) containing the IP addresses or hostnames of the nodes in the cluster. You can view the sample file, sample_machinefile, provided in the package to view the format in which the nodes have to be listed. ii. Now, execute the following command:

[root@host] # ./install.py -C -m <machinefilename> iii. Press ‘1’ to read the Chelsio End User License Agreement; Press ‘q’ to go back to installation menu. Press ‘2’ to Accept the EULA and start the installation or ‘3’ to Disagree and and exit from the installer. iv. Select the required T5/T4 configuration tuning option. The tuning options may vary depending on the Linux distribution.

Note

High capacity RDMA and High capacity TOE options are currently not supported with T5 Adapters. v. Select the required Cluster Configuration. vi. If you already have the required version of OFED software installed, you can skip this step.

To install OFED-3.5 choose the Install-OFED option. To install a different version, select

Choose-OFED-Version and then select the appropriate version. To skip this step, select

Skip-OFED.

Chelsio T5/T4 Unified Wire For Linux Page 28

Chapter I. Chelsio Unified Wire

Note OFED is currently not supported on RHEL 6.4 and SLES11SP3. vii. The selected components will now be installed.

3.2.3. CLI mode

i. Download the tarball ChelsioUwire-x.x.x.x.tar.gz from Chelsio Download Center, http://service.chelsio.com/ ii. Untar the tarball using the following command:

[root@host]# tar -zxvfm ChelsioUwire-x.x.x.x.tar.gz iii. Change your current working directory to Chelsio Unified Wire package directory and build the source using :

[root@host]# make iv. Install the drivers, tools and libraries using the following command:

[root@host]# make install v. The default configuration tuning option is Unified Wire (includes FCoE Initiator). The configuration tuning can be selected using the following commands:

[root@host]# make CONF=<T5/T4 configuration>

[root@host]# make CONF=<T5/T4 configuration> install

Important

Steps (iv)and (v) mentioned above will NOT install Bypass driver. It will have to be installed manually. Please refer to section 3.1.4 CLI mode (individual

drivers) for instructions on installing it.

Note

To view the different configuration tuning options, view help by typing

[root@host]#make help

Chelsio T5/T4 Unified Wire For Linux Page 29

Chapter I. Chelsio Unified Wire

Note

High capacity RDMA and High capacity TOE options are currently not supported with T5 Adapters. vi. Reboot your machine for changes to take effect.

3.2.3.1. Installation on updated kernels

If the kernel version on your Linux distribution is updated, please execute the following command to install the Unified Wire package:

[root@host]# make UNAME_R=<kernel_version>

Where kernel_version is the nearest supported kernel version.

For example, if you want to install the package on a RHEL 6 distribution updated to 2.6.32-

279.2.1.el6 kernel, run the following commands:

[root@host]# make UNAME_R=2.6.32-279.el6

[root@host]# make UNAME_R=2.6.32-279.el6 install

To view the list of the supported kernel versions, run the following command:

[root@host]# make list_kernels

Reboot your machine for changes to take effect.

3.2.4. CLI mode (individual drivers)

You can also choose to install drivers individually. Provided here are steps to build and install

NIC, TOE, iWARP, Bypass, WD-UDP, WD-TOE and UDP Segmentation Offload drivers. To know about other drivers, view help by running make help.

 To build and install NIC driver without offload support :

[root@host]# make nic

[root@host]# make nic_install

Chelsio T5/T4 Unified Wire For Linux Page 30

Chapter I. Chelsio Unified Wire

 To build and install NIC driver with offload support and Offload drivers:

[root@host]# make toe

[root@host]# make toe_install

 To build and install Offload drivers without IPv6 support:

[root@host]# make toe_ipv4

[root@host]# make toe_ipv4_install

 To build and install iWARP driver against outbox OFED:

[root@host]# make iwarp

[root@host]# make iwarp_install

 To build and install all drivers without IPv6 support:

[root@host]# make ipv6_disable=1

[root@host]# make ipv6_disable=1 install

The above step will not install Bypass and WD-TOE drivers.

 To build and install Bypass driver:

[root@host]# make bypass

[root@host]# make bypass_install

 To build and install WD-TOE driver:

[root@host]# make wdtoe

[root@host]# make wdtoe_install

Chelsio T5/T4 Unified Wire For Linux Page 31

Chapter I. Chelsio Unified Wire

To build and install WD-TOE and WD-UDP drivers together:

[root@host]# make wdtoe_wdudp

[root@host]# make wdtoe_wdudp_install

To build and install all drivers with DCBX support (only for T4 adapters)

[root@host]# make dcbx=1

[root@host]# make dcbx=1 install

The offload drivers support UDP Segmentation Offload with limited number of connections

(1024 connections).To build and install UDP Offload drivers which support large number of offload connections (approx 10K):

[root@host]# make udp_offload

[root@host]# make udp_offload_install

The default T5/T4 configuration tuning option is Unified Wire (includes FCoE Initiator). The configuration tuning can be selected using the following commands:

[root@host]# make CONF=<T5/T4 configuration> <Build Target>

[root@host]# make CONF=<T5/T4 configuration> <Install Target>

Note

To view the different configuration tuning options, view the help by typing

[root@host]#make help

Note

Note

If IPv6 is administratively disabled in the machine, by default the drivers will be built and installed without IPv6 Offload support.

High capacity RDMA and High capacity TOE options are currently not supported with T5 Adapters.

3.3.

Installing Chelsio Unified Wire from RPM

3.3.1. Menu Mode

i. Download the tarball specific to your operating system and architecture from Chelsio

Download Center, http://service.chelsio.com/

Chelsio T5/T4 Unified Wire For Linux Page 32

Chapter I. Chelsio Unified Wire ii. Untar the tarball:

E.g. For RHEL 6.3, untar using the following command:

[root@host]# tar -zxvfm ChelsioUwire-x.x.x.x-RHEL6.3_x86_64.tar.gz iii. Change your current working directory to Chelsio Unified Wire package directory and run the following command:

[root@host]# ./install.py iv. Press “Enter” to read EULA or “q” to quit. v. Press "y" to accept the EULA and continue or "q" to quit. vi. Select the required T5/T4 configuration tuning option as described below. Enter the corresponding number in the Input field and press Enter.

Note

High capacity RDMA and High capacity TOE options are currently not supported with T5 Adapters. vii. Next press 1 if you want to install OFED along with the configuration tuning option selected in step (vi) or 2 to skip OFED installation. Press Enter to continue with your choice. viii. Select the Installation type as described below. Enter the corresponding number in the Input field and press Enter.

1. Install all Chelsio drivers: Install all the drivers provided in the Unified Wire package.

2. Install NIC and TOE drivers: Install only the Network and TOE drivers.

3. Install bypass drivers and tools: Install only the bypass drivers and tools.

4. Install WD drivers and library: Install only Wire Direct drivers and libraries.

5. EXIT: Exit the installer.

Note

The Installation options may vary depending on the Configuration tuning option selected. ix. The selected components will now be installed. x. Reboot your machine for changes to take effect.

Note

If the installation aborts with the message "Resolve the errors/dependencies manually and restart the installation", please go through the install.log to resolve errors/dependencies and then start the installation again.

Chelsio T5/T4 Unified Wire For Linux Page 33

Chapter I. Chelsio Unified Wire

3.3.2. CLI mode

i. Download the tarball specific to your operating system and architecture from Chelsio

Download Center, http://service.chelsio.com/ ii. Untar the tarball:

E.g. For RHEL 6.3, untar using the following command:

[root@host]# tar -zxvfm ChelsioUwire-x.x.x.x-RHEL6.3_x86_64.tar.gz iii. Change your current working directory to Chelsio Unified Wire package directory and install

Unified Wire using:

[root@host]# ./install.py –i <nic_toe/all/bypass/udpso/wd>

nic_toe : NIC and TOE drivers only

all : all Chelsio drivers (including FCoE Initiator) built against inbox OFED

bypass : bypass drivers and tools

udpso : UDP segmentation offload capable NIC and TOE drivers only

wd : Wire Direct drivers and libraries only

Note

The Installation options may vary depending on the Linux distribution. iv. The default configuration tuning option is Unified Wire (includes FCoE Initiator). The configuration tuning can be selected using the following command:

[root@host]# ./install.py –i <Installation mode> -c <Configuration>

Note

To view the different configuration tuning options, view the help by typing

[root@host]# ./install.py –h

High capacity RDMA and High capacity TOE options are currently not

Note supported with T5 Adapters.

Chelsio T5/T4 Unified Wire For Linux Page 34

Chapter I. Chelsio Unified Wire v. To install OFED and Chelsio drivers built against OFED, run the above command with -o option.

[root@host]# ./install.py –i <Installation mode> -c <Configuration> -o vi. Reboot your machine for changes to take effect.

3.3.2.1. Driver installation on cluster nodes

Important

Please make sure that you have enabled password less authentication with ssh on the peer nodes for this feature to work. i. Create a file (machinefilename) containing the IP addresses or hostnames of the nodes in the cluster. You can view the sample file, sample_machinefile, provided in the package to view the format in which the nodes have to be listed. ii. Navigate to ChelsioUwire directory and execute the following command:

[root@host] # ./install.py -C -m <machinefilename> -i

<nic_toe/all/bypass/udpso/wd> -c <T5/T4 configuration> -o

Here, -o parameter will install OFED and Chelsio drivers built against OFED iii. Reboot your machine for changes to take effect.

3.4.

Firmware update

The T5 and T4 firmwares are installed on the system, typically in /lib/firmware/cxgb4 , and the driver will auto-load the firmwares if an update is required. The kernel must be configured to enable userspace firmware loading support:

Device Drivers -> Generic Driver Options -> Userspace firmware loading support

The firmware version can be verified using ethtool:

[root@host]# ethtool -i <iface>

Chelsio T5/T4 Unified Wire For Linux Page 35

Chapter I. Chelsio Unified Wire

4. Software/Driver Uninstallation

Similar to installation, the Chelsio Unified Wire package can be uninstalled using two main methods: from the source and RPM, based on the method used for installation. If you decide to use source, you can uninstall the package using CLI or GUI mode.

4.1.

Uninstalling Chelsio Unified Wire from source

4.1.1. GUI mode (with Dialog utility)

i. Change your current working directory to Chelsio Unified Wire package directory and run the following script to start the GUI installer:

[root@host]# ./install.py ii. Select “View EULA” to read the Chelsio End User License Agreement:

Chelsio T5/T4 Unified Wire For Linux Page 36

Chapter I. Chelsio Unified Wire iii. Select “Agree” to continue: iv. Select “uninstall” , Under “Choose an action”

Chelsio T5/T4 Unified Wire For Linux Page 37

Chapter I. Chelsio Unified Wire v. Select “all” to uninstall all the installed drivers, libraries and tools or select “custom” to remove specific components. vi. The selected components will now be uninstalled. vii. After successful uninstalltion, summary of the uninstalled components will be displayed.

Chelsio T5/T4 Unified Wire For Linux Page 38

Chapter I. Chelsio Unified Wire viii. Select “View log” to view uninstallation log or “Exit” to continue. ix.

Select “Yes” to exit the installer or “No” to go back.

Note Press Esc or Ctrl+C to exit the installer at any point of time.

4.1.2. CLI mode (without Dialog utility)

Run the following script with –u option to uninstall the Unified Wire Package:

[root@host]# ./install.py –u <target>

Note View help by typing [root@host]# ./install.py –h for more information

4.1.3. CLI mode

Change your current working directory to Chelsio Unified Wire package directory and uninstall using the following command:

[root@host]# make uninstall

Chelsio T5/T4 Unified Wire For Linux Page 39

Chapter I. Chelsio Unified Wire

Note

Uninstalling Unified Wire package will not uninstall Unified Wire Manager. Refer to the next section, CLI mode (individual drivers/software) to remove the software manually.

4.1.3.1. iWARP driver uninstallation on Cluster nodes

To uninstal iWARP drivers on multiple Cluster nodes with a single command, run the following command:

[root@host] # ./install.py -C -m <machinefilename> -u all

The above command will remove Chelsio iWARP (iw_cxgb4) and TOE (t4_tom) drivers from all the nodes listed in the machinefilename file.

4.1.4. CLI mode (individual drivers/software)

You can also choose to uninstall drivers/software individually. Provided here are steps to uninstall NIC, TOE, iWARP, Bypass, WD-TOE, UDP Segmentation Offload drivers and Unified

Wire Manager (UM). To know about other drivers, access help by running make help

 To uninstall NIC driver :

[root@host]# make nic_uninstall

 To uninstall offload driver:

[root@host]# make toe_uninstall

 To uninstall iWARP driver:

[root@host]# make iwarp_uninstall

 To uninstall Bypass driver:

[root@host]# make bypass_uninstall

Chelsio T5/T4 Unified Wire For Linux Page 40

Chapter I. Chelsio Unified Wire

 To uninstall UDP Segmentation Offload driver:

[root@host]# make udp_offload_uninstall

To uninstall WD-TOE driver:

[root@host]# make wdtoe_uninstall

vii. To uninstall WD-TOE and WD-UPD drivers together

[root@host]# make wdtoe_wdudp_uninstall

To uninstall Unified Wire Manager (UM)

[root@host]# make uninstall UM_UNINST=1

OR

[root@host]# make tools_uninstall UM_UNINST=1

4.2.

Uninstalling Chelsio Unified Wire from RPM

Change your current working directory to Chelsio Unified Wire package directory and and run the following command:

[root@host]# ./uninstall.py <inbox/ofed>

inbox : for removing all Chelsio drivers.

ofed : for removing OFED and Chelsio drivers.

Note

The uninstallation options may vary depending on Linux distribution. View help by typing [root@host]# ./uninstall.py –h for more information.

Chelsio T5/T4 Unified Wire For Linux Page 41

Chapter I. Chelsio Unified Wire

Note

Uninstalling Unified Wire package will not uninstall Unified Wire Manager. Refer to the Unified Wire Manager (UM) chapter to remove the software manually

( Click here ).

4.2.1.1. iWARP driver uninstallation on Cluster nodes

To uninstal iWARP drivers on multiple Cluster nodes with a single command, run the following:

[root@host] # ./install.py -C -m <machinefilename> -u

The above command will remove Chelsio iWARP (iw_cxgb4) and TOE (t4_tom) drivers from all the nodes listed in the machinefilename file.

Chelsio T5/T4 Unified Wire For Linux Page 42

Chapter I. Chelsio Unified Wire

5. Configuring Chelsio Network Interfaces

In order to test Chelsio adapters’ features it is required to use two machines both with Chelsio’s

(T5, T4 or both) network adapters installed. These two machines can be connected directly without a switch (back-to-back), or both connected to a switch. The interfaces have to be declared and configured. The configuration files for network interfaces on Red Hat Enterprise

Linux (RHEL) distributions are kept under /etc/sysconfig/network-scripts .

Note

Some operating systems may attempt to auto-configure the detected hardware and some may not detect all ports on a multi-port adapter. If this happens, please refer to the operating system documentation for manually configuring the network device.

5.1.

Configuring network-scripts

A typical interface network-script (e.g. eth0) on RHEL 6.3 looks like the following:

# file: /etc/sysconfig/network-scripts/ifcfg-eth0

DEVICE="eth0"

HWADDR=00:30:48:32:6A:AA

ONBOOT="yes"

NM_CONTROLLED="no"

BOOTPROTO="static"

IPADDR=10.192.167.111

NETMASK=255.255.240.0

Note On earlier versions of RHEL the NETMASK attribute is named IPMASK. Make sure you are using the right attribute name.

In the case of DHCP addressing the last two lines should be removed and

BOOTPROTO="static" should be changed to BOOTPROTO="dhcp"

The ifcfg-ethX files have to be created manually. They are required for bringing the interfaces up and down and attribute the desired IP addresses.

Chelsio T5/T4 Unified Wire For Linux Page 43

Chapter I. Chelsio Unified Wire

5.2.

Creating network-scripts

To spot the new interfaces, make sure the driver is unloaded first. To that point ifconfig -a | grep HWaddr should display all non-chelsio interfaces whose drivers are loaded, whether the interfaces are up or not.

[root@host]# ifconfig -a | grep HWaddr eth0 Link encap:Ethernet HWaddr 00:30:48:32:6A:AA

Then load the driver using the modprobe cxgb4 command (for the moment it does not make any difference whether we are using NIC-only or the TOE-enabling driver). The output of ifconfig should display the T5/T4 interfaces as:

[root@host]# ifconfig -a | grep HWaddr eth0 Link encap:Ethernet HWaddr 00:30:48:32:6A:AA eth1 Link encap:Ethernet HWaddr 00:07:43:04:6B:E9 eth2 Link encap:Ethernet HWaddr 00:07:43:04:6B:F1 eth3 Link encap:Ethernet HWaddr 00:07:43:04:6B:F9 eth4 Link encap:Ethernet HWaddr 00:07:43:04:6C:01

For each interface you can write a configuration file in /etc/sysconfig/network-scripts .

The ifcfg-eth1 could look like:

# file: /etc/sysconfig/network-scripts/ifcfg-eth1

DEVICE="eth1"

HWADDR=00:07:43:04:6B:E9

ONBOOT="no"

NM_CONTROLLED="no"

BOOTPROTO="static"

IPADDR=10.192.167.112

NETMASK=255.255.240.0

From now on, the eth1 interface of the adapter can be brought up and down through the ifup eth1 and ifdown eth1 commands respectively. Note that it is of course not compulsory to create a configuration file for every interface if you are not planning to use them all.

Chelsio T5/T4 Unified Wire For Linux Page 44

Chapter I. Chelsio Unified Wire

5.3.

Checking Link

Once the network-scripts are created for the interfaces you should check the link i.e. make sure it is actually connected to the network. First, bring up the interface you want to test using ifup eth1 .

You should now be able to ping any other machine from your network provided it has ping response enabled.

Chelsio T5/T4 Unified Wire For Linux Page 45

Chapter I. Chelsio Unified Wire

6. Software/Driver Update

For any distribution specific problems, please check README and Release Notes included in the release for possible workaround.

Please visit Chelsio support web site http://service.chelsio.com/ for regular updates on various software/drivers. You can also subscribe to our newsletter for the latest software updates.

Chelsio T5/T4 Unified Wire For Linux Page 46

Chapter II. Network (NIC/TOE)

II. Network (NIC/TOE)

Chelsio T5/T4 Unified Wire For Linux Page 47

Chapter II. Network (NIC/TOE)

1. Introduction

Chelsio’s T5 and T4 series of Unified Wire Adapters provide extensive support for NIC operation, including all stateless offload mechanisms for both IPv4 and IPv6 (IP, TCP and UDP checksum offload, LSO - Large Send Offload aka TSO - TCP Segmentation Offload, and assist mechanisms for accelerating LRO - Large Receive Offload).

A high performance fully offloaded and fully featured TCP/IP stack meets or exceeds software implementations in RFC compliance. Chelsio’s T5/T4 engine provides unparalleled performance through a specialized data flow processor implementation and a host of features designed for high throughput and low latency in demanding conditions and networking environments, using standard size Ethernet frames.

TCP offload is fully implemented in the hardware, thus freeing the CPU from TCP/IP overhead.

The freed CPU can be used for any computing needs. The TCP offload in turn removes network bottlenecks and enables applications to take full advantage of the networking capabilities.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with Chelsio

Network driver:

T520-LL-CR

T520-SO-CR*

T520-CR

T540-CR

T580-LP-CR

T420-CR

T440-CR

T422-CR

T420-SO-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

*Only NIC driver supported

Chelsio T5/T4 Unified Wire For Linux Page 48

Chapter II. Network (NIC/TOE)

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the Network driver is available for the following versions:

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04, 3.2.0-23

Kernel.org linux-2.6.35*

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

* Limited QA performed.

Chelsio T5/T4 Unified Wire For Linux Page 49

Chapter II. Network (NIC/TOE)

2. Software/Driver Loading

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail.

2.1.

Loading in NIC mode (without full offload support)

To load the Network driver without full offload support, run the following command:

[root@host]# modprobe cxgb4

2.2.

Loading in TOE mode (with full offload support)

To enable full offload support, run the following command:

[root@host]# modprobe t4_tom

Note Offload support needs to be enabled upon each reboot of the system. This can be done manually as shown above.

Chelsio T5/T4 Unified Wire For Linux Page 50

Chapter II. Network (NIC/TOE)

3. Software/Driver Unloading

3.1.

Unloading the NIC driver

To unload the NIC driver, run the following command:

[root@host]# rmmod cxgb4

3.2.

Unloading the TOE driver

Please reboot the system to unload the TOE driver.

Chelsio T5/T4 Unified Wire For Linux Page 51

Chapter II. Network (NIC/TOE)

4. Software/Driver Configuration and Fine-tuning

4.1.

Instantiate Virtual Functions (SR-IOV)

To instantiate the Virtual functions, load the cxgb4 driver with num_vf parameter with a non-zero value. For example:

[root@host]# modprobe cxgb4 num_vf=1,0,0,0

The number(s) provided for num_vf parameter specifies the number of Virtual Functions to be instantiated per Physical Function. The Virtual Functions can be assigned to Virtual Machines

(Guests). A maximum of 64 Virtual Functions can be instantiated with 16 Virtual Functions per

Physical Function. Loading the cxgb4 driver with num_vf parameter loads the cxgb4vf module

(the driver for Virtual Functions) in the host by default. Hence unload the cxgb4vf module (on the host) before assigning Virtual Functions to the Virtual Machines (Guests), using the following command:

[root@host]# rmmod cxgb4vf

Note

To get familiar with physical and virtual function terminologies, please refer the

PCI Express specification.

4.2.

Performance tuning

In order to auto tune the system and TOE devices for best performance, Chelsio recommends installing the tools which will copy t4_perftune.sh

script to /sbin directory. Run the script by using the following command:

[root@host]# t4_perftune.sh

This script will configure RSS and enable Interrupt Coalescing.

Chelsio T5/T4 Unified Wire For Linux Page 52

Chapter II. Network (NIC/TOE)

Receiver Side Scaling (RSS)

Receiver Side Scaling enables the receiving network traffic to scale with the available number of processors on a modern networked computer. RSS enables parallel receive processing and dynamic ally balances the load among multiple processors. Chelsio’s T5/T4 network controller fully supports Receiver Side Scaling for IPv4 and IPv6.

This script first determines the number of CPUs on the system and then each receiving queue is bound to an entry in the system interrupt table and assigned to a specific CPU. Thus, each receiving queue interrupts a specific CPU through a specific interrupt now. For example, on a 4core system, t4_perftune.sh

gives the following output:

[root@host]# t4_perftune.sh

Discovering Chelsio T4/T5 devices ...

Configuring Chelsio T4/T5 devices ...

Tuning eth7

IRQ table length 4

Writing 1 in /proc/irq/62/smp_affinity

Writing 2 in /proc/irq/63/smp_affinity

Writing 4 in /proc/irq/64/smp_affinity

Writing 8 in /proc/irq/65/smp_affinity eth7 now up and tuned

...

Because there are 4 CPUs on the system, 4 entries of interrupts are assigned. For other T5/T4 network interfaces, you should see similar output message.

Now the receiving traffic is dynamically assigned to one of the system’s CPUs through a T5/T4 queue. This achieves a balanced usage among all the processors. This can be verified, for example, by using the iperf tool. First set up a server on the receiver host:

[root@receiver_host]# iperf –s

Then on the sender host, send data to the server using the iperf client mode. To emulate a moderate traffic workload, use -P option to request 20 TCP streams from the server:

[root@sender_host]# iperf -c receiver_host_name_or_IP -P 20

Chelsio T5/T4 Unified Wire For Linux Page 53

Chapter II. Network (NIC/TOE)

Then on the receiver host, look at interrupt rate at /proc/interrupts :

[root@receiver_host]# cat /proc/interrupts | grep eth6

Id CPU0 CPU1 CPU2 CPU3 type interface

36: 115229 0 0 1 PCI-MSI-edge eth6 (queue 0)

37: 0 121083 1 0 PCI-MSI-edge eth6 (queue 1)

38: 0 0 105423 1 PCI-MSI-edge eth6 (queue 2)

39: 0 0 0 115724 PCI-MSI-edge eth6 (queue 3)

Now interrupts from eth6 are evenly distributed among the 4 CPUs.

Without T5/T4 ’s RSS support, the interrupts caused by network traffic may be distributed unevenly over CPUs. For your information, the traffic produced by the same iperf commands gives the following output in /proc/interrupts .

[root@receiver_host]# cat /proc/interrupts | grep eth6

Id CPU0 CPU1 CPU2 CPU3 type interface

36: 0 9 0 17418 PCI-MSI-edge eth6 (queue 0)

37: 0 0 21718 2063 PCI-MSI-edge eth6 (queue 1)

38: 0 7 391519 222 PCI-MSI-edge eth6 (queue 2)

39: 1 0 33 17798 PCI-MSI-edge eth6 (queue 3)

Here there are 4 receiving queues from the eth6 interface, but they are not bound to a specific

CPU or interrupt entry. Queue 2 has caused a very large number of interrupts on CPU2 while

CPU0 and CPU1 are barely used by any of the four queues. Enabling RSS is thus essential for best performance.

Note

Linux’s irqbalance may take charge of distributing interrupts among CPUs on a multiprocessor platform. However, irqbalance distributes interrupt requests from all hardware devices across processors. For a server with T5/T4 network card constantly receiving large volume of data at 10Gbps, the network interrupt demands are significantly high. Under such circumstances, it is necessary to enable RSS to balance the network load across multiple processors and achieve the best performance.

Chelsio T5/T4 Unified Wire For Linux Page 54

Chapter II. Network (NIC/TOE)

Interrupt Coalescing

The idea behind Interrupt Coalescing (IC) is to avoid flooding the host CPUs with too many interrupts. Instead of throwing one interrupt per incoming packet, IC waits for ‘n’ packets to be available in the Rx queues and placed into the host memory through DMA operations before an interrupt is thrown, reducing the CPU load and thus improving latency. It can be changed using the following command:

[root@host]# ethtool –C ethX rx-frames n

Note For more information, run the following command:

[root@host]# ethtool -h

Large Receive Offload / Generic Receive Offload

Large Receive Offload or Generic Receive Offload is a performance improvement feature at the receiving side. LRO/GRO aggregates the received packets that belong to same stream, and combines them to form a larger packet before pushing them to the receive host network stack.

By doing this, rather than processing every small packet, the receiver CPU works on fewer packet headers but with same amount of data. This helps reduce the receive host CPU load and improve throughput in a 10Gb network environment where CPU can be the bottleneck.

LRO and GRO are different names to refer to the same receiver packets aggregating feature.

LRO and GRO actually differ in their implementation of the feature in the Linux kernel. The feature was first added into the Linux kernel in version 2.6.24 and named Large Receive Offload

(LRO). However LRO only works for TCP and IPv4. As from kernel 2.6.29, a new protocolindependent implementation removing the limitation is added to Linux, and it is named Generic

Receive Offload (GRO). The old LRO code is still available in the kernel sources but whenever both GRO and LRO are presented GRO is always the preferred one to use.

Please note that if your Linux system has IP forwarding enabled, i.e. acting as a bridge or router, the LRO needs to be disabled. This is due to a known kernel issue.

Chelsio’s T5/T4 card supports both hardware assisted GRO/LRO and Linux-based GRO/LRO. t4_tom is the kernel module that enables the hardware assisted GRO/LRO. If it is not already in the kernel module list, use the following command to insert it:

Chelsio T5/T4 Unified Wire For Linux Page 55

Chapter II. Network (NIC/TOE)

[root@host]# lsmod | grep t4_tom

[root@host]# modprobe t4_tom

[root@host]# lsmod | grep t4_tom t4_tom 88378 0 [permanent] toecore 21618 1 t4_tom cxgb4 225342 1 t4_tom

Then T5/T4

’s hardware GRO/LRO implementation is enabled.

If you would like to use the Linux GRO/LRO for any reason, first the t4_tom kernel module needs to be removed from kernel module list. Please note you might need to reboot your system.

After removing the t4_tom module, you can use ethtool to check the status of current

GRO/LRO settings, for example:

[root@host]# ethtool -k eth6

Offload parameters for eth6: rx-checksumming: on tx-checksumming: on scatter-gather: on tcp-segmentation-offload: on udp-fragmentation-offload: off generic-segmentation-offload: on generic-receive-offload: on large-receive-offload: off

Now the generic-receive-offload option is on. This means GRO is enabled. Please note that there are two offload options here: generic-receive-offload and large-receiveoffload . This is because on this Linux system (RHEL6.0), the kernel supports both GRO and

LRO. As mentioned earlier, GRO is always the preferred option when both of them are present.

On other systems LRO might be the only available option. Then ethtool could be used to switch LRO on and off as well.

When Linux’s GRO is enabled, Chelsio’s T5/T4 driver provides two GRO-related statistics. They are displayed using the following command:

[root@host]# ethtool -S eth6

...

GROPackets : 0

GROMerged : 897723

...

Chelsio T5/T4 Unified Wire For Linux Page 56

Chapter II. Network (NIC/TOE)

GROPackets is the number of held packets. Those are candidate packets held by the kernel to be processed individually or to be merged to larger packets. This number is usually zero.

GROMerged is the number of packets that merged to larger packets. Usually this number increases if there is any continuous traffic stream present. ethtool can also be used to switch off the GRO/LRO options when necessary:

[root@host]# ethtool -K eth6 gro off

[root@host]# ethtool -k eth6

Offload parameters for eth6: rx-checksumming: on tx-checksumming: on scatter-gather: on tcp-segmentation-offload: on udp-fragmentation-offload: off generic-segmentation-offload: on generic-receive-offload: off large-receive-offload: off

The output above shows a disabled GRO.

Chelsio T5/T4 Unified Wire For Linux Page 57

Chapter III. Virtual Function Network (vNIC)

III. Virtual Function Network (vNIC)

Chelsio T5/T4 Unified Wire For Linux Page 58

Chapter III. Virtual Function Network (vNIC)

1. Introduction

The ever increasing network infrastructure of IT enterprises has lead to a phenomenal increase in maintenance and operational costs. IT managers are forced to acquire more physical servers and other data center resources to satisfy storage and network demands. To solve the Network and I/O overhead, users are opting for server virtualization which consolidates I/O workloads onto lesser physical servers thus resulting in efficient, dynamic and economical data center environments. Other benefits of Virtualization include improved disaster recovery, server portability, cloud computing, Virtual Desktop Infrastructure (VDI), etc.

Chelsio’s T5 and T4 Unified Wire family of Adapters deliver increased bandwidth, lower latency and lower power with virtualization features to maximize cloud scaling and utilization. The adapters support SFP+ and 10BASE-T media. The adapters also provide full support for PCI-

SIG SR-IOV to improve I/O performance on a virtualized system. User can configure up to 128

Virtual and 8 Physical functions (with 4 PFs as SR-IOV capable) along with 336 virtual MAC addresses.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with the Chelsio vNIC driver:

T520-LL-CR

T520-CR

T580-LP-CR

T420-CR

T440-CR

T422-CR

T420-SO-CR

T404-BT

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

Chelsio T5/T4 Unified Wire For Linux Page 59

Chapter III. Virtual Function Network (vNIC)

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the vNIC driver is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04, 3.2.0-23

Kernel.org linux-2.6.35

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

* Limited QA performed.

Chelsio T5/T4 Unified Wire For Linux Page 60

Chapter III. Virtual Function Network (vNIC)

2. Software/Driver Loading

The vNIC driver must be loaded or unloaded on the Guest OS by the root user. Any attempt to load the driver as a regular user will fail.

2.1.

Loading the driver

To load the driver, run the following command:

[root@host]# modprobe cxgb4vf

Chelsio T5/T4 Unified Wire For Linux Page 61

Chapter III. Virtual Function Network (vNIC)

3. Software/Driver Unloading

3.1.

Unloading the driver

To unload the driver, execute the following command:

[root@host]# rmmod cxgb4vf

Chelsio T5/T4 Unified Wire For Linux Page 62

Chapter III. Virtual Function Network (vNIC)

4. Software/Driver Configuration and Fine-tuning

4.1.

Instantiate Virtual Functions

To instantiate Chelsio Virtual Functions, please refer to the Network (NIC/TOE) section ( click here )

Chelsio T5/T4 Unified Wire For Linux Page 63

Chapter IV. iWARP (RDMA)

IV. iWARP (RDMA)

Chelsio T5/T4 Unified Wire For Linux Page 64

Chapter IV. iWARP (RDMA)

1. Introduction

Chelsio’s T5/T4 engine implements a feature rich RDMA implementation which adheres to the

IETF standards with optional markers and MPA CRC-32C.

The iWARP RDMA operation benefits from the virtualization, traffic management and QoS mechanisms provided by T5/T4 engine. It is possible to ACL process iWARP RDMA packets. It is also possible to rate control the iWARP traffic on a per-connection or per-class basis, and to give higher priority to QPs that implement distributed locking mechanisms. The iWARP operation also benefits from the high performance and low latency TCP implementation in the offload engine.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with Chelsio iWARP driver:

T520-LL-CR

T520-CR

T580-LP-CR

T540-CR

T420-CR

T440-CR

T422-CR

T404-BT

T440-LP-CR

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the iWARP driver is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1),2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27 *

Chelsio T5/T4 Unified Wire For Linux Page 65

Chapter IV. iWARP (RDMA)

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04, 3.2.0-23

Kernel.org linux-2.6.35

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work

*Limited QA performed

Chelsio T5/T4 Unified Wire For Linux Page 66

Chapter IV. iWARP (RDMA)

2. Software/Driver Loading

2.1.

Compiling and Loading iWARP driver

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail.

Change your current working directory to driver package directory and run the following command :

[root@host]# make

[root@host]# make install

To load the iWARP driver we need to load the NIC driver and core RDMA drivers first. Run the following commands:

[root@host]# modprobe cxgb4

[root@host]# modprobe iw_cxgb4

[root@host]# modprobe rdma_ucm

Set Chelsio driver option for MPI connection changes. Run the following command on all systems:

[root@host]# echo 1 > /sys/module/iw_cxgb4/parameters/peer2peer

Chelsio T5/T4 Unified Wire For Linux Page 67

Chapter IV. iWARP (RDMA)

3. Software/Driver Unloading

To unload the iWARP driver, run the following command:

[root@host]# rmmod iw_cxgb4

Chelsio T5/T4 Unified Wire For Linux Page 68

Chapter IV. iWARP (RDMA)

4. Software/Driver Configuration and Fine-tuning

4.1.

Testing connectivity with ping and rping

Load the NIC, iWARP & core RDMA modules as mentioned in Software/Driver Loading section. After which, you will see one or two ethernet interfaces for the T5/T4 device. Configure them with an appropriate ip address, netmask, etc. You can use the Linux ping command to test basic connectivity via the T5/T4 interface. To test RDMA, use the rping command that is included in the librdmacm-utils RPM:

Run the following command on the server machine:

[root@host]# rping -s -a server_ip_addr -p 9999

Run the following command on the client machine:

[root@host]# rping -c –Vv -C10 -a server_ip_addr -p 9999

You should see ping data like this on the client: ping data: rdma-ping-0: ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqr ping data: rdma-ping-1: BCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrs ping data: rdma-ping-2: CDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrst ping data: rdma-ping-3: DEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstu ping data: rdma-ping-4: EFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuv ping data: rdma-ping-5: FGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvw ping data: rdma-ping-6: GHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwx ping data: rdma-ping-7: HIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxy ping data: rdma-ping-8: IJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz ping data: rdma-ping-9: JKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyzA client DISCONNECT EVENT...

#

Chelsio T5/T4 Unified Wire For Linux Page 69

Chapter IV. iWARP (RDMA)

4.2.

Enabling various MPIs

4.2.1. DAPL Library configuration for Intel MPI and Platform MPI

You must set the iWARP (iw_cxgb4) module option peer2peer=1 on all systems. This can be done by writing to the /sys/module/ file system during boot.

E.g.: For RHEL 5 and SLES platforms use following commands:

[root@host]# echo 1 > /sys/module/iw_cxgb4/parameters/peer2peer

OR

You can add the following line to /etc/modprobe.conf

to set the option at module load time: options iw_cxgb4 peer2peer=1

To run Intel MPI over RDMA interface, DAPL 2.0 should be set up as follows:

Enable the Chelsio device by adding an entry at the beginning of the / etc/dat.conf

file for the

Chelsio interface. For instance, if your Chelsio interface name is eth2, then the following line adds a DAT version 2.0 device named

“chelsio2" for that interface: chelsio2 u2.0 nonthreadsafe default libdaplofa.so.2 dapl.2.0 "eth2 0" ""

4.2.2. Setting shell for Remote Login

User needs to set up authentication on the user account on all systems in the cluster to allow user to remotely logon or executing commands without password.

Quick steps to set up user authentication: i. Change to user home directory

[root@host]# cd ii. Generate authentication key

[root@host]# ssh-keygen -t rsa

Chelsio T5/T4 Unified Wire For Linux Page 70

Chapter IV. iWARP (RDMA) iii. Hit [Enter] upon prompting to accept default setup and empty password phrase iv. Create authorization file

[root@host]# cd .ssh

[root@host]# cat *.pub > authorized_keys

[root@host]# chmod 600 authorized_keys v. Copy directory .ssh to all systems in the cluster

[root@host]# cd

[root@host]# scp -r /root/.ssh remotehostname-or-ipaddress:

4.2.3. Configuration of various MPIs (Installation and Setup)

Intel-MPI

i. Download latest Intel MPI from the Intel website ii. Copy the license file (.lic file) into l_mpi_p_x.y.z

directory iii. Create machines.LINUX (list of node names) in l_mpi_p_x.y.z

iv. Select advanced options during installation and register the MPI. v. Install software on every node.

[root@host]# ./install.py vi. Set IntelMPI with mpi-selector (do this on all nodes).

[root@host]# mpi-selector --register intelmpi --source-dir

/opt/intel/impi/3.1/bin/

[root@host]# mpi-selector --set intelmpi vii. Edit .bashrc

and add these lines: export RSH=ssh export DAPL_MAX_INLINE=64 export I_MPI_DEVICE=rdssm:chelsio export MPIEXEC_TIMEOUT=180 export MPI_BIT_MODE=64 viii. Logout & log back in. ix. Populate mpd.hosts

with node names.

Chelsio T5/T4 Unified Wire For Linux Page 71

Chapter IV. iWARP (RDMA)

Note

The hosts in this file should be Chelsio interface IP addresses.

I_MPI_DEVICE=rdssm:chelsio assumes you have an entry in

/etc/dat.conf

named chelsio .

MPIEXEC_TIMEOUT value might be required to increase if heavy traffic is going across the systems. x. Contact Intel for obtaining their MPI with DAPL support. xi. To run Intel MPI applications: mpdboot -n -r ssh --ncpus= mpiexec -ppn -n 2 /opt/intel/impi/3.1/tests/IMB-3.1/IMB-MPI1

The performance is best with NIC MTU set to 9000 bytes.

Platform MPI Setup

i. After you have installed Platform MPI RPM and obtained the license from Platform computing, unset the previously selected MPI as mpi-selector –-unset <previous MPI> ii. Edit .bashrc

and add these lines: export MPI_ROOT=/opt/platform_mpi/ export PATH=$MPI_ROOT/bin:/opt/bin:$PATH export MANPATH=$MANPATH:$MPI_ROOT/share/man iii. Log out and log back in. iv. To run Platform MPI over RDMA interface, DAPL 2.0 should be set up as follows:

Enable the Chelsio device by adding an entry at the beginning of the / etc/dat.conf

file for the Chelsio interface. For instance, if your Chelsio interface name is eth2, then the following line adds a DAT 2.0 device named "chelsio2" for that interface: chelsio2 u2.0 nonthreadsafe default libdaplofa.so.2 dapl.2.0 "eth2 0" ""

Chelsio T5/T4 Unified Wire For Linux Page 72

Chapter IV. iWARP (RDMA) v. Start the License Manager and run the test as: mpirun -prot -e DAPL_MAX_INLINE=64 -UDAPL -hostlist <name of hosts separated by comma> -np 4 /opt/platform_mpi/tests/IMB-3.2/IMB-MPI1

Open MPI (Installation and Setup)

Open MPI iWARP support is only available in Open MPI version 1.3 or greater.

Open MPI will work without any specific configuration via the openib btl. Users wishing to performance tune the configurable options may wish to inspect the receive queue values. Those can be found in the "Chelsio T4" section of mca-btl-openib-device-params.ini

. Follow the steps mentioned below to install and configure Open MPI. i. If not alreay done, install mpi-selector tool. ii. Download the latest stable/feature version of openMPI from OpenMPI website

, http://www.open-mpi.org

iii. Untar and change your current working directory to openMPI package directory. iv. Configure and install as:

[root@host]# ./configure --with-openib=/usr CC=gcc CXX=g++ F77=gfortran

FC=gfortran --enable-mpirun-prefix-by-default --prefix=/usr/mpi/gcc/openmpix.y.z/ --with-openib-libdir=/usr/lib64/ --libdir=/usr/mpi/gcc/openmpix.y.z/lib64/ --with-contrib-vt-flags=--disable-iotrace

[root@host]# make

[root@host]# make install

The above step will install openMPI in /usr/mpi/gcc/openmpi-x.y.z/

Chelsio T5/T4 Unified Wire For Linux Page 73

Chapter IV. iWARP (RDMA) v. Next, create a shell script , mpivars.csh, with the following entry:

# path if ("" == "`echo $path | grep /usr/mpi/gcc/openmpi-x.y.z/bin`") then

set path=(/usr/mpi/gcc/openmpi-x.y.z/bin $path) endif

# LD_LIBRARY_PATH if ("1" == "$?LD_LIBRARY_PATH") then

if ("$LD_LIBRARY_PATH" !~ */usr/mpi/gcc/openmpi-x.y.z/lib64*) then

setenv LD_LIBRARY_PATH /usr/mpi/gcc/openmpix.y.z/lib64:${LD_LIBRARY_PATH}

endif else

setenv LD_LIBRARY_PATH /usr/mpi/gcc/openmpi-x.y.z/lib64 endif

# MPI_ROOT setenv MPI_ROOT /usr/mpi/gcc/openmpi-x.y.z vi. Simlarly, create another shell script, mpivars.sh, with the following entry:

# PATH if test -z "`echo $PATH | grep /usr/mpi/gcc/openmpi-x.y.z/bin`"; then

PATH=/usr/mpi/gcc/openmpi-x.y.z/bin:${PATH}

export PATH fi

# LD_LIBRARY_PATH if test -z "`echo $LD_LIBRARY_PATH | grep /usr/mpi/gcc/openmpi-

x.y.z/lib64`"; then

LD_LIBRARY_PATH=/usr/mpi/gcc/openmpi-

x.y.z/lib64${LD_LIBRARY_PATH:+:}${LD_LIBRARY_PATH}

export LD_LIBRARY_PATH fi

# MPI_ROOT

MPI_ROOT=/usr/mpi/gcc/openmpi-x.y.z export MPI_ROOT vii. Next, copy the two files created in steps (v) and (vi) to /usr/mpi/gcc/openmpi-x.y.z/bin and

/usr/mpi/gcc/openmpi-x.y.z/etc

Chelsio T5/T4 Unified Wire For Linux Page 74

Chapter IV. iWARP (RDMA) viii. Register OpenMPi with MPI-selector:

[root@host]# mpi-selector --register openmpi --source-dir

/usr/mpi/gcc/openmpi-x.y.z/bin ix. Verify if it is listed in mpi-selector:

[root@host]# mpi-selector --l x. Set OpenMPI:

[root@host]# mpi-selector --set openmpi –yes xi. Logut and log back in.

MVAPICH2 (Installation and Setup)

i. Download the latest MVAPICH2 software package from http://mvapich.cse.ohio-state.edu/ ii. Untar and change your current working directory to MVAPICH2 package directory. iii. Configure and install as:

[root@host]# ./configure --prefix=/usr/mpi/gcc/mvapich2-x.y/ --withdevice=ch3:mrail --with-rdma=gen2 --enable-shared --with-iblibpath=/usr/lib64/ -enable-rdma-cm --libdir=/usr/mpi/gcc/mvapich2-x.y/lib64

[root@host]# make

[root@host]# make install

The above step will install MVAPICH2 in /usr/mpi/gcc/mvapich2-x.y/

Chelsio T5/T4 Unified Wire For Linux Page 75

Chapter IV. iWARP (RDMA) iv. Next, create a shell script , mpivars.csh, with the following entry:

# path if ("" == "`echo $path | grep /usr/mpi/gcc/mvapich2-x.y/bin`") then

set path=(/usr/mpi/gcc/mvapich2-x.y/bin $path) endif

# LD_LIBRARY_PATH if ("1" == "$?LD_LIBRARY_PATH") then

if ("$LD_LIBRARY_PATH" !~ */usr/mpi/gcc/mvapich2-x.y/lib64*) then

setenv LD_LIBRARY_PATH /usr/mpi/gcc/mvapich2x.y/lib64:${LD_LIBRARY_PATH}

endif else

setenv LD_LIBRARY_PATH /usr/mpi/gcc/mvapich2-x.y/lib64 endif

# MPI_ROOT setenv MPI_ROOT /usr/mpi/gcc/mvapich2-x.y v. Simlarly, create another shell script, mpivars.sh, with the following entry:

# PATH if test -z "`echo $PATH | grep /usr/mpi/gcc/ mvapich2-x.y/bin`"; then

PATH=/usr/mpi/gcc/mvapich2-x.y/bin:${PATH}

export PATH fi

# LD_LIBRARY_PATH if test -z "`echo $LD_LIBRARY_PATH | grep /usr/mpi/gcc/mvapich2x.y/lib64`"; then

LD_LIBRARY_PATH=/usr/mpi/gcc/mvapich2- x.y/lib64${LD_LIBRARY_PATH:+:}${LD_LIBRARY_PATH}

export LD_LIBRARY_PATH fi

# MPI_ROOT

MPI_ROOT=/usr/mpi/gcc/mvapich2-x.y export MPI_ROOT

Chelsio T5/T4 Unified Wire For Linux Page 76

Chapter IV. iWARP (RDMA) vi. Next, copy the two files created in steps (iv) and (v) to /usr/mpi/gcc/mvapich2-x.y/bin and

/usr/mpi/gcc/mvapich2-x.y/etc vii. Add the following entries in .bashrc file: export MVAPICH2_HOME=/usr/mpi/gcc/mvapich2-x.y/ export MV2_USE_IWARP_MODE=1 export MV2_USE_RDMA_CM=1 viii. Register MPI:

[root@host]# mpi-selector --register mvapich2 --source-dir

/usr/mpi/gcc/mvapich2-x.y/bin/ xii. Verify if it is listed in mpi-selector:

[root@host]# mpi-selector --l ix. Set MVAPICH2:

[root@host]# mpi-selector --set mvapich2 –yes x. Logut and log back in. xi. Populate mpd.hosts

with node names. xii. On each node, create /etc/mv2.conf with a single line containing the IP address of the local

T4/T5 interface. This is how MVAPICH2 picks which interface to use for RDMA traffic.

4.2.4. Building MPI tests

i. Download

Intel’s MPI Benchmarks from http://software.intel.com/en-us/articles/intel-mpibenchmarks ii. Untar and change your current working directory to src directory.

Chelsio T5/T4 Unified Wire For Linux Page 77

Chapter IV. iWARP (RDMA) iii. Edit make_mpich file and set MPI_HOME variable to the MPI which you want to build the benchmarks tool against. For example, in case of openMPI-1.6.4 set the variable as:

MPI_HOME=/usr/mpi/gcc/openmpi-1.6.4/ iv. Next, build and install the benchmarks using:

[root@host]# gmake -f make_mpich

The above step will install IMB-MPI1, IMB-IO and IMB-EXT benchmarks in the current working directory (i.e. src). v. Change your working directory to the MPI installation directory. In case of OpenMPI, it will be /usr/mpi/gcc/openmpi-x.y.z/ vi. Create a directory called tests and then another directory called imb under tests. vii. Copy the benchmarks built and installed in step (iv) to the imb directory. viii. Follow steps (v), (vi) and (vii) for all the nodes.

4.2.5. Running MPI applications

Run Open MPI application as: mpirun --host node1,node2 -mca btl openib,sm,self /usr/mpi/gcc/openmpix.y.z/tests/imb/IMB-MPI1

Note

For OpenMPI/RDMA clusters with node counts greater than or equal to 8 nodes, and process counts greater than or equal to 64, you may experience the following RDMA address resolution error when running MPI jobs with the default OpenMPI settings:

The RDMA CM returned an event error while attempting to make a connection.

This type of error usually indicates a network configuration error.

Local host: core96n3.asicdesigners.com

Local device: Unknown

Error name: RDMA_CM_EVENT_ADDR_ERROR

Peer: core96n8

Chelsio T5/T4 Unified Wire For Linux Page 78

Chapter IV. iWARP (RDMA)

Workaround: Increase the OpenMPI rdma route resolution timeout. The default is 1000, or

1000ms. Increase it to 30000 with this parameter:

--mca btl_openib_connect_rdmacm_resolve_timeout 30000

Run MVAPICH2 application as : mpirun_rsh -ssh -np 8 -hostfile mpd.hosts $MVAPICH2_HOME/tests/imb/IMB-MPI1

4.3.

Setting up NFS-RDMA

4.3.1. Starting NFS-RDMA

Server-side settings

Follow the steps mentioned below to set up an NFS-RDMA server. i. Make entry in /etc/exports file for the directories you need to export using NFS-RDMA on server as:

/share/rdma

/share/rdma1

*(fsid=0,async,insecure,no_root_squash)

*(fsid=1,async,insecure,no_root_squash)

Note that for each directory you export, you should have DIFFERENT fsid’s. ii. Load the iwarp modules and make sure peer2peer is set to 1. iii. Load xprtrdma and svcrdma modules as:

[root@host]# modprobe xprtrdma

[root@host]# modprobe svcrdma iv. Start the nfs service as:

[root@host]# service nfs start

Chelsio T5/T4 Unified Wire For Linux Page 79

Chapter IV. iWARP (RDMA)

All services in NFS should start without errors. v. Now we need to edit the file portlist in the path /proc/fs/nfsd/ vi. Include the rdma port 2050 into this file as:

[root@host]# echo rdma 2050 > /proc/fs/nfsd/portlist vii. Run exportfs to make local directories available for Network File System (NFS) clients to mount.

[root@host]# exportfs

Now the NFS-RDMA server is ready.

Client-side settings

Follow the steps mentioned below at the client side. i. Load the iwarp modules and make sure peer2peer is set to 1. Make sure you are able to ping and ssh to the server Chelsio interface through which directories will be exported. ii. Load the xprtrdma module.

[root@host]# modprobe xprtrdma iii. Run the showmount command to show all directories from server as:

[root@host]# showmount –e <server-chelsio-ip> iv. Once the exported directories are listed, mount them as:

[root@host]# mount.nfs <serverip>:<directory> <mountpoint-on-client> -o vers=3,rdma,port=2050,wsize=65536,rsize=65536

Chelsio T5/T4 Unified Wire For Linux Page 80

Chapter V. WD-UDP

V. WD-UDP

Chelsio T5/T4 Unified Wire For Linux Page 81

Chapter V. WD-UDP

1. Introduction

Chelsio WD-UDP (Wire Direct-User Datagram Protocol) with Multicast is a user-space UDP stack with Multicast address reception and socket acceleration that enables users to run their existing UDP socket applications unmodified.

It features software modules that enable direct wire access from user space to the Chelsio

T5/T4 network adapter with complete bypass of the kernel, which results in an ultra-low, three microsecond deterministic latency 10Gb Ethernet solution for high frequency trading and other delay-sensitive applications.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with Chelsio WD-

UDP driver:

T520-LL-CR

T520-CR

T580-LP-CR

T540-CR

T420-CR

T440-CR

T422-CR

T404-BT

T440-LP-CR

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the WD-UDP driver is available for the following versions:

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1),2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Chelsio T5/T4 Unified Wire For Linux Page 82

Chapter V. WD-UDP

Ubuntu 12.04 , 3.2.0-23

Kernel.org linux-2.6.35

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

*Limited QA performed

Chelsio T5/T4 Unified Wire For Linux Page 83

Chapter V. WD-UDP

2. Software/Driver Compiling and Loading

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail. i. Change your current working directory to driver package directory & run the following command :

[root@host]# make

[root@host]# make install ii. RDMA core modules from the OFED package should be loaded before proceeding. To load the WD-UDP driver, use the following commands which will automatically load RDMA core modules:

[root@host]# modprobe cxgb4

[root@host]# modprobe iw_cxgb4

[root@host]# modprobe rdma_ucm

Chelsio T5/T4 Unified Wire For Linux Page 84

Chapter V. WD-UDP

3. Software/Driver Unloading

To unload the WD-UDP driver, run the following command:

[root@host]# rmmod iw_cxgb4

Chelsio T5/T4 Unified Wire For Linux Page 85

Chapter V. WD-UDP

4. Software/Driver Configuration and Fine-tuning

4.1.

Accelerating UDP Socket communications

The libcxgb4_sock library is a LD_PRELOAD-able library that accelerates UDP Socket communications transparently and without recompilation of the user application. This section describes how to use libcxgb4_sock.

By preloading libcxgb4_sock, all sockets created by the application are intercepted and possibly accelerated based on the user’s configuration. Once accelerated, data for the UDP endpoint are transmitted or received via HW queues allocated specifically for the accelerated endpoint, bypassing the kernel, the host networking stack and sockets framework, and enabling ultra-low latency and high bandwidth utilization.

Due to HW resource limitations, only a small number of queues can be allocated for UDP acceleration. Therefore only performance critical UDP applications should use libcxgb4_sock.

Only 64 IPv4 UDP sockets / 28 IPv6 UDP Sockets can be accelerated per Chelsio T5/T4 device used.

4.1.1. Application Requirements

Certain application behavior is not supported by libcxb4_sock in this release. If your application does any of the following, it will not work with libcxgb4_sock:

Calling fork() after creating UDP sockets and using the UDP socket in the child process.

Using multiple threads on a single UDP socket without serialization. For instance, having one thread sending concurrently with another thread receiving. If your application does this, you need to serialize these paths with a spin or mutex lock.

Only 1 UDP endpoint is allowed to bind to a given port per host. So if you have multiple processes on the same host binding to the same UDP port number, you cannot use

libcxgb4_sock.

Applications must have root privileges to use libcxgb4_sock.

Applications requiring bonded T5/T4 interfaces are not currently supported.

The performance benefit observed with libcxgb4_sock will vary based on your application’s behavior. While all UDP IO is handled properly, only certain datagrams are accelerated. Non accelerated IO is handled by libcxgb4_sock via the host networking stack seamlessly. Both

Unicast and Multicast datagrams can be accelerated, but the datagrams must meet the following criteria:

Non fragmented. In other words, they fit in a single IP datagram that is <= the T5/T4 device

MTU.

Routed through the T5/T4 acceleration device. If the ingress datagram arrives via a device other than the T5/T4 acceleration device, then it will not utilize the acceleration path. On

Chelsio T5/T4 Unified Wire For Linux Page 86

Chapter V. WD-UDP egress, if the destination IP address will not route out via the T5/T4 device, then it too will not be accelerated.

4.1.2. Using libcxgb4_sock

The libcxgb4_sock library utilizes the Linux RDMA Verbs subsystem, and thus requires the

RDMA modules be loaded. Ensure that your systems load the iw_cxgb4 and rdma_ucm modules:

[root@host]# modprobe iw_cxgb4

[root@host]# modprobe rdma_ucm

To preload libcxgb4_sock, use one of the commands mentioned below when starting your application:

[root@host]# LD_PRELOAD=libcxgb4_sock.so <pathto>/your_application

OR

[root@host]# wdload <pathto>/your_application

Note

If you encounter error while using wdload on kernels built on RHEL 5.x distribution, then run the above command as :

[root@host]# NUMA=0 wdload <pathto>/your_application

If you want to use libcxgb4_sock ’s debug capabilities, make the following entry in the

/etc/syslog.conf file:

*.debug /var/log/cxgb4.log then, restart the service:

[root@host]# /etc/init.d/syslog restart

Chelsio T5/T4 Unified Wire For Linux Page 87

Chapter V. WD-UDP

Now, preload libcxgb4_sock_debug using the command mentioned below when starting your application: root@host]# LD_PRELOAD=libcxgb4_sock_debug.so CXGB4_SOCK_DEBUG=-1

<pathto>/your_application

In addition to preloading libcxgb4_sock.so

, you must create a configuration file that defines which UDP endpoints should be accelerated, their vlan and priority if any, as well as which

T5/T4 interface/port should be used. The file /etc/libcxgb4_sock.conf

contains these endpoint entries. Create this file on all systems using libcxgb4_sock. Here is the syntax:

#

# Syntax:

#

# endpoint {attributes} ...

# where attributes include:

#

#

#

# interface = interface-name port = udp-port-number vlan = vlan-id priority = vlan-priority

#

# e.g.

# endpoint {

#

#

# } interface=eth2.5 port = 8000 vlan = 5 priority=1

# endpoint { interface=eth2 port=9999}

#

# endpoints that bind to port 0 (requesting the host allocate a port)

# can be accelerated with port=0:

#

# endpoint {interface=eth1 port=0}

#

Assume your T5/T4 interface is eth2. To accelerate all applications that preload libcxgb4_sock using eth2, you only need one entry in /etc/libcxgb4_sock.conf

: endpoint {interface=eth2 port=0}

Chelsio T5/T4 Unified Wire For Linux Page 88

Chapter V. WD-UDP

If you have eth2 and eth3 configured for example, you can define certain endpoints to eth2 and others to eth3: endpoint {interface=eth2 port=9999} endpoint {interface=eth3 port=8888}

For VLAN support, create your VLANs using the normal OS service (like vconfig, for example), then add entries to define the VLAN and priority for each endpoint to be accelerated: endpoint {interface = eth2.5 port=10000} endpoint {interface = eth2.7 priority=3 port=9000}

The maximum MTU supported for a 4K host page size is 4032. If the MTU is > 4032, follow the steps mentioned below:

[root@host]# echo 1024 > /proc/sys/vm/nr_hugepages

[root@host]# CXGB4_SOCK_HUGE_PAGES=1 wdload <pathto>/your_application

Note

In order to offload IPv6 UDP sockets, please select “low latency networking” as

T5/T4 configuration tuning option during installation.

4.1.3. Using per process config file

The libcxgb4_sock library utilizes the Linux RDMA Verbs subsystem, and thus requires the

RDMA modules be loaded. Ensure that your systems load the iw_cxgb4 and rdma_ucm modules:

[root@host]# modprobe iw_cxgb4

[root@host]# modprobe rdma_ucm

For individual interfaces, create a config file for “accelerate all” configuration on both nodes:

[root@host]# cat /root/config1 endpoint {interface=eth1 port=0}

Chelsio T5/T4 Unified Wire For Linux Page 89

Chapter V. WD-UDP

To preload libcxb4_sock and load the config file, use the LD_PRELOAD environment variable along with the config file when starting your application:

[root@host]# CXGB4_SOCK_CFG=<path to config file>

LD_PRELOAD=libcxgb4_sock.so <pathto>/your_application

4.1.4. Example with hpcbench/udp

The udp benchmark from the hpcbench suite can be used to show the benefits of libcxgb4_sock. The hpcbench suite can be found at:

Source : http://hpcbench.sourceforge.net/index.html

Sample: http://hpcbench.sourceforge.net/udp.html

The nodes in this example, r9 and r10, have T5/T4 eth1 configured and the ports are connected point-to-point.

[root@r9 ~]# ifconfig eth1|grep inet

inet addr:192.168.2.111 Bcast:192.168.2.255 Mask:255.255.255.0

inet6 addr: fe80::7:4300:104:465a/64 Scope:Link

[root@r9 ~]#

[root@r10 ~]# ifconfig eth1|grep inet

inet addr:192.168.2.112 Bcast:192.168.2.255 Mask:255.255.255.0

inet6 addr: fe80::7:4300:104:456a/64 Scope:Link

[root@r10 ~]#

For this benchmark, we need a simple “accelerate all” configuration on both nodes:

[root@r9 ~]# cat /etc/libcxgb4_sock.conf endpoint {interface=eth1 port=0}

[root@r9 ~]#

[root@r10 ~]# cat /etc/libcxgb4_sock.conf endpoint {interface=eth1 port=0}

[root@r10 ~]#

On R10, we run udpserver on port 9000 without libcxgb4_sock preloaded, and on port 90001 with preload:

Chelsio T5/T4 Unified Wire For Linux Page 90

Chapter V. WD-UDP

[root@r10 ~]# /usr/local/src/hpcbench/udp/udpserver -p 9000 &

[1] 11453

[root@r10 ~]# TCP socket listening on port [9000]

[root@r10 ~]# LD_PRELOAD=libcxgb4_sock.so

/usr/local/src/hpcbench/udp/udpserver -p 9001 &

[2] 11454

[root@r10 ~]# TCP socket listening on port [9001]

[root@r10 ~]#

Then on r9, we run udptest to port 9000 to see the host stack UDP latency:

[root@r9 ~]# /usr/local/src/hpcbench/udp/udptest -r 5 -a -h 192.168.1.112 -p

9000

Running the same test with libcxgb4_sock:

[root@r9 ~]# LD_PRELOAD=libcxgb4_sock.so /usr/local/src/hpcbench/udp/udptest

-r 5 -a -h 192.168.1.112 -p 9001

4.1.5. Performance tuning on 2.6.18 kernel

To get better performance with WD-UDP using the 2.6.18 kernel, load the iw_cxgb4 modules with the ocqp_support=0 parameter. For example, modprobe iw_cxgb4 ocqp_support=0

4.1.6. Determining if the application is being offloaded

To see if the application is being offloaded, open a window on one of the machines, and run tcpdump against the Chelsio interface. If you see minimal UDP output on the interface, then the

UDP traffic is being properly offloaded.

Chelsio T5/T4 Unified Wire For Linux Page 91

Chapter VI. WD-TOE

VI. WD-TOE

Chelsio T5/T4 Unified Wire For Linux Page 92

Chapter VI. WD-TOE

1. Introduction

Chelsio WD-TOE (Wire Direct-Transmission Control Protocol) with a user-space TCP stack enables users to run their existing TCP socket applications unmodified.

It features software modules that enable direct wire access from user space to the Chelsio

T5/T4 network adapter with complete bypass of the kernel, which results in a low latency 10Gb

Ethernet solution for high frequency trading and other delay-sensitive applications.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with the WD-TOE driver.

T520-LL-CR

T520-CR

T580-LP-CR

T420-CR

T440-CR

T422-CR

T420-SO-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the WD-TOE driver is available for the following version(s):

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5*

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5*

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6*

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7*

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Kernel.org linux-3.4

Chelsio T5/T4 Unified Wire For Linux Page 93

Chapter VI. WD-TOE

Kernel.org linux-3.6

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

*Limited QA performed

Chelsio T5/T4 Unified Wire For Linux Page 94

Chapter VI. WD-TOE

2. Software/Driver Loading

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail.

Run the following commands to load the driver:

[root@host]# modprobe cxgb4

[root@host]# modprobe t4_tom

Chelsio T5/T4 Unified Wire For Linux Page 95

Chapter VI. WD-TOE

3. Software/Driver Unloading

Reboot the system to unload the driver.

Chelsio T5/T4 Unified Wire For Linux Page 96

Chapter VI. WD-TOE

4. Software/Driver Configuration and Fine-tuning

4.1.

Running the application

To run an application with WD-TOE library, use the following command:

[root@host]# PROT=TCP wdload <path to application>

Example: To run Netperf application with WD-TOE library. i. Start netserver at the PEER, using the following command:

[root@host]# PROT=TCP wdload netserver –D -4 ii. On the Test Machine, run the following command to run netperf application.

[root@host]# PROT=TCP wdload netperf -H <PEER_IP> -t TCP_RR -l 10

Chelsio T5/T4 Unified Wire For Linux Page 97

Chapter VII. iSCSI PDU Offload Target

VII. iSCSI PDU Offload Target

Chelsio T5/T4 Unified Wire For Linux Page 98

Chapter VII. iSCSI PDU Offload Target

1. Introduction

This section describes how to install and configure iSCSI PDU Offload Target software for use as a key element in your iSCSI SAN. The software runs on Linux-based systems that use

Chelsio or non-Chelsio based Ethernet adapters. However to guarantee highest performance,

Chelsio recommends using Chelsio adapters. Chelsio’s adapters include offerings that range from stateless offload adapters (regular NIC) to the full line of TCP/IP Offload Engine (TOE) adapters.

The software implements RFC 3720, the iSCSI standard of the IETF. The software has been fully tested for compliance to that RFC and others and it has been exhaustively tested for interoperability with the major iSCSI vendors.

The software implements most of the iSCSI protocol in software running in kernel mode on the host with the remaining portion, which consists of the entire fast data path, in hardware when used with Chelsio’s TOE adapters. When standard NIC Adapters are used the entire iSCSI protocol is executed in software.

The performance of this iSCSI stack is outstanding and when used with Chelsio’s hardware it is enhanced further. Because of the tight integration with Chelsio’s TOE adapters, this software has a distinct performance advantage over the regular NIC. The entire solution, which includes this software, Chelsio TOE hardware, an appropriate base computer system – including a high end disk subsystem, has industry leading performance. This can be seen when the entire solution is compared to others based on other technologies currently available on the market in terms of throughput and IOPS.

1.1.

Features

Chels io’s iSCSI driver stack supports the iSCSI protocol in the Target mode. From henceforth

“iSCSI Software Entity” term refers to the iSCSI target.

The Chelsio iSCSI PDU Offload Target software provides the following high level features:

Expanded NIC Support

Chelsio TCP Offload Engine (TOE) Support

 T5/T4 Based HBAs (T5/T4xx Series cards)

Chelsio Terminator ASIC Support

Offloads iSCSI Fast Data Path with Direct Data Placement (DDP)

Offloads iSCSI Header and Data Digest Calculations

Offload Speeds at 1 Gb and 10 Gb

Offloads TCP/IP for NAS simultaneously with iSCSI

Target Specific features

Chelsio T5/T4 Unified Wire For Linux Page 99

Chapter VII. iSCSI PDU Offload Target

Full compliance with RFC 3720

Error Recovery Level 0 (ERL 0)

CHAP Support, including Mutual Authentication

Internet Storage Name Service (iSNS) Client

Target Access Control List (ACL)

Multiple Connections per Session

Multiple Targets

Multiple LUNs per Target

Multi Path I/O (MPIO)

Greater than 2 TB Disk Support

Reserve / Release for Microsoft Cluster© Support

Persistent Reservation

Dynamic LUN Resizing

iSCSI Target Redirection

Multiple Target device types

 Block

 Virtual Block (LVM, Software RAID, EVMS, etc.)

 Built in RAM Disk

 Built in zero copy RAM Disk

Supports iSCSI Boot Initiators

An Intuitive and Feature Rich Management CLI

This chapter will cover these features in detail.

1.2.

Hardware Requirements

1.2.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with iSCSI PDU

Offload Target software:

T520-LL-CR

T520-CR

T580-LP-CR

T540-CR

T420-CR

T440-CR

T422-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

Chelsio T5/T4 Unified Wire For Linux Page 100

Chapter VII. iSCSI PDU Offload Target

T420-CX

1.2.2. Adapter Requirements

The Chelsio iSCSI PDU Offload Target software can be used with or without hardware protocol offload technology. There are four modes of operation using the iSCSI PDU Offload Target software on Ethernet-based adapters:

Regular NIC – The software can be used in non-offloaded (regular NIC) mode. Please note however that this is the least optimal mode of operating the software in terms of performance.

TOE HW Acceleration – In TOE mode the software takes advantage of the TCP/IP Offload capability of Chelsio’s TOE adapter (without the additional benefit of iSCSI HW acceleration).

iSCSI HW Acceleration – In addition to offloading the TCP/IP protocols in hardware (TOE), this mode also takes a dvantage of Chelsio’s ASIC capability of hardware assisted iSCSI data and header digest calculations as well as using the direct data placement (DDP) feature.

Mix of iSCSI HW Acceleration and TOE HW Acceleration – Using a special software mode the stack can be configured to change between iSCSI acceleration or just TOE acceleration, depending if digests are used or not.

1.2.3. Storage Requirements

When using the Chelsio iSCSI target, a minimum of one hardware storage device is required.

This device can be any of the device types that are supported (block, virtual block, RAM disk).

Multiple storage devices are allowed by configuring the devices to one target or the devices to multiple targets. The software allows multiple targets to share the same device but use caution when doing this.

Chelsio’s implementation of the target iSCSI stack has flexibility to accommodate a large range of configurations. For quick testing, using a RAM Disk as the block storage device works nicely.

For deployment in a production environment a more sophisticated system would be needed.

That typically consists of a system with one or more storage controllers with multiple disk drives attached running software or hardware based RAID.

Chelsio T5/T4 Unified Wire For Linux Page 101

Chapter VII. iSCSI PDU Offload Target

1.3.

Software Requirements

chiscsi_base.ko

is iSCSI non-offload target mode driver and chiscsi_t4.ko

is iSCSI PDU offload target mode driver.

cxgb4, toecore, t4_tom and chiscsi_base modules are required by chiscsi_t4.ko

module to work in offloaded mode. Whereas in iscsi non-offloaded target (NIC) mode, only cxgb4 is needed by chiscsi_base.ko

module.

1.3.1. Linux Requirements

Currently the iSCSI PDU Offload Target software is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5 *

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5 *

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6*

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27*

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Kernel.org linux-2.6.35*

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7*

Other kernel versions have not been tested and are not guaranteed to work.

*Limited QA performed

Note chiscsi_t4 compiles on all Kernel.org Linux Kernels 2.6.18+ Kernels, but only tested and guaranteed to work with above mentioned OSes.

1.3.2. Requirements for Installing the iSCSI Software

When installing the iSCSI software, it is required that the system have Linux kernel source or its headers installed in order to compile the iSCSI software as a kernel module. The source tree may be only header files, as for RHEL5 as an example, or a complete tree. The source tree needs to be configured and the header files need to be compiled. Additionally, the Linux kernel must be configured to use modules.

License Key Requirement

The iSCSI PDU offload software provided in the Unified Wire package is an evaluation copy fully functional for a period of 30-days. After the trial period, a license key file is required to continue using the software . To obtain the key file a binary program called “ chinfotool ” must be run on the host system where the iSCSI software is to be installed. That program generates

Chelsio T5/T4 Unified Wire For Linux Page 102

Chapter VII. iSCSI PDU Offload Target an information file that contains data about the system. The information file must be sent back to

Chelsio where a license key file will be generated. The key file will be sent back and must be installed on the system in order to unlock the software. Details on this mechanism can be found in the next section on software installation.

Chelsio T5/T4 Unified Wire For Linux Page 103

Chapter VII. iSCSI PDU Offload Target

2. Software/Driver Loading

There are two main steps to installing the Chelsio iSCSI PDU Offload Target software. They are:

1. Installing the iSCSI software

– The majority of this section deals with how to install the iSCSI software.

2. Configuring the iSCSI software – Information on configuring the software can be found in a section further into this user ’s guide.

2.1.

Latest iSCSI Software Stack Driver Software

The iSCSI software stack comes bundled in the Chelsio Unified Wire package which can be downloaded from the Chelsio support website ( http://service.chelsio.com

). The license key must be obtained through interaction with the Chelsio sales channel and/or the support organization.

Please contact [email protected]

or [email protected]

for more information.

The iSCSI software is available for use with most installations of the Linux kernel version 2.6.

The software is dependent on the underlying NIC adapter driver and thus the limitation on what version of the 2.6 Linux kernel it can run on is mostly dependent on the NIC driver’s limitations.

The iSCSI module will be installed in the

/lib/modules/<linux_kernel_version>/kernel/drivers/iscsi directory. The modules database will be updated by the installer. This allows the iSCSI module to be located when using the modprobe utility. The actual module chiscsi_t4.ko

can be found inside the package under /chiscsi-5.xxx/kernel.

The iscsictl tool and the chisns tool will be installed in /sbin . The chisns tool starts the iSNS client. The iscsictl tool is provided for configuring and managing the iSCSI targets and iSNS client. It also provides control for iSCSI global settings.

1. Loading the Kernel module

For RHEL distributions, run modprobe as follows:

[root@host]# modprobe chiscsi_t4

For SLES11 sp1, run modprobe as follows:

[root@host]# modprobe chiscsi_t4 --allow-unsupported

Chelsio T5/T4 Unified Wire For Linux Page 104

Chapter VII. iSCSI PDU Offload Target

Note i. While using rpm-tar-ball for installation a. Uninstallation will result into chiscsi.conf file renamed into chiscsi.conf.rpmsave, but if again uninstallation is done then it will lead to overwriting of the old chiscsi.rpmsave file. b. Its advised to take a backup of chiscsi.conf file before you do an uninstallation and installation of new/same unified wire package. As re-installing/upgrading unified-wire package may lead to loss of chiscsi.conf file. ii. Installation/uninstallation using source-tar-ball will neither remove the conf file nor rename it. It will always be intact. However it’s recommended to always take a backup of your configuration file for both methods of installation.

A sample iSCSI configuration file will be installed in /etc/chelsio-iscsi/chiscsi.conf

. This file should be edited using a standard text editor and customized to fit your environment.

2. Set iSCSI service to automatically start at bootup

The chelsio-target service scripts are installed to /etc/init.d

and the parameters for the script are installed at /etc/sysconfig/chiscsi . The script is installed as a system service.

To auto-start the iSCSI target service at a certain runlevel, e.g. runlevel 3, chkconfig can be used on Red Hat and Novell / SuSE based systems as follows:

[root@host]# chkconfig –-level 3 chelsio-target on

The chelsio-target service scripts do basic checks before starting the iSCSI target service, loads the kernel module, and starts all the targets configured by default. It can also be used to stop the targets, and restart/reload configuration.

Note

For the script to execute properly, make sure the following flag is set on all kernel.org kernels.

# CONFIG_MODULE_FORCE_LOAD=y

Chelsio T5/T4 Unified Wire For Linux Page 105

Chapter VII. iSCSI PDU Offload Target

2.2.

Obtaining the iSCSI Software License

A license file is required for each copy of the Chelsio iSCSI PDU Offload Target software installed.

The license is tied to the selected NIC present in the system. The license file will be generated depending on your requirement for a Chelsio iSCSI Target.

2.2.1. Linux Requirements

To obtain an iSCSI license key file, which could be either a production or an evaluation version, please follow the steps below.

1. Prepare License information file

A license information file is required for each license. Run the following command to obtain the license information file:

[root@host]# chinfotool

The chinfotool scans and lists all the NICs in the system, and prompts the user to select one

NIC, to which the keyfile will be tied. At the end, it prints out a summary of the license file.

Note

Only one NIC per system needs to be selected. As long as the selected NIC is in the system, the Chelsio iSCSI software will work on any of the NICs in the system.

For the selected NIC, the following information is gathered:

PCI vendor ID

MAC address

Link speed

Privacy notice: The above information is used by Chelsio Communications strictly for iSCSI license management. Chelsio Communications, Inc. does not share, sell, rent, or lease the information to any third party.

Chelsio T5/T4 Unified Wire For Linux Page 106

Chapter VII. iSCSI PDU Offload Target

[root@host]# chinfotool

Scanning System for network devices....

License key will be tied to any of the following interfaces.

Please select the interface

1. Interface eth1 with INTEL Adapter

Linkspeed is 1000 Mbps/s MAC is: 00:30:48:00:00:10.

2. Interface eth2 with CHELSIO Adapter

Linkspeed is 10000 Mbps/s MAC is: 00:07:43:00:00:10.

ENTER NUMBER:2

*******************************************************

SUMMARY:

Interface with MAC Address 00:07:43:00:00:10 (10000 Mbps) is selected

Please send info file "/etc/xyz_chelsio_infofile" to Chelsio Communications at [email protected]

Chelsio Support will send back the keyfile.

The keyfile unlocks the software as long as the above interface is present

*******************************************************

The generated file hostname_chelsio_infofile needs to be sent to Chelsio at [email protected]

as instructed by the chinfotool .

Note

Please be sure that the selected NIC Adapter is present in the system at all times when using Chelsio’s iSCSI software. If it is removed the license will be invalid and the process of obtaining a new license file will need to be restarted.

That includes using chinfotool to rescan the system and obtaining a new keyfile from Chelsio support.

2. Install the license file received from Chelsio

Once the information file (generated by chinfotool in the previous step) is processed by

Chelsio, the user will receive a license file tailored to the system. Copy the file to

/etc/chelsio-iscsi/chiscsi.key

.

[root@host]# cp -f chiscsi.key /etc/chelsio-iscsi/chiscsi.key

The iSCSI software is now unlocked and is ready to be configured and used.

Chelsio T5/T4 Unified Wire For Linux Page 107

Chapter VII. iSCSI PDU Offload Target

3. Software/Driver Unloading

Use the following command to unload the module:

[root@host]# rmmod chiscsi_t4

Chelsio T5/T4 Unified Wire For Linux Page 108

Chapter VII. iSCSI PDU Offload Target

4. Software/Driver Configuration and Fine-tuning

The Chelsio iSCSI software needs configuration before it can become useful. The following sections describe how this is done.

There are two main components used in configuring the Chelsio iSCSI software: the

configuration file and the iSCSI control tool. This section describes in some detail what they are and their relationship they have with one another.

4.1.

Command Line Tools

There are two command line tools, one for control of the iSNS client and one for control of the iSCSI target nodes.

4.1.1. iscsictl

The Chelsio iSCSI control tool, iscsictl , is a Command Line Interface (CLI) user space program that allows administrators to:

Start/Stop the iSCSI Target

Start the iSNS client

Get/Set the iSCSI driver global settings

Get/Set/Remove the iSCSI Target configuration settings

Retrieve active sessions ’ information of an iSCSI Target

Manually flush data to the iSCSI Target disks

Reload the iSCSI configuration file

Update the iSCSI configuration file

Save the current iSCSI configuration to a file

4.1.2. chisns

The Chelsio iSNS client, chisns , can be started independently of iscsictl .

4.2.

iSCSI Configuration File

The iSCSI configuration file is the place where information about the Chelsio iSCSI software is stored. The information includes global data that pertains to all targets as well as information on each specific iSCSI target node. Most of the information that can be placed in the configuration file has default values that only get overwritten by the values set in the configuration file.

There are only a few global configuration items that can be changed. An example is

“iscsi_offload_mode” which specifies the hardware protocol offload method that should be used

(e.g. “TOE”, “ULP”, or “AUTO” modes) for all iSCSI targets.

Chelsio T5/T4 Unified Wire For Linux Page 109

Chapter VII. iSCSI PDU Offload Target

There are many specific parameters that can be configured, some of which are iSCSI specific and the rest bein g Chelsio specific. An example of an iSCSI specific item is “HeaderDigest” which is defaulted to “None” but can be overridden to “CRC32C”. An example of a Chelsio specific configurable item is “ACL” (for Access Control List). “ACL” is one of the few items that have no default.

Before starting any iSCSI target, an iSCSI configuration file must be created. An easy way to create this file is to use the provided sample configuration file and modify it. This file can be named anything and placed in any directory but it must be explicitly specified when using iscsictl by using the -f option. To avoid this, put configuration file in the default directory

(/etc/chelsio-iscsi ) and name it the default file name ( chiscsi.conf

).

4.2.1. “On the fly” Configuration Changes

Parameters for the most part can be changed while an iSCSI node is running. However, there are exceptions and restrictions to this rule that are explained in a later section that describes the details of the iSCSI control tool iscsictl .

4.3.

A Quick Start Guide for Target

This section describes how to get started quickly with a Chelsio iSCSI target. It includes:

Basic editing of the iSCSI configuration file.

Basic commands of the iSCSI control tool including how to start and stop a target.

4.3.1. A Sample iSCSI Configuration File

The default Chelsio iSCSI configuration file is located at /etc/chelsio-iscsi/chiscsi.conf

.

If this file doesn’t already exist then one needs to be created.

To configure an iSCSI target, there are three required parameters (in the form of key=value pairs) needed as follows:

TargetName – A worldwide unique iSCSI target name.

PortalGroup – The portal group tag associating with a list of target IP address (es) and port number(s) that service the login request. The format of this field is a Chelsio specific iSCSI driver parameter which is described in detail in the configuration file section.

TargetDevice – A device served up by the associated target. A device can be:

A block device (for example, /dev/sda)

A virtual block device (for example, /dev/md0)

A RAM disk

A regular file

Chelsio T5/T4 Unified Wire For Linux Page 110

Chapter VII. iSCSI PDU Offload Target

A target can serve multiple devices, each device will be assigned a Logical Unit Number (LUN) according to the order it is specified (i.e., the first device specified is assigned LUN 0, the second one LUN 1, …, and so on and so forth). Multiple TargetDevice key=value pairs are needed to indicate multiple devices.

Here is a sample of a minimum iSCSI target configuration located at /etc/chelsioiscsi/chiscsi.conf

: target:

TargetName=iqn.2006-02.com.chelsio.diskarray.san1

TargetDevice=/dev/sda

[email protected]:3260

The TargetDevice value must match with the storage device in the system. The PortalGroup value must have a matching IP address of the Ethernet adapter card in the system.

For more information about TargetDevice configuration see

Target Storage Device

Configuration .

4.3.2. Basic iSCSI Control

Control of the Chelsio iSCSI software is done through iscsictl , the command line interface control tool. The following are the basic commands needed for effective control of the target.

Start Target: To start all of the iSCSI targets specified in the iSCSI configuration file, execute iscsictl with the “-S” option followed by “ target=ALL ”.

[root@host]# iscsictl –f /etc/chelsio-iscsi/chiscsi.conf –S target=ALL

To start a specific target execute iscsictl with “ -S ” followed by the target.

[root@host]# iscsictl –f /etc/chelsio-iscsi/chiscsi.conf –S target=iqn.2006-02.com.chelsio.diskarray.san1

Stop Target: To stop the all the iSCSI target(s), execute iscsictl with “ -s ” option followed by

“ target=ALL ”.

[root@host]# iscsictl –s target=ALL

Chelsio T5/T4 Unified Wire For Linux Page 111

Chapter VII. iSCSI PDU Offload Target

To stop a specific target execute iscsictl with “ -s ” followed by the target name.

[root@host]# iscsictl –s target=iqn.2006-02.com.chelsio.diskarray.san1

View Configuration: To see the configuration of all the active iSCSI targets, execute iscsictl with “ -c ” option.

[root@host]# iscsictl –c

To see the more detailed configuration settings of a specific target, execute iscsictl with “ -c ” option followed by the target name.

[root@host]# iscsictl –c target=iqn.2006-02.com.chelsio.diskarray.san1

View Global Settings: To see Chelsio global settings, execute iscsictl with “ -g ” option.

[root@host]# iscsictl –g

Change Global Settings: To change Chelsio global settings, execute iscsictl with “ -G ” option.

[root@host]# iscsictl –G iscsi_offload_mode=AUTO

View Help: To print help to stdout, execute iscsictl with “ -h ” option.

[root@host]# iscsictl –h

Chelsio T5/T4 Unified Wire For Linux Page 112

Chapter VII. iSCSI PDU Offload Target

4.4.

The iSCSI Configuration File

The iSCSI configuration file consists of a series of blocks consisting of the following types of iSCSI entity blocks:

1. global

2. target

There can be only one global entity block whereas multiple target entity blocks are allowed. The global entity block is optional but there must be at least one target entity block.

An entity block begins with a block type (global or target). The content of each entity block is a list of parameters specified in a "key=value" format. An entity block ends at the beginning of the next entity block or at the end-of-file.

The parameter list in an entity block contains both:

iSCSI parameters that override the default values

Parameters that facilitate passing of control information to the iSCSI module

All lines in the configuration file that begin with “ # ” character are treated as comments and will be ignored. White space is not significant except in key=value pairs.

For the “key=value” parameters the <value> portion can be a single value or a list of multiple values. When <value> is a list of multiple values, they must be listed on one line with a comma

“,” to separate their values. Another way to list the values instead of commas is to list their values as key=value pairs repeatedly, each on a new line, until they are all listed.

There are three categories of “key=value” parameter, the first category belongs to the global entity block whereas the second and third categories belong to target and initiator entity blocks:

1. The Chelsio Global Entity Settings of key=value pairs

2. The iSCSI Entity Settings of key=value pairs

3. The Chelsio Entity Settings of key=value pairs

The following sub-sections describe these three categories and list in tables the details of their key=value parameters.

4.4.1. Chelsio System Wide Global Entity Settings

Description

Chelsio System Wide Global Entity Parameters pass system control information to the iSCSI software which affects all targets in the same way. More detail of these parameters below can be found in a later section entitled “System Wide Parameters”.

Chelsio T5/T4 Unified Wire For Linux Page 113

Chapter VII. iSCSI PDU Offload Target

Table of Chelsio Global Entity Settings

Key iscsi_offload_mode iscsi_auth_order

Valid

Values

“AUTO”

“TOE”

“ULP”

“ACL”

“CHAP”

Default

Value

“AUTO”

“CHAP”

Multiple

Values

No

No

No

Description

Defines the offload mode

AUTO: iSCSI software will make the decision. If the connection goes through

Chelsio’sHBA which has the iSCSI acceleration enabled, then ULP.

TOE: Use Chelsio HBA TCP

Offloading Engine (TOE) capabilities.

ULP: Use Chelsio iSCSI hardware acceleration in terminator ASIC including Direct

Data Placement.

NOTES: CLI will allow the user to change the offload mode to

TOE even though there is no

TOE functionality existing. This is because, with a TOE network adaptor card, its TOE capability could be disabled anytime, in which case the traffic will be treated as normal NIC traffic.

Authorization order for login verification on the target. Valid only when a target’s

ACL_Enable=Yes

ACL: ACL first then CHAP

CHAP: CHAP first then ACL

Applies to Target(s) Only

The target vendor ID part of the device identification sent by an iSCSI target in response of SCSI

Inquiry command. iscsi_target_vendor_id a string of maximum of 8 characters

“CHISCSI”

4.4.2. iSCSI Entity Settings

Description iSCSI Entity Parameters pass iSCSI protocol control information to the Chelsio iSCSI module.

This information is unique for each entity block. The parameters follow the IETF iSCSI standard

RFC 3720 in both definition and syntax. The descriptions below are mostly from this RFC.

Chelsio T5/T4 Unified Wire For Linux Page 114

Chapter VII. iSCSI PDU Offload Target

Table of iSCSI Entity Settings

Key

MaxConnections

InitialR2T

ImmediateData

FirstBurstLength

MaxBurstLength

DefaultTime2Wait

DefaultTime2Retain

MaxOutstandingR2T

DataPDUInOrder

DataSequenceInOrder

ErrorRecoveryLevel

HeaderDigest

DataDigest

AuthMethod

TargetName

TargetAlias

Valid

Values

1 to 65535

“Yes”

“No”

“Yes”

“No”

512 to

16777215

(2

24

- 1)

512 to

16777215

(2

24

- 1)

0 to 3600

0 to 3600

1 to 65535

“Yes”

“No”

“Yes”

“No”

0 to 2

MaxRecvDataSegmentLength 512 to

OFMarker

16777215

(2

24

- 1)

“Yes”

“No”

Default

Value

1

Multiple

Values

No

“Yes”

“Yes”

No

No

Description

Initiator and target negotiate the maximum number of connections requested/acceptable.

To turn on or off the default use of R2T for unidirectional and the output part of bidirectional commands.

To turn on or off the immediate data.

65536 No

262144

2

20

1

“Yes”

“Yes”

0

No

No

No

“None”

“CRC32C”

“None”

“CRC32C”

“CHAP” and “None”

“<target name>”

“<target alias>”

“None”

Yes

“None”

Yes

“None,

CHAP”

Yes

No

No

8192

“No”

No

No

No

No

No

No

The maximum negotiated SCSI data in bytes of unsolicited data that an iSCSI initiator may send to a target during the execution of a single SCSI command.

The maximum negotiated SCSI data in bytes, of a Data-In or a solicited Data-

Out iSCSI sequence between the initiator and target.

The minimum time, in seconds, to wait before attempting an explicit / implicit logout or connection reset between initiator and target.

The maximum time, in seconds, after an initial wait.

The maximum number of outstanding

R2Ts per task.

To indicate the data PDUs with sequence must be at continuously increasing order or can be in any order.

Chelsio only supports “Yes”.

To indicate the Data PDU sequences must be transferred in continuously non-decreasing sequence offsets or can be transferred in any order.

Chelsio only supports “Yes”.

To negotiate the recovery level supported by the node.

Chelsio only supports 0.

To enable or disable iSCSI header

Cyclic integrity checksums.

To enable or disable iSCSI data Cyclic integrity checksums.

To choose an authentication method during login phase.

A worldwide unique iSCSI target name.

Target only.

A human-readable name or description of a target. It is not used as an identifier, nor is it for authentication.

Target only.

To declare the maximum data segment length in bytes it can receive in an iSCSI PDU.

To turn on or off the initiator to target markers on the connection.

Chelsio only supports “No”.

Chelsio T5/T4 Unified Wire For Linux Page 115

Chapter VII. iSCSI PDU Offload Target

IFMarker

OFMarkInt

IFMarkInt

“Yes”

“No”

1 to 65535

1 to 65535

4.4.3. Chelsio Entity Settings

Description

“No”

2048

2048

No

No

No

To turn on or off the target to initiator markers on the connection.

Chelsio only supports “No”.

To set the interval for the initiator to target markers on a connection.

To set the interval for the target to initiator markers on a connection.

Chelsio Entity Parameters pass control information to the Chelsio iSCSI module. The parameters are specific to Chelsio’simplementation of the iSCSI node (target or initiator) and are unique for each entity block. The parameters consist of information that can be put into three categories:

1. Challenge Handshake Authentication Protocol (CHAP).

2. Target specific settings. All of the following parameters can have multiple instances in one target entity block (i.e., they can be declared multiple times for one particular target).

Portal Group

Storage Device

Access Control List (ACL)

Table of Chelsio Entity Settings

Key Valid Values

Default

Value

Multiple

Values

Description

Auth_CHAP_Target

Auth_CHAP_Initiator

Chelsio CHAP Parameter (Target)

“<user id>”

:“<secret>”

No

“<user id>”

:“<secret>”

Yes

CHAP user id and secret for the target.

<user id> must be less than 256 characters. Commas “,” are not allowed.

<secret> must be between 6 and 255 characters. Commas “,” are not allowed.

The target user id and secret are used by the initiator to authenticate the target while doing mutual chap.

NOTE: The double quotes are required as part of the format.

CHAP user id and secret for the initiator.

<user id> must be less than 256 characters. Commas “,” are not

Chelsio T5/T4 Unified Wire For Linux Page 116

Chapter VII. iSCSI PDU Offload Target

Auth_CHAP_Challenge

Length

Auth_CHAP_Policy

PortalGroup

ShadowMode allowed.

<secret> must be between 6 and 255 characters. Commas “,” are not allowed.

The initiator user id and secret are used by the target to authenticate the initiator.

NOTE: The double quotes are required as part of the format.

CHAP challenge length 16 to 1024 16 No

“Oneway” or

“Mutual”

“Oneway”

No Oneway or Mutual (two-way) CHAP

Chelsio Target Specific Parameter

<portal group tag>

@<target IP address>

[:<port number>]

.

.

.

[,<target IP address>

[:<port number>]]

[,timeout=

<timeout value in milliseconds>

]

[,[portalgrou ptag1, portalgroupta g2,… portalgroupta

gn]

Yes

“Yes”

“No”

“No”

No

The portal group name associates the given target with the given list of IP addresses (and optionally, port numbers) for servicing login requests.

It‟s required to have at least one per target.

<portal group tag> is a unique tag identifying the portal group. It must be a positive integer.

<target IP address> is the IP address associated with the portal group tag.

<port number> is the port number associated with the portal group tag. It is optional and if not specified the wellknown iSCSI port number of 3260 is used.

<timeout> is optional, it applies to all the portals in the group.

The timeout value is in milliseconds and needs to be multiple of 100ms.

It is used to detect loss of communications at the iSCSI level.

NOTE: There can be multiple target IP address/port numbers per portal group tag. This enables a target to operate on multiple interfaces for instance.

<portalgrouptagX>The portalgroup to which login requests should be redirected to.

NOTE: There can be multiple redirection target portalgroups specified for a particular target portal group and the redirection will happen

to these in a round robin manner.

To turn ShadowMode on or off for iSCSI Target Redirection

Chelsio T5/T4 Unified Wire For Linux Page 117

Chapter VII. iSCSI PDU Offload Target

TargetSessionMaxCmd

1 to 2048

TargetDevice* <path/name>

[,FILE|MEM|BL

K]

[,NULLRW]

[,SYNC]

[,RO]

[,size=xMB]

[,ScsiID=xxxx xx]

[,WWN=xxxxxxx xx]

64

Chelsio T5/T4 Unified Wire For Linux

No

No

The maximum number of outstanding iSCSI commands per session.

A device served up by the associated target.

The device mode can be a:

Block Device (e.g. /dev/sda)

Virtual Block Device (e.g.

/dev/md0)

 RamDisk

 Regular File

<path/name> is the path to the device

- with the exception of when a RAM

Disk is specified, where it is a unique name given to the device. If multiple

RAM Disks are used for a target then each name must be unique within the target.

NULLRW specifies that random data is returned for reads, and for writes data is dropped. Useful for testing network performance.

SYNC specifies that the device will function in the write-through mode (i.e., the data will be flushed to the device before the response is returned to the initiator).

NOTE: SYNC is only applicable with

FILE mode.

RO specifies the device as a read-only device.

FILE specifies this device should be accessed via the kernel‟s VFS layer.

This mode is the most versatile, and it is the default mode in the cases where there is no mode specified.

BLK specifies this device should be accessed via the kernel‟s block layer.

This mode is suitable for high-speed storage device such as RAID

Controllers.

MEM specifies this device should be created as a RAM Disk. size =xMB is used with “MEM”, to specify the RamDisk size. If not specified, the default RamDisk size is

16MB (16 Megabytes). The minimum value of x is 1 (1MB) and the maximum value is limited by system memory.

ScsiID=xxxxxx is a 16 character unique value set for multipath aware iSCSI initiator host.

Page 118

Chapter VII. iSCSI PDU Offload Target

ACL_Enable

ACL

“Yes”

“No”

[iname=<name1

>][;<sip=<sip

1>][;dip=<dip

1>][;lun=<lun

_list:permiss ions>]

Chelsio T5/T4 Unified Wire For Linux

“No”

No

Yes

WWN=xxxxxx is a 16 character unique value set for multipath aware iSCSI initiator host.

When multipath aware initiator host is accessing the storage Logical Unit

Number( LUN) via multiple iSCSI session, ScsiID and WWN values must be set for the TargetDevice. These values will be returned in Inquiry response (VPD 0x83).

Multiple TargetDevice key=value pairs are needed to indicate multiple devices.

There can be multiple devices for any particular target. Each device will be assigned a Logical Unit Number (LUN) according to the order it is specified

(i.e., the first device specified is assigned LUN 0, the second one LUN

1, …, and so on and so forth).

NOTE: FILE mode is the most versatile mode, if in doubt use FILE mode.

*During MPIO i.e. when two iSCSI targets are using the same

TargetDevice, intermittently it is observed that LUN shows up twice on the initiator, to work around this behavior, try using the same

TargetAlias on both the targets.

Defines if Chelsio’sAccess Control List

(ACL) method will be enforced on the target:

Yes: ACL is enforced on the target

No: ACL is not enforced on the target

The ACL specifies which initiators and how they are allowed to access the

LUNs on the target. iname=<Initiator Name> specifies one or more initiator names, the name must be a fully qualified iSCSI initiator name.

sip=<Source IP address> specifies one or more IP addresses the initiators are connecting from.

Dip=<Destination IP address> specifies one or more IP addresses that the iSCSI target is listening on

(i.e., the target portal IP addresses).

NOTE: when configuring an ACL at least one of the above three must be provided:

 iname, and/or

Page 119

Chapter VII. iSCSI PDU Offload Target

RegisteriSNS

 sip, and/or

 dip.

lun=<lun list>:<permission> controls how the initiators access the luns.

The supported value for <lun list> is

ALL.

<permissions> can be:

R: Read Only

RW or WR: Read and Write

If permissions are specified then the associated LUN list is required.

If no lun=<lun list>:[R|RW] is specified then it defaults to ALL:RW.

NOTE: For the Chelsio Target

Software release with lun-masking included,

<lun list> is in the format of <0..N |

0~N | ALL>

Where:

0..N: only one value from 0 through N

0~N: a range of values between 0 through N

ALL: all currently supported LUNs.

Multiple lists of LUN numbers are allowed. When specifying the list

separate the LUN ranges by a comma.

To turn on or off exporting of target information via iSNS

“Yes”

“No”

“Yes”

No

4.4.4. Sample iSCSI Configuration File

Following is a sample configuration file. While using iSCSI node (target), irrelevant entity block can be removed or commented.

#

# Chelsio iSCSI Global Settings

# global:

iscsi_offload_mode=AUTO

iscsi_auth_order=ACL

#

# an iSCSI Target “iqn.2006-02.com.chelsio.diskarray.san1”

# being served by the portal group "5". Setup as a RAM Disk.

# target:

TargetName=iqn.2006-02.com.chelsio.diskarray.san1

Chelsio T5/T4 Unified Wire For Linux Page 120

Chapter VII. iSCSI PDU Offload Target

# lun 0: a ramdisk with default size of 16MB

TargetDevice=ramdisk,MEM

[email protected]:3260

#

# an iSCSI Target “iqn.2005-8.com.chelsio:diskarrays.san.328”

# being served by the portal group "1" and "2"

# target:

#

# iSCSI configuration

#

TargetName=iqn.2005-8.com.chelsio:diskarrays.san.328

TargetAlias=iTarget1

MaxOutstandingR2T=1

MaxRecvDataSegmentLength=8192

HeaderDigest=None,CRC32C

DataDigest=None,CRC32C

ImmediateData=Yes

InitialR2T=No

FirstBurstLength=65535

MaxBurstLength=262144

#

# Local block devices being served up

# lun 0 is pointed to /dev/sda

# lun 1 is pointed to /dev/sdb

TargetDevice=/dev/sda,ScsiID=aabbccddeeffgghh,WWN=aaabbbcccdddeeef

TargetDevice=/dev/sdb

#

# Portal groups served this target

#

[email protected]:3260

[email protected]:3260

#

# CHAP configuration

Chelsio T5/T4 Unified Wire For Linux Page 121

Chapter VII. iSCSI PDU Offload Target

#

Auth_CHAP_Policy=Mutual

Auth_CHAP_target=“iTarget1ID”:“iTarget1Secret”

Auth_CHAP_Initiator=“iInitiator1”:“InitSecret1”

Auth_CHAP_Initiator=“iInitiator2”:“InitSecret2”

Auth_CHAP_ChallengeLength=16

#

# ACL configuration

#

# initiator “iqn.2006-02.com.chelsio.san1” is allowed full access

# to this target

ACL=iname=iqn.2006-02.com.chelsio.san1

# any initiator from IP address 102.50.50.101 is allowed full access

# of this target

ACL=sip=102.50.50.101

# any initiator connected via the target portal 102.60.60.25 is

# allowed full access to this target

ACL=dip=102.60.60.25

# initiator “iqn.2005-09.com.chelsio.san2” from 102.50.50.22 and

# connected via the target portal 102.50.50.25 is allowed read only

# access of this target

ACL=iname=iqn.2006-

02.com.chelsio.san2;sip=102.50.50.22;dip=102.50.50.25;lun=ALL:R

4.5.

Challenge-Handshake Authenticate Protocol (CHAP)

CHAP is a protocol that is used to authenticate the peer of a connection and uses the notion of a challenge and response, (i.e., the peer is challenged to prove its identity).

The Chelsio iSCSI software supports two CHAP methods: one-way and mutual. CHAP is not supported for discovery sessions.

4.5.1. Oneway CHAP authentication

With one-way CHAP (also called unidirectional CHAP) the target uses CHAP to authenticate the initiator. The initiator does not authenticate the target. This method is the default method.

Chelsio T5/T4 Unified Wire For Linux Page 122

Chapter VII. iSCSI PDU Offload Target

For one-way CHAP, the initiator CHAP id and secret are configured and stored on a per-initiator with Chelsio Entity parameter “Auth_CHAP_Initiator”.

4.5.2. Mutual CHAP authentication

With mutual CHAP (also called bidirectional CHAP), the target and initiator use CHAP to authenticate each other.

For mutual CHAP, in addition to the initiator CHAP id and secret, the target CHAP id and secret are required. They are configured and stored on a per target basis with Chelsio Entity parameter

“Auth_CHAP_Target”.

4.5.3. Adding CHAP User ID and Secret

A single Auth_CHAP_Target key and multiple Auth_CHAP_Initiator keys could be configured per target: target:

TargetName=iqn.2006-02.com.chelsio.diskarray.san1

TargetDevice=/dev/sda [email protected]:8000

Auth_CHAP_Policy=Oneway

Auth_CHAP_Initiator=”remoteuser1”:”remoteuser1_secret”

Auth_CHAP_Initiator=”remoteuser2”:”remoteuser2_secret”

Auth_CHAP_Target=“targetid1”:”target1_secret”

In the above example, target iqn.2005-com.chelsio.diskarray.san1

has been configured to authenticate two initiators, and its own id and secret are configured for use in the case of mutual

CHAP.

4.5.4. AuthMethod and Auth_CHAP_Policy Keys

By setting the iSCSI keys AuthMethod and Auth_CHAP_Policy , a user can choose whether to enforce CHAP and if mutual CHAP needs to be performed.

The AuthMethod key controls if an initiator needs to be authenticated or not. The default setting of AuthMethod is None,CHAP

The Auth_CHAP_Policy key controls which CHAP authentication (one-way or mutual) needs to be performed if CHAP is used. The default setting of Auth_CHAP_Policy is Oneway

On an iSCSI node, with AuthMethod=None,CHAP

Auth_CHAP_Policy=Oneway , Chelsio iSCSI target will accept a relevant initiator if it does a) no CHAP b) CHAP Oneway or c) CHAP Mutual

Chelsio T5/T4 Unified Wire For Linux Page 123

Chapter VII. iSCSI PDU Offload Target

Auth_CHAP_Policy=Mutual , the Chelsio iSCSI target will accept a relevant initiator if it does a) no CHAP or b) CHAP Mutual

With AuthMethod=None , regardless the setting of the key Auth_CHAP_Policy, the Chelsio iSCSI target will only accept a relevant initiator if it does no CHAP.

With AuthMethod=CHAP , CHAP is enforced on the target: i. Auth_CHAP_Policy=Oneway , the iSCSI target will accept a relevant initiator only if it does a) CHAP Oneway or b) CHAP Mutual ii. Auth_CHAP_Policy=Mutual , the iSCSI node will accept a relevant initiator only if it does a) CHAP Mutual

4.6.

Target Access Control List (ACL) Configuration

The Chelsio iSCSI target supports iSCSI initiator authorization via an Access Control List (ACL).

ACL configuration is supported on a per-target basis. The creation of an ACL for a target establishes:

Which iSCSI initiators are allowed to access it

The type of the access: read-writeor read-only

Possible SCSI layer associations of LUNs with the initiator

More than one initiator can be allowed to access a target and each initiator‟s access rights can be independently configured.

The format for ACL rule is as follows:

ACL=[iname=<initiator name>][;<sip=<source ip addresses>]

[;dip=<destination ip addresses>][;lun=<lun_list>:<permissions>] target:

TargetName=iqn.2006-02.com.chelsio.diskarray.san1

TargetDevice=/dev/sda

[email protected]:3260

[email protected]:3260

# initiator “iqn.2006-02.com.chelsio.san1” is allowed

# full read-write access to this target

Chelsio T5/T4 Unified Wire For Linux Page 124

Chapter VII. iSCSI PDU Offload Target

ACL=iname=iqn.2006-02.com.chelsio.san1

# any initiator from IP address 102.50.50.101 is allowed full

# read-write access of this target

ACL=sip=102.50.50.101

# any initiator connected via the target portal 102.60.60.25

# is allowed full read-write access to this target

ACL=dip=102.60.60.25

# initiator “iqn.2005-09.com.chelsio.san2” from 102.50.50.22

# and connected via the target portal 102.50.50.25 is allowed

# read only access of this target

ACL=iname=iqn.2006-

02.com.chelsio.san2;sip=102.50.50.22;dip=102.50.50.25;lun=ALL:R

4.6.1. ACL Enforcement

To toggle ACL enforcement on a per-target base, a Chelsio keyword ACL_Enable is provided:

Setting ACL_Enable=Yes enables the target to perform initiator authorization checking for all the initiators during login phase. And in addition, once the initiator has been authorized to access the target, the access rights will be checked for each individual LU the initiator trying to access.

Setting ACL_Enable=No disable the target to perform initiator authorization checking.

When a target device is marked as read-only (RO ), it takes precedence over ACL’s write permission (i.e., all of ACL write permission of an initiator is ignored).

4.7.

Target Storage Device Configuration

An iSCSI Target can support one or more storage devices. The storage device can either be the built-in RAM disk or actual backend storage.

Configuration of the storage is done through the Chelsio configuration file via the key-value pair

TargetDevice .

When option NULLRW is specified, on writes the data is dropped without being copied to the storage device, and on reads the data is not actually read from the storage device but instead random data is used. This option is usefuly for measuring network performance.

Chelsio T5/T4 Unified Wire For Linux Page 125

Chapter VII. iSCSI PDU Offload Target

The details of the parameters for the key TargetDevice are found in the table of Chelsio Entity

Settings section earlier in this document.

4.7.1. RAM Disk Details

For the built-in RAM disk:

The minimum size of the RAM disk is 1 Megabyte (MB) and the maximum is limited by system memory.

To use a RAM disk with a Windows Initiator, it is recommended to set the size >= 16MB.

To configure an ramdisk specify MEM as the device mode:

TargetDevice=<name>,MEM,size=xMB

Where: <name> Is a unique name given to the RAM disk. This name identifies this x particular ramdisk. If multiple RAM disks are configured for the same target, the name must be unique for each RAM Disk.

Is the size of the RAM disk in MB. It’s an integer between (1-max), where max is limited by system memory. If this value is not specified the default value is 16 MB. target:

#<snip>

# 16 Megabytes RAM Disk named ramdisk1

TargetDevice=ramdisk1,MEM,size=16MB

#<snip>

4.7.2. FILE Mode Storage Device Details

The FILE mode storage device is the most common and versatile mode to access the actual storage attached to the target system:

The FILE mode can accommodate both block devices and virtual block devices.

The device is accessed in the exclusive mode. The device should not be accessed (or active) in any way on the target system.

Each device should be used for one and only one iSCSI target.

“SYNC” can be used with FILE mode to make sure the data is flushed to the storage device before the Target responds back to the Initiator.

To configure a FILE storage device specify FILE as the device mode:

TargetDevice=<path to the storage device>[,FILE][,SYNC]

Chelsio T5/T4 Unified Wire For Linux Page 126

Chapter VII. iSCSI PDU Offload Target

Where: <path>

SYNC

Is the path to the actual storage device, such as /dev/sdb for a block device or /dev/md0 for a software RAID. The path must exist in the system.

When specified, the Target will flush all the data in the system cache to the storage driver before sending response back to the Initiator.

4.7.3. Example Configuration of FILE Mode Storage

Below is an example: target:

#<snip>

# software raid /dev/md0 is accessed in FILE mode

TargetDevice=/dev/md0,FILE

#<snip>

4.7.4. BLK Mode Storage Device Details

The BLK mode storage device is suitable for high-speed storage attached to the target system:

The BLK mode can accommodate only block devices.

The device is accessed in the exclusive mode. The device should not be accessed (or active) in any way on the target system.

Each device should be used for one and only one iSCSI target.

To configure a block storage device specify BLK as the device mode:

TargetDevice=<path to the storage device>,BLK

Where: target:

<path> Is the path to the actual storage device, such as /dev/sdb. The path must exist in the system.

#<snip>

# /dev/sdb is accessed in BLK mode

TargetDevice=/dev/sdb,BLK

#<snip>

Chelsio T5/T4 Unified Wire For Linux Page 127

Chapter VII. iSCSI PDU Offload Target

4.8.

Target Redirection Support

An iSCSI Target can redirect an initiator to use a different IP address and port (often called a portal) instead of the current one to connect to the target. The redirected target portal can either be on the same machine, or a different one.

4.8.1. ShadowMode for Local vs. Remote Redirection

The ShadowMode setting specifies whether the Redirected portal groups should be present on the same machine or not. If ShadowMode is enabled, the redirected portal groups are on a different system. If it is disabled then the redirected portal groups must be present on the same system otherwise the target would fail to start.

Below is an example with ShadowMode enabled: target:

#<snip>

# any login requests received on 10.193.184.81:3260 will be

# redirected to 10.193.184.85:3261.

[email protected]:3260,[2]

[email protected]:3261

# the PortalGroup “2” is NOT presented on the same system.

ShadowMode=Yes

#<snip>

Below is an example with ShadowMode disabled: target:

#<snip>

# any login requests received on 10.193.184.81:3260 will be

# redirected to 10.193.184.85:3261

[email protected]:3260,[2]

[email protected]:3261

# the PortalGroup “2” IS present on the same system

ShadowMode=No

#<snip>

Chelsio T5/T4 Unified Wire For Linux Page 128

Chapter VII. iSCSI PDU Offload Target

4.8.2. Redirecting to Multiple Portal Groups

The Chelsio iSCSI Target Redirection allows redirecting all login requests received on a particular portal group to multiple portal groups in a round robin manner.

Below is an example Redirection to Multiple Portal Groups: target:

#<snip>

# any login requests received on 10.193.184.81:3260 will be

# redirected to 10.193.184.85:3261 and 10.193.184.85:3262 in a

# Round Robin Manner.

[email protected]:3260,[2,3]

[email protected]:3261

[email protected]:3262

ShadowMode=No

#<snip>

4.9.

The com mand line interface tools “iscsictl” & “chisns”

4.9.1. iscsictl

iscsictl is the tool Chelsio provides for controlling the iSCSI target. It is a Command Line

Interface (CLI) that is invoked from the console. Its usage is as follows: iscsictl <options> <mandatory parameters> [optional parameters]

The mandatory and optional parameters are the key=value pair(s) defined in RFC3720, or the

var=const pair(s) defined for Chelsio iSCSI driver implementation. In this document, the key=value is referred to as “pair”, and var=const is referred to as “parameter” to clarify between iSCSI protocol‟s pair value(s), and Chelsio iSCSI driver‟s parameter value(s). Note that all

value and const are case sensitive.

4.9.2. chisns

chisns is the command line tool for controlling the iSNS client. This is a simple tool that starts the iSNS client with a client and server parameter.

Chelsio T5/T4 Unified Wire For Linux Page 129

Chapter VII. iSCSI PDU Offload Target

4.9.3. iscsictl options

Options

-h

-v

-f

Mandatory

Parameters

<[path/] filename>

Optional

Parameters

-k <key>[=<val>]

-c target=<name>

[,name2

.

.

.

,<nameN>]

Description

Display the help messages.

Display the version.

Specifies a pre-written iSCSI configuration text file, used to start, update, save, or reload the iSCSI node(s).

This option must be specified with one of the following other options: “-S”, “-U”, or “-W”. For the

-S” option “-f” must be specified first. All other options will ignore this “-f” option.

If the “-f” option is not specified with the commands above the default configuration file will be used. It‟s name and location is:

/etc/chelsio-iscsi/chiscsi.conf

The configuration file path and filename must conform to Linux standards.

For the format of the iSCSI configuration file, please see “Format of The iSCSI Configuration File” section earlier in this document.

Specifies an iSCSI Entity or Chelsio Entity parameter.

This option can be specified after “-c” option to retrieve a parameter setting..

Display the Chelsio iSCSI target configuration.

target=<name> parameter:

Where name is the name of the node whose

information will be returned. name can be one or

more string of names, separated by a comma,

<name1[,name2,…,nameN] | ALL>

A name of ALL returns information on all targets.

ALL is a reserved string that must be uppercase.

Example:

iscsictl -c target=iqn.com.cc.it1

Iscsictl -c target=iqn.com.cc.target1 -k

TargetAlias

The <name> parameter can also be specified as one or more parameter on the same command line, separated by a comma, target=<name1>, <name2>, … ,<nameN>

The target=<name> parameter(s) are optional and if not specified all active Chelsio iSCSI targets(s) configuration(s) will be displayed.

If target=ALL is specified or no parameters are specified the output will be abbreviated. Specify specific targets to get detailed configuration data.

Chelsio T5/T4 Unified Wire For Linux Page 130

Chapter VII. iSCSI PDU Offload Target

-F

-g

-G <var=const> target=<name>

-k lun=<value>

If the target=<name> option is specified, the -k

<key> option can optionally be specified along with this option to display only the selected entity parameter setting.

Example: iscsictl -c target=iqn.com.cc.target1 -k

HeaderDigest

Flush the cached data to the target disk(s).

target=<name> parameter:

Where name is the name of the target to be flushed.

name can be one or more string of names, separated by a comma,

<name1[,name2,…,nameN] | ALL>

A name of ALL will cause all the target data to be flushed. ALL is a reserved string that must be uppercase.

The target=name parameter is optional. If no

target=name parameter is specified, it is the same as specifying target=ALL.

The -k lun=<value> option is optional. It can be used to further specify a particular lun to be flushed.

Example:

To flush all the targets in the system: iscsictl -F

To flush a particular target: iscsictl -F target=iqn.com.cc.it1

To flush only the lun 0 of a particular target:

iscsictl -F target=iqn.com.cc.it1 -k lun=0

Display the Chelsio iSCSI global settings.

The global parameters printed are: iscsi_offload_mode iscsi_auth_order iscsi_target_vendor_id *

*The iscsi_target_vendor_id will be printed only if it has been changed from the default value.

Set the Chelsio iSCSI global settings.

var=const parameter:

Where var=const can be:

iscsi_offload_mode=<AUTO | TOE | ULP>

iscsi_auth_order=<ACL | CHAP> iscsi_target_vendor_id=<string of max. 8 characters>

Example: iscsictl -G iscsi_auth_order=ACL

The var=const parameter(s) are mandatory.

If the var=const parameter is not specified, the command will be denied.

Chelsio T5/T4 Unified Wire For Linux Page 131

Chapter VII. iSCSI PDU Offload Target

-s target=<name>

-S target=<name>

Chelsio T5/T4 Unified Wire For Linux

If any of the specified var=const parameter is invalid, the command will reject only the invalid parameters, but will continue on and complete all other valid parameters if any others are specified.

Stop the specified active iSCSI targets.

target=<name> parameter:

See the description of option -c for the target=<name> parameter definition.

The target=<name> parameter is mandatory. If no

target=<name> parameter is specified, the command will be denied.

If the target=<name> parameter is specified, only the specified targets from the target=<name> parameters will be stopped.

If target=ALL is specified, all active targets will stop.

Start or reload the iSCSI targets.

target=<name> parameter:

Where name is the name of the target(s) that will be started or reloaded.

The target=<name> parameter can be specified as one or more parameter on the same command line, separated by a space, target=<name1> target=<name2> … target=<nameN>

The target=<name> parameter can also be, target=ALL

A name of ALL starts or reloads all targets specified in the configuration file. ALL is a reserved string that must be uppercase.

The target=<name> parameter is optional.

If this command line option is specified without the -f option, the default configuration file

/etc/chelsio-iscsi/chiscsi.conf will be used.

Rules,

1. If the target=<name> parameter is specified, only the targets from the list will be started or reloaded.

2. If target=ALL is specified, all targets specified from the iSCSI configuration file will be started or reloaded.

3. If the target=<name> parameter is not specified, all active targets configurations will be reloaded from the configuration file while those targets are running.

All non-active targets specified will not be loaded / started.

For Rules 1-3, if the specified targets are currently active (running), they will get reloaded.

For Rules 1 & 2, if the specified targets are not

Page 132

Chapter VII. iSCSI PDU Offload Target

-r

-U

-W

-h target=<name> -k initiator=<name> server=<IP address>

[:<port>] id=<isns entity id> query=<query interval> currently active, they will be started.

For Rules 2 & 3, please note the differences

– they are not the same!

The global settings are also reloaded from the configuration file with this option.

Retrieve active iSCSI sessions under a target.

target=<name> parameter:

Where name must be a single target name.

If target=<name> parameter is specified as target=<name>, the sessions can be further filtered based on the remote node name with optional –k initiator=<name> option.

Examples:

iscsictl -r target=iqn.com.cc.it1 iscsictl -r target=iqn.com.cc.it1 -k initiator=iqn.com.cc.ii1

The first target=<name> parameter is mandatory. If it is not specified, the command will be denied.

Update the specified iSCSI configuration file to include the current iSCSI global settings and the active iSCSI targets‟ configuration.

Will not delete the nonactive targets’ configuration from the specified file. The order of the targets’ configuration might get re-organized afterward.

The -f option MUST be specified along with this option.

Overwrite the specified iSCSI configuration file with

ONLY the current iSCSI global settings and the active iSCSI targets‟ configuration to the specified iSCSI configuration file.

Will delete any nonactive targets’ configuration from the specified file.

The -f option MUST be specified along with this option.

Display the help messages.

Start the Chelsio iSNS client.

server=<IP address>[:<port>] where server is the iSNS server address. The port is optional and if it‟s not specified it defaults to 3205. The server with the ip address is mandatory and if it‟s not specified the, the command will be denied.

id=<isns entity id> where id is the iSNS entity ID used to register with the server. It defaults to

<hostname>.

query=<query interval> where query is the initiator query interval (in seconds). It defaults to 60 seconds.

Examples: chisns server=192.0.2.10 chisns server=192.0.2.10:3205 id=isnscln2 query=30

Chelsio T5/T4 Unified Wire For Linux Page 133

Chapter VII. iSCSI PDU Offload Target

In the first example the minimum command set is given where the IP address of the iSNS server is specified.

In the second example a fully qualified command is specified by also setting three optional parameters.

Here, the mandatory IP address and the corresponding optional port number are specified.

Also set is the iSNS entity ID to “isnscln2” as well as the query interval to 30 seconds.

4.10.

Rules of Target Reload (i.e. “on the fly” changes)

After a target has been started its settings can be modified via reloading of the configuration file

(i.e., iscsictl -S ).

The following parameters cannot be changed once the target is up and running otherwise the target reload would fail:

TargetName

TargetSessionMaxCmd

The following parameters CAN be changed by reloading of the configuration file. The new value will become effective IMMEDIATELY for all connections and sessions:

TargetDevice

PortalGroup

ACL_Enable

ACL

The following parameter CAN be changed by reloading of the configuration file. The new value will NOT affect any connections and sessions that already completed login phase:

TargetAlias

MaxConnections

InitialR2T

ImmediateData

FirstBurstLength

MaxBurstLength

MaxOutstandingR2T

HeaderDigest

DataDigest

MaxRecvDataSegmentLength

AuthMethod

Auth_CHAP_Initiator

Auth_CHAP_Target

Chelsio T5/T4 Unified Wire For Linux Page 134

Chapter VII. iSCSI PDU Offload Target

Auth_CHAP_ChallengeLength

Auth_CHAP_Policy

The following parameters SHOULD NOT be changed because only one valid value is supported:

DataPDUInOrder (support only "Yes")

DataSequenceInOrder (support only "Yes")

ErrorRecoveryLevel

OFMarker

(support only "0")

(support only "No")

IFMarker (support only "No")

The following parameters can be changed but would not have any effect because they are either not supported or they are irrelevant:

DefaultTime2Wait (not supported)

DefaultTime2Retain (not supported)

OFMarkInt (irrelevant because OFMarker=No)

IFMarkInt (irrelevant because IFMarker=No)

4.11.

System Wide Parameters

The Chelsio iSCSI software provides three system wide parameters that can be controlled through the configuration file, through the use of the command line “ iscsictl -G ”. The finer points of these parameters are described in detail here:

4.11.1. iscsi_offload_mode

iscsictl is the tool Chelsio provides for controlling the iSCSI target. It is a Command Line

Interface (CLI) that is invoked from the console. Its usage is as follows:

Options : “AUTO”, “TOE”, or “ULP”, defaults to “AUTO”

TOE: When selected, the iSCSI software attempts to offload the TCP connection. If the connection can be offloaded, it operates that connection with TCP/IP offloaded in the

Chelsio TOE hardware.

ULP: This is also known as iSCSI Hardware Acceleration. Like TOE mode, when selected the software attempts to offload the TCP connection. If the connection can be offloaded, it operates that connection with TCP/IP offloaded in the Chelsio TOE hardware. Additionally

(and differing from TOE mode), it also enables hardware calculations of the header and data digest and performs Direct Data Placement (DDP) into host memory.

AUTO: When selected the iSCSI software chooses either TOE mode or ULP mode automatically. It will select ULP mode when the connection is going through a Chelsio iSCSI acceleration enabled HBA. The purpose of this mode is to optimize the performance of the

Chelsio acceleration hardware.

Chelsio T5/T4 Unified Wire For Linux Page 135

Chapter VII. iSCSI PDU Offload Target

Note iscsi_offload_mode has no meaning when the iSCSI software is used on a non-

TOE based NIC.keyfile from Chelsio support.

4.11.2. iscsi_auth_order

Options : “ACL” or “CHAP”, defaults to “CHAP”

On an iSCSI target when ACL_Enable is set to Yes , iscsi_auth_order decides whether to perform CHAP first then ACL or perform ACL then CHAP.

ACL: When setting iscsi_auth_order=ACL , initiator authorization will be performed at the start of the login phase for an iSCSI normal session: upon receiving the first iscsi_login_request, the target will check its ACL. If this iSCSI connection does not match any ACL provisioned, the login attempt will be terminated.

CHAP: When setting iscsi_auth_order=CHAP , initiator authorization will be performed at the end of the login phase for an iSCSI normal session: before going to the full feature phase, the target will check its ACL. If this iSCSI connection does not match any ACL provisioned, the login attempt will be terminated.

Note iscsi_auth_order has no meaning when ACL_Enable is set to No on a target.

4.11.3. iscsi_target_vendor_id

Options: A string of maximum of 8 characters, defaults to CHISCSI

The iscsi_target_vendor_id is part of the device identification sent by an iSCSI target in response of a SCSI Inquiry request.

Chelsio T5/T4 Unified Wire For Linux Page 136

Chapter VIII. iSCSI PDU Offload Initiator

VIII. iSCSI PDU Offload Initiator

Chelsio T5/T4 Unified Wire For Linux Page 137

Chapter VIII. iSCSI PDU Offload Initiator

1. Introduction

The Chelsio T5/T4 series Adapters support iSCSI acceleration and iSCSI Direct Data

Placement (DDP) where the hardware handles the expensive byte touching operations, such as

CRC computation and verification, and direct DMA to the final host memory destination:

iSCSI PDU digest generation and verification

On transmitting, Chelsio hardware computes and inserts the Header and Data digest into the PDUs. On receiving, Chelsio hardware computes and verifies the Header and Data digest of the PDUs.

Direct Data Placement (DDP)

Chelsio hardware can directly place the iSCSI Data-In or Data-Out PDU's payload into preposted final destination host-memory buffers based on the Initiator Task Tag (ITT) in Data-In or Target Task Tag (TTT) in Data-Out PDUs.

PDU Transmit and Recovery

On transmitting, Chelsio hardware accepts the complete PDU (header + data) from the host driver, computes and inserts the digests, decomposes the PDU into multiple TCP segments if necessary, and transmit all the TCP segments onto the wire. It handles TCP retransmission if needed.

On receiving, Chelsio hardware recovers the iSCSI PDU by reassembling TCP segments, separating the header and data, calculating and verifying the digests, then forwarding the header to the host. The payload data, if possible, will be directly placed into the pre-posted host DDP buffer. Otherwise, the payload data will be sent to the host too.

The cxgb4i driver interfaces with open-iSCSI initiator and provides the iSCSI acceleration through Chelsio hardware wherever applicable.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with iSCSI PDU

Offload Initiator Software:

T520-LL-CR

T520-CR

T580-LP-CR

T540-CR

T420-CR

T440-CR

T422-CR

Chelsio T5/T4 Unified Wire For Linux Page 138

Chapter VIII. iSCSI PDU Offload Initiator

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the iSCSI PDU Offload Initiator software is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5 *

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6*

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27 *

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04 , 3.2.0-23

Kernel.org linux-2.6.35*

Kernel.org linux-3.4

Kernel.org linux-3.6

Kernel.org linux-3.7*

Other kernel versions have not been tested and are not guaranteed to work.

*Limited QA performed

Chelsio T5/T4 Unified Wire For Linux Page 139

Chapter VIII. iSCSI PDU Offload Initiator

2. Software/Driver Loading

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail.

Run the following command to load the driver:

[root@host]# modprobe cxgb4i

On SLES, load the driver with --allow option:

[root@host]# modprobe cxgb4i --allow

If installing the Chelsio Unified Wire package doesn’t install the cxgb4i driver properly on SLES, then follow these steps to configure your source tree and then re-install the package:

[root@host]# cd /usr/src/linux

[root@host]# make clean

[root@host]# make oldconfig

[root@host]# make scripts

Note

Sometimes open-iSCSI fails to compile because of dependencies on libopenssl-devel (for SLES11SP1) or openssl-devel(for RHEL 6) packages, so install these packages first and then install the Chelsio Unified Wire package (such issue is not seen while installing package using RPM).

If loading of cxgb4i displays "unkown symbols found" error in dmesg, then kill iscsid, unload all open iSCSI modules and related/dependent modules (like multipath, bnx2i etc) and then reload cxgb4i driver.

If loading the driver fails, then follow the steps mentioned below and try again. i. Run the following command to view all the loaded iSCSI modules

[root@host]# lsmod | grep iscsi

Chelsio T5/T4 Unified Wire For Linux Page 140

Chapter VIII. iSCSI PDU Offload Initiator ii. Now, unload them using the following command:

[root@host]# rmmod <modulename>

Chelsio T5/T4 Unified Wire For Linux Page 141

Chapter VIII. iSCSI PDU Offload Initiator

3. Software/Driver Unloading

To unload the driver, execute the following commands:

[root@host]# rmmod cxgb4i

[root@host]# rmmod libcxgbi

Chelsio T5/T4 Unified Wire For Linux Page 142

Chapter VIII. iSCSI PDU Offload Initiator

4. Software/Driver Configuration and Fine-tuning

4.1.

Accelerating open-iSCSI Initiator

The following steps need to be taken to accelerate the open-iSCSI initiator:

4.1.1. Configuring iscsid.conf file

Edit the iscsi/iscsid.conf file and change the setting for MaxRecvDataSegmentLength : node.conn[0].iscsi.MaxRecvDataSegmentLength = 8192

The login would fail for a normal session if MaxRecvDataSegmentLength is too big. A error message in the format of ERR! MaxRecvSegmentLength <X> too big. Need to be <=

<Y>. would be logged to dmesg.

Important

Always take a backup of iscsid.conf

file before installing Chelsio

Unified Wire Package. Although the file is saved to iscsid.rpmsave after uninstalling the package using RPM, you are still advised to take a backup.

In case of RHEL 6, uninstalling the package using RPM will result into uninstallation of open-iSCSI admin utility. In order to get back the utility, please install open-iscsi-utils provided with OS CD/DVD.

4.1.2. Configuring interface (iface) file

Create an interface file located under iface directory for the new transport class cxgb4i in the following format: iface.iscsi_ifacename = <iface file name> iface.hwaddress = <MAC address> iface.transport_name = cxgb4i iface.net_ifacename = <ethX> iface.ipaddress = <iscsi ip address>

Chelsio T5/T4 Unified Wire For Linux Page 143

Chapter VIII. iSCSI PDU Offload Initiator

E.g.:- iface.iscsi_ifacename = cxgb4i.00:07:43:04:5b:da iface.hwaddress = 00:07:43:04:5b:da iface.transport_name = cxgb4i iface.net_ifacename = eth3 iface.ipaddress = 102.2.2.137

Alternatively, you can create the file automatically by executing the following command:

[root@host]# iscsiadm -m iface

Here,

 iface.iscsi_ifacename

denotes the name of interface file in /etc/iscsi/ifaces/.

 iface.hwaddress

denotes the MAC address of the Chelsio interface via which iSCSI traffic will be running.

 iface.transport_name

denotes the transport name, which is cxgb4i.

 iface.net_ifacename

denotes the Chelsio interface via which iSCSI traffic will be running.

 iface.ipaddress

denotes the IP address which is assigned to the interface.

Note i. The interface file needs to be created in /etc/iscsi/iscsid.conf

. ii. If iface.ipaddress

is specified, it needs to be either the same as the ethX's IP address or an address on the same subnet. Make sure the IP address is unique in the network.

4.1.3. Discovery and Login

i. Starting iSCSI Daemon

Start Daemon from /sbin by using the following command:

[root@host]# iscsid -f -d 3

Note If iscsid is already running, then kill the service and start it as shown above after installing the Chelsio Unified Wire package.

Chelsio T5/T4 Unified Wire For Linux Page 144

Chapter VIII. iSCSI PDU Offload Initiator

ii. Discovering iSCSI Targets

To discovery an iSCSI target execute a command in the following format: iscsiadm -m discovery -t st -p <target ip address>:<target port no> -I

<cxgb4i iface file name>

E.g.:-

[root@host]# iscsiadm -m discovery -t st -p 102.2.2.155:3260 -I cxgb4i.00:07:43:04:5b:da iii. Logging into an iSCSI Target

Log into an iSCSI target using the following format: iscsiadm -m node -T <iqn name of target> -p <target ip address>:<target port no> -I <cxgb4i iface file name> -l

E.g.:-

[root@host]# iscsiadm -m node -T iqn.2004-05.com.chelsio.target1 -p

102.2.2.155:3260,1 -I cxgb4i.00:07:43:04:5b:da -l iv. Logging out from an iSCSI Target

Log out from an iSCSI Target by executing a command in the following format: iscsiadm -m node -T <iqn name of target> -p <target ip address>:<target port no> -I <cxgb4i iface file name> -u

E.g.:-

[root@host]# iscsiadm -m node -T iqn.2004-05.com.chelsio.target1 -p

102.2.2.155:3260,1 -I cxgb4i.00:07:43:04:5b:da -u

Note Other options can be found by typing iscsiadm --help

Chelsio T5/T4 Unified Wire For Linux Page 145

Chapter VIII. iSCSI PDU Offload Initiator

4.2.

Auto login from cxgb4i initiator at OS bootup

For iSCSI auto login (via cxgb4i) to work on OS startup, please add the following line to start() in /etc/rc.d/init.d/iscsid file on RHEL: modprobe -q cxgb4i

E.g.:- force_start() {

echo -n $"Starting $prog: "

modprobe -q iscsi_tcpmodprobe -q ib_iser

modprobe -q cxgb4i

modprobe -q cxgb3i

modprobe -q bnx2i

modprobe -q be2iscsi

daemon brcm_iscsiuio

daemon $prog

retval=$?

echo

[ $retval -eq 0 ] && touch $lockfile

return $retval

}

Chelsio T5/T4 Unified Wire For Linux Page 146

Chapter IX. FCoE Full Offload Initiator

IX. FCoE Full Offload Initiator

Chelsio T5/T4 Unified Wire For Linux Page 147

Chapter IX. FCoE Full Offload Initiator

1. Introduction

Fibre Channel over Ethernet (FCoE) is a mapping of Fibre Channel over selected full duplex

IEEE 802.3 networks. The goal is to provide I/O consolidation over Ethernet, reducing network complexity in the Datacenter.

Chelsio FCoE initiator maps Fibre Channel directly over Ethernet while being independent of the

Ethernet forwarding scheme. The FCoE protocol specification replaces the FC0 and FC1 layers of the Fibre Channel stack with Ethernet. By retaining the native Fibre Channel constructs,

FCoE will integrate with existing Fibre Channel networks and management software.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with FCoE full offload Initiator driver:

T520-LL-CR

T520-CR

T420-CR

T440-CR

T422-CR

T404-BT

T440-LP-CR

T420-LL-CR

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the FCoE full offload Initiator driver is available for the following versions:

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6*

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27*

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Kernel.org linux-2.6.35*

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

* Limited QA performed.

Chelsio T5/T4 Unified Wire For Linux Page 148

Chapter IX. FCoE Full Offload Initiator

2. Software/Driver Loading

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail.

To load the driver, execute the following:

[root@host]# modprobe csiostor

Note To load the driver on SLES11sp1 and SLES11sp2, execute the following command:

[root@host]# modprobe csiostor --allow

Chelsio T5/T4 Unified Wire For Linux Page 149

Chapter IX. FCoE Full Offload Initiator

3. Software/Driver Unloading

To unload the driver:

[root@host]# modprobe -r csiostor

Note If multipath services are running, unload of FCoE driver is not possible. Stop the multipath service and then unload the driver.

Chelsio T5/T4 Unified Wire For Linux Page 150

Chapter IX. FCoE Full Offload Initiator

4. Software/Driver Configuration and Fine-tuning

4.1.

Configuring Cisco Nexus 5010 switch

4.1.1. Configuring the DCBX parameters

Note

By default the Cisco Nexus switch enables DCBX functionality and configures

PFC for FCoE traffic making it no drop with bandwidth of 50% assigned to FCoE class of traffic and another 50% for the rest(like NIC). If you wish to configure

custom bandwidth, then follow the procedure below.

In this procedure, you may need to adjust some of the parameters to suit your environment, such as VLAN IDs, Ethernet interfaces, and virtual Fibre Channel interfaces.

To enable PFC, ETS, and DCB functions on a Cisco Nexus 5000 series switch: i. Open a terminal configuration setting. switch# config terminal switch(config)# ii. Configure qos class-maps and set the traffic priorities: NIC uses priority 0 and FcoE uses priority 3. switch(config)#class-map type qos class-nic switch(config-cmap-qos)# match cos 0 switch(config-cmap-qos)# class-map type qos class-fcoe switch(config-cmap-qos)# match cos 3 iii. Configure queuing class-maps. switch(config)#class-map type queuing class-nic switch(config-cmap-que)#match qos-group 2 iv. Configure network-qos class-maps. switch(config)#class-map type network-qos class-nic switch(config-cmap-nq)#match qos-group

Chelsio T5/T4 Unified Wire For Linux Page 151

Chapter IX. FCoE Full Offload Initiator v. Configure qos policy-maps. switch(config)#policy-map type qos policy-test switch(config-pmap-qos)#class type qos class-nic switch(config-pmap-c-qos)#set qos-group 2 vi. Configure queuing policy-maps and assign network bandwidth. Divide the network bandwidth between FcoE and NIC traffic. switch(config)#policy-map type queuing policy-test switch(config-pmap-que)#class type queuing class-nic switch(config-pmap-c-que)#bandwidth percent 50 switch(config-pmap-c-que)#class type queuing class-fcoe switch(config-pmap-c-que)#bandwidth percent 50 switch(config-pmap-c-que)#class type queuing class-default switch(config-pmap-c-que)#bandwidth percent 0 vii. Configure network-qos policy maps and set up the PFC for no-drop traffic class. switch(config)#policy-map type network-qos policy-test switch (config-pmap-nq)#class type network-qos class-nic switch(config-pmap-nq-c)#pause no-drop

Note

By default FCoE is set to pause no drop. In such a trade off, one may want to

set NIC to drop instead. viii. Apply the new policy (PFC on NIC and FcoE traffic) to the entire system. switch(config)#system qos switch(config-sys-qos)#service-policy type qos input policy-test switch(config-sys-qos)#service-policy type queuing output policy-test switch(config-sys-qos)#service-policy type queuing input policy-test switch(config-sys-qos)#service-policy type network-qos policy-test

4.1.2. Configuring the FCoE/FC ports

In this procedure, you may need to adjust some of the parameters to suit your environment, such as VLAN IDs, Ethernet interfaces, and virtual Fibre Channel interfaces

Chelsio T5/T4 Unified Wire For Linux Page 152

Chapter IX. FCoE Full Offload Initiator i. Following steps will enable FCoE services on a particular VLAN and does a VSAN-VLAN mapping. Need not do these steps every time, unless a new mapping has to be created. switch(config)# vlan 2 switch(config-vlan)# fcoe vsan 2 switch(config-vlan)#exit ii. Following steps help in creating a virtual fibre channel (VFC) and binds that VFC to a

Ethernet interface so that the Ethernet port begins functioning as a FCoE port. switch(config)# interface vfc 13 switch(config-if)#bind interface ethernet 1/13 switch(config-if)# no shutdown switch(config-if)#exit switch(config)#vsan database switch(config-vsan-db)# vsan 2 switch(config-vsan-db)#vsan 2 interface vfc 13 switch(config-vsan-db)#exit

Note

If you are binding the VFC to a MAC address instead of an ethernet port then

make sure the ethernet port is part of both default VLAN and FCoE VLAN. iii. Assign VLAN ID to the Ethernet port on which FCoE service was enabled in step1. switch(config)# interface ethernet 1/13 switch(config-if)#switchport mode trunk switch(config-if)#switchport trunk allowed vlan 2 switch(config-if)#no shutdown switch(config)#exit

Chelsio T5/T4 Unified Wire For Linux Page 153

Chapter IX. FCoE Full Offload Initiator iv. Enabling DCBX: switch(config)# interface ethernet 1/13 switch(config-if)# priority-flow-control mode auto switch(config-if)# flowcontrol send off switch(config-if)# flowcontrol receive off switch(config-if)# lldp transmit switch(config-if)# lldp receive switch(config-if)# no shutdown v. On the FC Ports, if a FC target is connected then perform the following steps - switch(config)#vsan database switch(config-vsan-db)#vsan 2 switch(config-vsan-db)# vsan 2 interface fc 2/2 switch(config-vsan-db)#exit switch(config)interface fc 2/2 switch(config-if)# switchport mode auto switch(config-if)# switchport speed auto switch(config-if)# no shutdown. vi. If you have not created a zone then make sure the default-zone permits the VSAN created, otherwise the initiator and the target on that particular VSAN although FLOGI’d into the switch will not talk to each other. To enable it, execute the below command. switch(config)# zone default-zone permit vsan 2

4.1.3. Configuring the Brocade 8000 switch

i. Configure LLDP for FCoE.Example of configuring LLDP for 10-Gigabit Ethernet interface. switch(config)#protocol lldp switch(conf-lldp)#advertise dcbx-fcoe-app-tlv switch(conf-lldp)#advertise dcbx-fcoe-logical-link-tlv

Chelsio T5/T4 Unified Wire For Linux Page 154

Chapter IX. FCoE Full Offload Initiator ii. Create a CEE Map to carry LAN and SAN traffic if it does not exist. Example of creating a

CEE map. switch(config)# cee-map default switch(conf-cee-map)#priority-group-table 1 weight 40 pfc switch(conf-cee-map)#priority-group-table 2 weight 60 switch(conf-cee-map)#priority-table 2 2 2 1 2 2 2 2 iii. Configure the CEE interface as a Layer 2 switch port. Example of configuring the switch port as a 10-Gigabit Ethernet interface. switch(config)#interface tengigabitethernet 0/16 switch(config-if-te-0/16)#switchport switch(config-if-te-0/16)#no shutdown switch(config-if)#exit iv. Create an FCoE VLAN and add an interface to it. Example of creating a FCoE VLAN and adding a single interface. switch(config)#vlan classifier rule 1 proto fcoe encap ethv2 switch(config)#vlan classifier rule 2 proto fip encap ethv2 switch(config)#vlan classifier group 1 add rule 1 switch(config)#vlan classifier group 1 add rule 2 switch(config)#interface vlan 1002 switch(conf-if-vl-1002 )#fcf forward switch(conf-if-vl-1002 )#interface tengigabitethernet 0/16 switch(config-if-te-0/16)#switchport switch(config-if-te-0/16)#switchport mode converged switch(config-if-te-0/16)#switchport converged allowed vlan add 1002 switch(config-if-te-0/16)#vlan classifier activate group 1 vlan 1002 switch(config-if-te-0/16)#cee default switch(config-if-te-0/16)#no shutdown switch(config-if-te-0/16)#exit

Note

Unlike cisco, only one VLAN ID can carry FCoE traffic for now on Brocade 8000.

It is their limitation.

Chelsio T5/T4 Unified Wire For Linux Page 155

Chapter IX. FCoE Full Offload Initiator v. Save the Configuration switch#copy running-config startup-config

4.2.

FCoE fabric discovery verification

The verification is done using “cxgbtool”.

4.2.1. Verifying the DCBX parameters

To verify the current DCBX information being exchanged, execute the below commands using cxgbtool. i. First step is to find the adapter number. Find it using the following command

[root@host]# cxgbtool stor –s ii. Now execute the following command to check the DCBX information.

[root@host]# cxgbtool stor –a <adapter_no> --dcb-params

Chelsio T5/T4 Unified Wire For Linux Page 156

Chapter IX. FCoE Full Offload Initiator

4.2.2. Verifying Local Ports

Once connected to the switch, use the following command to see if the FIP has gone through and a VN_Port MAC address has been assigned.

Verify if all the FCoE ports are online/ready and a successful FIP has taken place using the following command. The wwpn and state of the initiator local port can be found under sysfs.

[root@host]# cat /sys/class/fc_host/hostX/port_name

Chelsio T5/T4 Unified Wire For Linux Page 157

Chapter IX. FCoE Full Offload Initiator

Note

The hosts under fc_host depends on the number of ports on the adapter

used.

Inorder to identify chelsio fc_host from other vendor fc_host, the WWPN always begins with 0x5000743

Alternatively, the local port information can also be found using:

[root@host]# cxgbtool stor –a <adapter_no> --show-lnode

Chelsio T5/T4 Unified Wire For Linux Page 158

Chapter IX. FCoE Full Offload Initiator

Chelsio T5/T4 Unified Wire For Linux Page 159

Chapter IX. FCoE Full Offload Initiator

4.2.3. Verifying the target discovery

To check the targets being discovered use cxgbtool. To check the list of targets that are being discovered from a particular FCoE port, use the following commands from cxgbtool. i. Check for the adapter number using the following command.

[root@host]# cxgbtool stor –s ii. To check the list of targets discovered from a particular FCoE port, first find out the wwpn of the initiator local port under sysfs. The hosts under fc_host depends on the number of ports on the adapter used.

[root@host]# cat /sys/class/fc_host/hostX/port_name iii. After finding the localport, go to the corresponding Remote port under sysfs # cat

/sys/class/fc_remote_ports/rport-X:B:R where X is the Host ID, B is the bus ID and R is the remote port.

Note

R can correspond to NameServer, Management Server and other initiator ports logged in to the switch and targets.

Alternatively the locaports can also be found using cxgbtool

[root@host]# cxgbtool stor –a <adapter no> --show-lnode

Chelsio T5/T4 Unified Wire For Linux Page 160

Chapter IX. FCoE Full Offload Initiator

After finding out the wwpn of the local node, to verify the list of targets being discovered, use the following command.

[root@host]# cxgbtool stor –a <adapter_no> --show-rnode --wwn=<wwpn of lnode>

Chelsio T5/T4 Unified Wire For Linux Page 161

Chapter IX. FCoE Full Offload Initiator

4.3.

Formatting the LUNs and Mounting the Filesystem

Use lsscsi –g to list the LUNs discovered by the initiator

[root@host]# lsscsi –g

Chelsio T5/T4 Unified Wire For Linux Page 162

Chapter IX. FCoE Full Offload Initiator

Alternatively, the LUNs discovered by the Chelsio FCoE initiators can be accessed via easilyidentifiable ‘udev’ path device files like:

[root@host]# ls /dev/disk/by-path/pci-0000:04:00.0-csio-fcoe

<local_wwpn>:<remote_wwpn>:<lun_wwn>

Chelsio T5/T4 Unified Wire For Linux Page 163

Chapter IX. FCoE Full Offload Initiator

4.4.

Creating Filesystem

Create an ext3 filesystem using the following command:

[root@host]# mkfs.ext3 /dev/sdx

4.5.

Mounting the formatted LUN

The formatted LUN can be mounted on the specified mountpoint using the following command:

[root@host]# mount /dev/sdx /mnt

Chelsio T5/T4 Unified Wire For Linux Page 164

Chapter IX. FCoE Full Offload Initiator

Chelsio T5/T4 Unified Wire For Linux Page 165

Chapter X. Offload Bonding driver

X. Offload Bonding driver

Chelsio T5/T4 Unified Wire For Linux Page 166

Chapter X. Offload Bonding driver

1. Introduction

The Chelsio Offload bonding driver provides a method to aggregate multiple network interfaces into a single logical bonded interface effectively combining the bandwidth into a single connection. It also provides redundancy in case one of link fails.

The traffic running over the bonded interface can be fully offloaded to the T5/T4 Adapter, thus freeing the CPU from TCP/IP overhead.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with the Chelsio

Offload Bonding driver:

T520-LL-CR

T520-CR

T580-LP-CR

T540-CR

T420-CR

T440-CR

T422-CR

T420-SO-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the Offload Bonding driver is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Chelsio T5/T4 Unified Wire For Linux Page 167

Chapter X. Offload Bonding driver

Ubuntu 12.04, 3.2.0-23

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

*Limited QA performed

Chelsio T5/T4 Unified Wire For Linux Page 168

Chapter X. Offload Bonding driver

2. Software/Driver Loading

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail.

To load the driver (with offload support), run the following command:

[root@host]# modprobe bonding

Chelsio T5/T4 Unified Wire For Linux Page 169

Chapter X. Offload Bonding driver

3. Software/Driver Unloading

To unload the driver, run the following command:

[root@host]# rmmod bonding

Chelsio T5/T4 Unified Wire For Linux Page 170

Chapter X. Offload Bonding driver

4. Software/Driver Configuration and Fine-tuning

4.1.

Offloading TCP traffic over a bonded interface

To offload TCP traffic over a bonded interface, use the following method: i. Load the network driver with TOE support:

[root@host]# modprobe t4_tom ii. Create a bonded interface:

[root@host]# modprobe bonding mode=1 miimon=100 iii. Bring up the bonded interface and enslave the interfaces to the bond:

[root@host]# ifconfig bond0 up

[root@host]# ifenslave bond0 ethX ethY iv. Assign IPv4/IPv6 address to the bonded interface:

[root@host]# ifconfig bond0 X.X.X.X/Y

All TCP traffic will be offloaded over the bonded interface now.

Chelsio T5/T4 Unified Wire For Linux Page 171

Chapter XI. UDP Segmentation Offload and Pacing

XI. UDP Segmentation Offload and Pacing

Chelsio T5/T4 Unified Wire For Linux Page 172

Chapter XI. UDP Segmentation Offload and Pacing

1. Introduction

Chelsio’s T5/T4 series of adapters provide UDP segmentation offload and per-stream rate shaping to drastically lower server CPU utilization, increase content delivery capacity, and improve service quality.

Tailored for UDP content, UDP Segmentation Offload (USO) technology moves the processing required to packetize UDP data and rate control its transmission from software running on the host to the network adapter. USO increases performance and dramatically reduces CPU overhead, allowing significantly higher capacity using the same server hardware. Without USO support, UDP server software running on the host needs to packetize payload into frames, process each frame individually through the network stack and schedule individual frame transmission, resulting in millions of system calls, and packet traversals through all protocol layers in the operating system to the network device. In contrast, USO implements the network protocol stack in the adapter, and the host server software simply hands off unprocessed UDP payload in large I/O buffers to the adapter.

The following figure compares the traditional datapath on the left to the USO datapath on the right, showing how per-frame processing is eliminated. In this example, the video server pushes

5 frames at a time. In an actual implementation, a video server pushes 50 frames or more in each I/O, drastically lowering the CPU cycles required to deliver the content.

Pacing is beneficial for several reasons, one example is for Content Delivery Networks

(CDNs)/Video On Demand (VOD) providers to avoid receive buffer overflows, smooth out network traffic, or to enforce Service Level Agreements (SLAs). Without dedicated hardware based pacing support, the video server must perform this in software, which is a CPU intensive task and can be prohibitive at 10Gb and higher rates.

Chelsio T5/T4 Unified Wire For Linux Page 173

Chapter XI. UDP Segmentation Offload and Pacing

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with the UDP

Segmentation Offload and Pacing driver.

T520-LL-CR

T520-CR

T580-LP-CR

T420-CR

T440-CR

T422-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the UDP Segmentation Offload and Pacing driver is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04, 3.2.0-23

Kernel.org linux-2.6.35

Kernel.org linux-3.4

Kernel.org linux-3.6

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

Chelsio T5/T4 Unified Wire For Linux Page 174

Chapter XI. UDP Segmentation Offload and Pacing

2. Software/Driver Loading

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail.

Run the following commands to load the driver:

[root@host]# modprobe cxgb4

[root@host]# modprobe t4_tom

Though normally associated with the Chelsio TCP Offload engine, the t4_tom module is required in order to allow for the proper redirection of UDP socket calls.

Chelsio T5/T4 Unified Wire For Linux Page 175

Chapter XI. UDP Segmentation Offload and Pacing

3. Software/Driver Unloading

Reboot the system to unload the driver.

Chelsio T5/T4 Unified Wire For Linux Page 176

Chapter XI. UDP Segmentation Offload and Pacing

4. Software/Driver Configuration and Fine-tuning

4.1.

Modifying the application

To use the UDP offload functionality, the application needs to be modified. Follow the steps mentioned below: i. Determine the UDP socket file descriptor in the application through which data is sent ii. Declare and initialize two variables in the application: int fs=1316; int cl=1;

Here,

fs is the UDP packet payload size in bytes that is transmitted on the wire. The minimum value of fs is 256 bytes.

cl is the UDP traffic class(scheduler-class-index) that the user wishes to assign the data stream to. This value needs to be in the range of 0 to 14.

The application will function according to the parameters set for that traffic class. iii. Add socket option definitions:

In order to use setsockopt() to set the options to the UDP socket, the following three definitions need to be made:

SO_FRAMESIZE used for setting frame size, which has the value 291.

SOL_SCHEDCLASS used for setting UDP traffic class, which has the value 290.

IPPROTO_UDP used for setting the type of IP Protocol.

# define SO_FRAMESIZE 291

# define SOL_SCHEDCLASS 290

# define IPPROTO_UDP 17 iv. Use the setsockopt() function to set socket options:

//Get the UDP socket descriptor variable setsockopt (sockfd , IPPROTO_UDP, SO_FRAMESIZE, &fs, sizeof(fs)); setsockopt (sockfd , IPPROTO_UDP, SOL_SCHEDCLASS, &cl, sizeof(cl));

Chelsio T5/T4 Unified Wire For Linux Page 177

Chapter XI. UDP Segmentation Offload and Pacing

Here:

sockfd : The file descriptor of the UDP socket

&fs / &cl : Pointer to the framesize and class variables

sizeof(fs) / sizeof(cl) : The size of the variables v. Now, compile the application.

4.1.1. UDP offload functionality for RTP data

In case of RTP data, the video server application sends the initial sequence number and the

RTP payload. The USO engine segments the payload data, increments the sequence number and sends out the data.

In order to use the UDP offload functionality for RTP data, make the following additions to the steps mentioned above:

1. In step (ii), declare and initialize a new variable in the application: int rtp_header_size=16;

Here, rtp_header_size is the RTP header size in bytes that the application sends.

2. In step (iii), define a new macro, UDP_RTPHEADERLEN used for setting RTP header length with the value 292.

# define UDP_RTPHEADERLEN 292

3. In step (iv), define a new socket option: setsockopt (sockfd,17,UDP_RTPHEADERLEN,&rtp_header_size, sizeof(rtp_header_size));

Here,

&rtp_header_size : pointer to the RTP header length variable

sizeof(rtp_header_size) : the size of the RTP header length variable

Chelsio T5/T4 Unified Wire For Linux Page 178

Chapter XI. UDP Segmentation Offload and Pacing

4.2.

Configuring UDP Pacing

Now that the application has been modified to associate the application’s UDP socket to a particular UDP traffic class, the pacing of that socket’s traffic can be set using the cxgbtool utility. The command and its parameters are explained below:

[root@host]# cxgbtool <ethX> sched-class params type packet level cl-rl mode flow rate-unit bits rate-mode absolute channel <Channel No.> class

<scheduler-class-index> max-rate <maximum-rate> pkt-size <Packet size>

Here,

ethX is the Chelsio interface

Channel No. is the port on which data is flowing (0-3)

scheduler-class-index is the UDP traffic class (0-14) set in the SOL_SCHEDCLASS socket option in the application in section 4.1.

maximum-rate is the bit rate (Kbps) for this UDP stream. This value should be in the range of 50 Kbps to 50 Mbps.

Packet size is the UDP packet payload size in bytes; it should be equal to the value set in the SO_FRAMESIZE socket option in the application in section 4.1.

Example:

The user wants to transfer UDP data on port 0 of the adapter using the USO engine. The application has been modified as shown in section 4.1. In order to set a bit rate of 10Mbps for traffic class 1 with payload size of 1316 on port 0, the following invocation of cxgbtool is used:

[root@host]# cxgbtool ethX sched-class params type packet level cl-rl mode flow rate-unit bits rate-mode absolute channel 0 class 1 max-rate

10000 pkt-size 1316

Note

To get an accurate bit rate per class, data sent by the application to the sockets should be a multiple of the value set for the “pkt-size” parameter.

In above example, IO size sent by application should be a multiple of 1316.

Note

Linux Unified Wire currently supports 10240 offload UDP connections. If the application needs to establish more than 10240 UDP connections, it can check the return code of ENOSPC from a send() or sendto() call and close

this socket and open a new one that uses the kernel UDP stack.

Chelsio T5/T4 Unified Wire For Linux Page 179

Chapter XII. Offload IPv6 driver

XII. Offload IPv6 driver

Chelsio T5/T4 Unified Wire For Linux Page 180

Chapter XII. Offload IPv6 driver

1. Introduction

The growth of the Internet has created a need for more addresses than are possible with IPv4.

Internet Protocol version 6 (IPv6) is a version of the Internet Protocol (IP) designed to succeed the Internet Protocol version 4 (IPv4).

Chelsio’s Offload IPv6 feature provides support to fully offload IPv6 traffic to the T5/T4 adapter.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with Chelsio

Offload IPv6 feature:

T520-LL-CR

T520-CR

T580-LP-CR

T540-CR

T420-CR

T440-CR

T422-CR

T420-SO-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the Offload IPv6 feature is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04, 3.2.0-23

Chelsio T5/T4 Unified Wire For Linux Page 181

Chapter XII. Offload IPv6 driver

Kernel.org linux-2.6.35

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

*Limited QA performed

Chelsio T5/T4 Unified Wire For Linux Page 182

Chapter XII. Offload IPv6 driver

2. Software/Driver Loading

IPv6 must be enabled in your system (enabled by default) to use the Offload IPv6 feature.Also,

Unified Wire package must be installed with IPv6 support (see Software/Driver Installation ).

After installing Unified Wire package and rebooting the host, load the NIC (cxgb4) and TOE

(t4_tom) drivers. The drivers must be loaded by the root user. Any attempt to load the drivers as a regular user will fail.

[root@host]# modprobe cxgb4

[root@host]# modprobe t4_tom

All the IPv6 traffic will be offloaded now.

Chelsio T5/T4 Unified Wire For Linux Page 183

Chapter XII. Offload IPv6 driver

3. Software/Driver Unloading

To disable Offload IPv6 feature, unload NIC and TOE drivers:

3.1.

Unloading the NIC driver

To unload the NIC driver, run the following command:

[root@host]# rmmod cxgb4

3.2.

Unloading the TOE driver

Please reboot the system to unload the TOE driver.

Chelsio T5/T4 Unified Wire For Linux Page 184

Chapter XIII. Bypass Driver

XIII. Bypass Driver

Chelsio T5/T4 Unified Wire For Linux Page 185

Chapter XIII. Bypass Driver

1. Introduction

Chelsio’s B420 and B404 Bypass Adapters are Ethernet cards that provide bypass functionality and an integrated L2, L3, and L4 Ethernet switch. The integrated switch allows for selective bypass on a per-packet basis at line rate.

To use the bypass adapters, you must have both the Chelsio NIC driver and the bypass CLI user space application loaded.

1.1.

Features

B404 is a four-port 1G and B420 is a two-port 10G, short-range or long range, low profile PCI-

Express Host Bus Adapters (HBAs) using T4 ASIC.

Bypass

Software programmable behavior on power fails – either Bypass Mode or Drop

Mode.

Firmware control of Bypass / Normal / Drop Modes when T4 timer expires.

Bypass control via software

Drop Mode control by putting the associated PHY in reset.

Selective Bypass – Programmable HW traffic classification and redirection without host intervention in normal mode

Product function

:

The Bypass Adapters can operate in 4 modes.

Bypass Mode

Normal Mode

Disconnect Mode

Selective Bypass Mode

Bypass Mode

In this mode the Bypass adapter switches the packet from one port to another port. That is, in

B420, port 0 to port 1 and vice versa; and in B404, port 0 to port 1 and port 2 to port 3 and vice versa.

Normal Mode

The Bypass Adapters can be programmed to function as a NIC. In this mode all the packets are redirected to the host.

Chelsio T5/T4 Unified Wire For Linux Page 186

Chapter XIII. Bypass Driver

Disconnect Mode

The Bypass cards can also be programmed to drop all the packets.

Selective Bypass

In Normal mode, the Bypass Adapters can be programmed to perform redirection of packets depending on the certain portion of the packet. The specification of the match criteria is called a

rule. When a rule is matched an action is applied to the ingress packet. The actions that are supported are drop, forward and input.

The drop action causes the packet to be discarded.

The forward action causes the packet to bypass the host from any port to any port.

Finally, the input action directs the packet to the host where it can be processed by the application.

1.2.

Hardware Requirements

1.2.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with Chelsio

Bypass driver:

B404-BT

B420-SR

Chelsio T5/T4 Unified Wire For Linux Page 187

Chapter XIII. Bypass Driver

1.3.

Software Requirements

1.3.1. Linux Requirements

Currently the Bypass driver is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5 *

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5 *

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27 *

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04, 3.2.0-23

Kernel.org linux-2.6.35*

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

* Limited QA performed.

Chelsio T5/T4 Unified Wire For Linux Page 188

Chapter XIII. Bypass Driver

2. Software/Driver Loading

The driver must be loaded by the root user. Any attempt to load the driver as a regular user will fail.

Run the following command to load the Bypass driver:

[root@host]# modprobe cxgb4

Chelsio T5/T4 Unified Wire For Linux Page 189

Chapter XIII. Bypass Driver

3. Software/Driver Unloading

Run the following command to unload the Bypass driver:

[root@host]# rmmod cxgb4

Chelsio T5/T4 Unified Wire For Linux Page 190

Chapter XIII. Bypass Driver

4. Software/Driver Configuration and Fine-tuning

4.1.

Starting ba server

4.1.1. For IPv4 only

Execute the following command to start the ba server only for IPv4:

[root@host]# ba_server –i ethX

4.1.2. For IPv4 and IPv6

Execute the following command to start the ba server for IPv4 and IPv6:

[root@host]# ba_server -6 –i ethX

4.2.

Bypass API (CLI)

A CLI will be created that implements the Bypass API as specified below. This CLI will then communicate the requests to the SDK server. The API will contain the following elements:

Bypass Management (watchdog, state)

Redirect Management

Bypass management provides a means of setting the watchdog timeout as well as enabling and disabling it. It allows setting a default state and a current state.

The redirect management element will interface with the SDK server to take the appropriate actions to manage tables and rules.

4.2.1. Bypass Management

The bypass CLI will have the following syntax:

[root@host]# bypass ethX get|set --key [value]

Chelsio T5/T4 Unified Wire For Linux Page 191

Chapter XIII. Bypass Driver

E.g.

[root@host]# bypass eth1 set --current_state normal

Here is a list of Bypass commands:

Bypass command list

Command bypass ethX get bypass ethX set bypass ethX get bypass bypass bypass ethX ethX ethX

set

get

set

Key default_state default_state current_state current_state watchdog watchdog bypass ethX get watchdog_timeout bypass ethX set watchdog_timeout

Value bypass|disconnect bypass|disconnect| normal enable|disable|lock| ping

Return bypass|disconnect bypass|disconnect| normal enabled|disabled timeout in milliseconds timeout in milliseconds (maximum is 60000)

4.2.2. Operating the Switch (Examples)

The switch has three modes of operation; bypass, disconnect, and normal. These three modes can be selected using the bypass CLI. The mode can be selected for the current operating state or for the default state that is applied when powering up or after failure detection.

4.2.3. Bypass State

To obtain the current state of the switch:

[root@host]# bypass ethX get --current_state

To set the state of the switch:

[root@host]# bypass ethX set --current_state [bypass|disconnect|normal]

Chelsio T5/T4 Unified Wire For Linux Page 192

Chapter XIII. Bypass Driver

Getting the default state:

[root@host]# bypass ethX get --default_state

Setting the default state:

[root@host]# bypass ethX set --default_state [bypass |disconnect]

4.2.4. Using the bypass watchdog timer

The watchdog timer is used to ensure that if there is a software failure, the switch will enter the default state.

First set the timeout value for the watchdog in milliseconds:

[root@host]# bypass ethX set --watchdog_timeout 3000

Start pinging the watchdog:

[root@host]# while true; do bypass ethX set --watchdog ping; sleep 1; done

Enable the watchdog:

[root@host]# bypass ethX set --watchdog enable

4.2.5. Redirect Management

The bypass adapter has the ability to redirect packets based upon rules. Rules can be grouped into tables. When a rule is added into a table, it is assigned an index. Within a table, rules are evaluated by index number from low to high. When a rule matches the action associated with it, that rule is performed and higher number rules are not evaluated. Rules are not evaluated until a table is activated.

Chelsio T5/T4 Unified Wire For Linux Page 193

Chapter XIII. Bypass Driver

The redirect CLI has the following syntax:

[root@host]# redirect ethX command --key [value] …

Redirect Command List

Command Key Value Return redirect list ethX redirect ethX add update match list of all configured tables and rules table index proto table id (defaults to table 1)

Add a rule to a table.

Update the specified rule with new keys.

Match specified keys to a rule in a table. rule index. If not specified the rule is appended. Otherwise it is inserted at the specified index. icmp|icmp6|tcp|udp|any

(defaults to any) srcaddr dstaddr vlan action port srcmask

Source IP address

Destination IP address

VLAN id input|forward|drop (default input)

If action is “input”, then this is the port to which the packet is directed.

Source IP address mask

(Default is 255.255.255.255) dstmask Destination IP address mask

(Default is 255.255.255.255) srcport

Source port Number dstport

Destination Port number srcportmask

Source port Number Mask* dstportmask

Destination port Number

Mask* ipv6

Use this option if IPv6 address are used while srcaddr6 adding Redirect rule

IPv6 Source IP address dstaddr6

IPv6 Destination IP address srcmask6 dstmask6

IPv6 Source IP address mask

(Default:

FFFF:FFFF:FFFF:FFFF:FFF

F:FFFF:FFFF:FFFF)

IPv6 Destination IP address mask

(Default:FFFF:FFFF:FFFF:F

FFF:FFFF:FFFF:FFFF:FFFF)

Chelsio T5/T4 Unified Wire For Linux Page 194

Chapter XIII. Bypass Driver redirect ethX delete redirect ethX purge redirect ethX move table index table table redirect ethX count old_id new_id table index redirect ethX create_table redirect ethX activate_table redirect ethx deactivate_table redirect ethX delete_table redirect ethX dump table table table table table id rule index table id table id

Existing rule index.

New rule index. table id rule index table id table id table id table id

Delete the table

Delete a rule from a table.

Remove all rules from the specified table

Check that the new rule id is valid and move the source rule to new id if it doesn’t contain any existing rule

Display a count of the number of packets and bytes has matched a rule.

Create a new table that is used to hold a set of rules.

Activate a table so that all the rules in it will be active.

Deactivate a table and its associated rules.

Delete a table and the rules in it. dump all tables and rules as commands

* The matching algorithm for srcportmask and dstportmask is

(Ingress_Packet-Field & Filter-Mask) == (Filter-Value & Filter-Mask)

4.2.6. Managing persistent tables and rules

This example creates a new table with table id 1 and then adds a rule to drop all ICMP packets for port 0. The rules in a table are not active until the table is activated.

Create a table:

[root@host]# redirect ethX create_table --table 1

Activate the table:

[root@host]# redirect ethX activate_table --table 1

Add a rule to the newly created table:

[root@host]# redirect ethX add --table 1 --proto icmp –-action drop –-port 0

Chelsio T5/T4 Unified Wire For Linux Page 195

Chapter XIII. Bypass Driver

The redirect dump command can be used to save the currently configured tables and rules into a shell script.

To make the current configured rules & tables persistent, redirect the output to /etc/ba.cfg

file only:

[root@host]# redirect ethX dump > /etc/ba.cfg where the /etc/ba.cfg

is read by the bad service at boot time.

To apply the saved configuration after machine reboots, start the bad service. This service is available only in the IPv4 mode. For IPv6, the ba_server needs to be started manually.

[root@host]# service bad start

To stop the service:

[root@host]# service bad stop

To restart the service:

[root@host]# service bad restart

Chelsio T5/T4 Unified Wire For Linux Page 196

Chapter XIV. WD Sniffing and Tracing

XIV. WD Sniffing and Tracing

Chelsio T5/T4 Unified Wire For Linux Page 197

Chapter XIV. WD Sniffing and Tracing

1. Theory of Operation

The objective of these utilities (wd_sniffer and wd_tcpdump_trace) is to provide sniffing and tracing capabilities by making use of T4's hardware features.

Sniffer- Involves targeting specific multicast traffic and sending it directly to user space. a) Get a Queue (raw QP) idx. b) Program a filter to redirect specific traffic to the raw QP queue.

Tracer - All tapped traffic is forwarded to user space and also pushed back on the wire via the internal loop back mechanism a) Get a Queue (raw QP) idx b) Set the T4 adapter in loop back c) Connect Client A and B to ports 0 and 1 or ports 2 and 3. d) Enable tracing.

In either mode the targeted traffic bypasses the kernel TCP/IP stack and is delivered directly to user space by means of an RX queue.

Chelsio T5/T4 Unified Wire For Linux Page 198

Chapter XIV. WD Sniffing and Tracing

Schematic diagram of T4 sniffer and tracer

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with the tools:

T420-CR

T440-CR

T422-CR

T404-BT

T420-BCH

Chelsio T5/T4 Unified Wire For Linux Page 199

Chapter XIV. WD Sniffing and Tracing

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the WD Sniffing and Tracing utility is available for the following version:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1),2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27 *

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Kernel.org linux-2.6.35

Kernel.org linux-3.4

Kernel.org linux-3.6*

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

*Limited QA performed

Chelsio T5/T4 Unified Wire For Linux Page 200

Chapter XIV. WD Sniffing and Tracing

2. Installation and Usage

2.1.

Installing basic support

iw_cxgb4 (Chelsio iWARP driver) and cxgb4 (Chelsio NIC driver) drivers have to be compiled and loaded before running the utilities. Refer to the Software/Driver Loading section for each driver and follow the instructions mentioned before proceeding.

2.2.

Using Sniffer (wd_sniffer)

1. Setup:

Wire filter sniffing requires 2 systems with one machine having a T4 card.

The machines should be setup in the following manner:

Machine A <---------> Machine B

192.168.1.100 192.168.1.200

2. Procedure:

On the Device Under Test (DUT), start sniffer.

[root@host]# wd_sniffer -T 20 -s 1000 -I <MAC address of interface to sniff>

Start traffic on the PEER and watch the sniffer.

The sniffer will receive all packets as fast as possible, update the packet count, and then discard the data. Performance is a full 10Gbps for packet size 1000.

2.3.

Using Tracer (wd_tcpdump_trace)

1. Setup:

Wire tapping requires 3 systems with one machine having a T4 card with two or more ports. The machines should be setup in the following manner:

Chelsio T5/T4 Unified Wire For Linux Page 201

Chapter XIV. WD Sniffing and Tracing

DUT: Machine B

PEER: Machine A <-----> (port 0) (port 1) <-----> PEER: Machine C

192.168.1.100 IP-dont-care IP-dont-care 192.168.1.200

2. Procedure:

Run wd_tcpdump_trace -i iface on the command prompt where iface is one of the interfaces whose traffic you want to trace. In the above diagram its port 0 or port 1.

[root@host]# wd_tcpdump_trace -i <iface>

Try ping or ssh between machines A and B. The traffic should successfully make it from end to end and wd_tcpdump_trace on the DUT should show the tapped traffic.

Chelsio T5/T4 Unified Wire For Linux Page 202

Chapter XV. Classification and Filtering

XV. Classification and Filtering

Chelsio T5/T4 Unified Wire For Linux Page 203

Chapter XV. Classification and Filtering

1. Introduction

Classification and Filtering feature enhances network security by controlling incoming traffic as they pass through network interface based on source and destination addresses, protocol, source and receiving ports, or the value of some status bits in the packet. This feature can be used in the ingress path to:

Steer ingress packets that meet ACL (Access Control List) accept criteria to a particular receive queue.

Switch (proxy) ingress packets that meet ACL accept criteria to an output port, with optional header rewrite.

Drop ingress packets that fail ACL accept criteria.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with the

Classification and Filtering feature:

T520-LL-CR

T520-SO-CR

T520-CR

T540-CR

T580-LP-CR

T420-CR

T440-CR

T422-CR

T420-SO-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the Classification and Filtering feature is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Chelsio T5/T4 Unified Wire For Linux Page 204

Chapter XV. Classification and Filtering

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04, 3.2.0-23

Kernel.org linux-2.6.35

Kernel.org linux-3.4

Kernel.org linux-3.6

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

Chelsio T5/T4 Unified Wire For Linux Page 205

Chapter XV. Classification and Filtering

2. Usage

2.1.

Configuration

The Classification and Filtering feature is configured by specifying the filter selection combination set in the T4 firmware configuration file t4-config.txt located in /lib/firmware/cxgb4/

The following combination is set by default and packets will be matched accordingly: filterMode = fragmentation, mpshittype, protocol, vlan, port, fcoe

Where,

fragmentation: Fragmented IP packets

mpshittype : MAC address "match type" (0=unicast, 1=unicast hash, 2=multicast, 3=multicast hash, 4=PROM, 5=hyper PROM, 6=broadcast, 7=none)

protocol : IP protocol number (ICMP=1, TCP=6, UDP=17, etc)

vlan : Inner VLAN Tag

port : Packet ingress port number

fcoe : Fibre Channel over Ethernet frames

2.2.

Creating Filter Rules

Network driver (cxgb4 ) must be installed and loaded before setting the filter rule. If you haven’t done already, run the Unified Wire Installer with the appropriate T4 configuration tuning option to install the Network Driver.

Next, run the following command to load the network driver:

[root@host]# modprobe cxgb4

Now, create filter rules using cxgbtool:

[root@host]# cxgbtool ethx filter <index> action [pass/drop/switch]

Chelsio T5/T4 Unified Wire For Linux Page 206

Chapter XV. Classification and Filtering

Where,

ethX : Chelsio interface

index : positive integer set as filter id

action : Ingress packet disposition

pass : Ingress packets will be passed through set ingress queues

switch : Ingress packets will be routed to an output port with optional header rewrite.

drop : Ingress packets will be dropped.

Note

In case of multiple filter rules, the rule with the lowest filter index takes higher priority.

2.2.1. Examples

Drop action

[root@host]# cxgbtool ethX filter 0 action drop fip 192.168.1.5

The above filter rule will drop all ingress packets from IP 192.168.1.5

Pass action

[root@host]# cxgbtool ethX filter 0 action pass lport 10001 fport 355 queue 2

The above filter rule will pass all ingress packets that match local port 10001 and remote port

355 to ingress queue 2 for load balancing.

Switch action

[root@host]# cxgbtool ethX filter 0 action switch iport 0 eport 1 vlan 3

The above filter rule will route all ingress packets that match VLAN id 3 from port 0 of Chelsio adapter to port 1.Remaining packets will be sent to the host.

Chelsio T5/T4 Unified Wire For Linux Page 207

Chapter XV. Classification and Filtering

For offloaded ingress packets, use the prior argument with the above command:

[root@host]# cxgbtool ethx filter <index> action <pass/drop/switch> prio 1

Note

For more information on additional parameters, refer to cxgbtool manual by running the man cxgbtool command

2.3.

Listing Filter Rules

To list the filters set, run the following command:

[root@host]# cxgbtool ethX filter show

2.4.

Removing Filter Rules

To remove a filter, run the following command with the corresponding filter rule index

[root@host]# cxgbtool ethX filter index <delete|clear>

Note

For more information on additional parameters, refer to cxgbetool manual by running the man cxgbtool command

Chelsio T5/T4 Unified Wire For Linux Page 208

XVI. Traffic Management

XVI. Traffic Management

Chelsio T5/T4 Unified Wire For Linux Page 209

XVI. Traffic Management

1. Introduction

Traffic Management capabilities built-in to Chelsio T5/T4 CNAs can shape transmit data traffic through the use of sophisticated queuing and scheduling algorithms built-in to the ASIC hardware which provides fine-grained software control over latency and bandwidth parameters such as packet rate and byte rate. These features can be used in a variety of data center application environments to solve traffic management problems.

Traffic Management features in Chelsio ‘s adapters allows the user to control three main things:

Guarantee low latency in the presence of other traffic

Control max bandwidth that a connection or a flow (a group of connections) can use

Allocate available bandwidth to several connection or flows based on desired levels of performance

Once the offload transmit traffic shaping classes have been configured, individual offloaded connections (flows) may be assigned to a traffic shaping class in order to manage the flows according to the class configuration. The mechanism to accomplish this "flow to class" mapping assignment is the Connection Offload Policy (COP) configuration system.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with the Traffic

Management feature.

T520-LL-CR

T520-CR

T580-LP-CR

T420-CR

T440-CR

T422-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

Chelsio T5/T4 Unified Wire For Linux Page 210

XVI. Traffic Management

1.2.

Software Requirements

1.2.1. Linux Requirements

Currently the Traffic Management feature is available for the following versions:

Redhat Enterprise Linux 5 update 7 kernel (RHEL5.7), 2.6.18-274.el5

Redhat Enterprise Linux 5 update 8 kernel (RHEL5.8), 2.6.18-308.el5

Redhat Enterprise Linux 6 update 3 kernel (RHEL6.3), 2.6.32-279.el6

Redhat Enterprise Linux 6 update 4 kernel (RHEL6.4), 2.6.32-358.el6

Suse Linux Enterprise Server 11 SP1 kernel (SLES11SP1), 2.6.32.12-0.7

Suse Linux Enterprise Server 11 SP2 kernel (SLES11SP2), 3.0.13-0.27

Suse Linux Enterprise Server 11 SP3 kernel (SLES11SP3), 3.0.76-0.11

Ubuntu 12.04, 3.2.0-23

Kernel.org linux-2.6.35

Kernel.org linux-3.4

Kernel.org linux-3.6

Kernel.org linux-3.7

Other kernel versions have not been tested and are not guaranteed to work.

Chelsio T5/T4 Unified Wire For Linux Page 211

XVI. Traffic Management

2. Software/Driver Loading

Traffic Management can be performed on non-offloaded connections as well as on offloaded connections.

The drivers must be loaded by the root user. Any attempt to load the drivers as a regular user will fail.

Run the following commands to load the drivers:

[root@host]# modprobe cxgb4

[root@host]# modprobe t4_tom

Chelsio T5/T4 Unified Wire For Linux Page 212

XVI. Traffic Management

3. Software/Driver Unloading

Reboot the system to unload the driver.

Chelsio T5/T4 Unified Wire For Linux Page 213

XVI. Traffic Management

4. Software/Driver Configuration and Fine-tuning

4.1.

Traffic Management Rules

Traffic Management supports the following types of scheduler hierarchy levels which can be configured using the cxgbtool utility: i. Class Rate Limiting ii. Class Weighted Round Robin iii. Channel Rate Limiting

4.1.1. Class Rate Limiting

This scheduler hierarchy level can be used to rate limit individual traffic classes or individual connections (flow) in a traffic class.

Class rate limiting can be configured using the following command:

[root@host]# cxgbtool <ethX> sched-class params type packet level cl-rl mode <scheduler-mode> rate-unit <scheduler-rate-unit> rate-mode

<scheduler-rate-mode> channel <Channel No.> class <scheduler-class-index> max-rate <maximum-rate> pkt-size <Packet size>

Here,

ethX is the Chelsio interface

scheduler-mode specifies whether the rule is configured for individual traffic classes or individual connections (flow) in a traffic class. Possible values include flow or class.

scheduler-rate-unit specifies whether the rule is configured for bit-rate or packet rate .

Possible values include bits or pkts

scheduler-rate-mode specifies whether the rule is configured to support a percent of the channel rate or an effective rate. Possible values include relative or absolute.

Channel No. is the port on which data is flowing (0-3).

scheduler-class-index is the TCP traffic class (0-14)

maximum-rate is the bit rate (Kbps) for this TCP stream.

Packet size is the TCP mss size in bytes; for example – for an MTU of 1500, use a packet size of 1460.

Note flow scheduler-mode and pkts scheduler-rate-unit are not supported in this release

Chelsio T5/T4 Unified Wire For Linux Page 214

XVI. Traffic Management

4.1.2. Class Weighted Round Robin

Incoming traffic flows from various applications can be prioritized and provisioned using a weighted round-robin scheduling algorithm.

Class weighted round robin can be configured using the following command:

[root@host]# cxgbtool <ethX> sched-class params type packet level cl-wrr mode class rate-unit bits rate-mode absolute channel <Channel No.> class

<scheduler-class-index> weight <Y>

Here,

ethX is the Chelsio interface

Channel No. is the port on which data is flowing (0-3).

scheduler-class-index is the TCP traffic class (0-14)

weight is the weight to be used for a weighted-round-robin scheduling hierarchy. Possible values include 1 to 99.

Note

Class Weighted Round Robin level is not supported on T5 adapters.

4.1.3. Channel Rate Limiting

This scheduler hierarchy level can be used to rate limit individual channels.

Channel rate limiting can be configured using the following command:

[root@host]# cxgbtool <ethX> sched-class params type packet level ch-rl mode class rate-unit <scheduler-rate-unit> rate-mode <scheduler-rate-mode> channel <Channel No.> class 0 max-rate <maximum-rate> pkt-size <Packet size>

Here,

ethX is the Chelsio interface

scheduler-rate-unit specifies whether the traffic management rule is configured for bit-rate or packet rate . Possible values include bits or pkts

scheduler-rate-mode specifies whether the traffic management rule is configured to support a percent of the channel rate or an effective rate. Possible values include relative or absolute.

Channel No. is the port on which data is flowing (0-3).

Chelsio T5/T4 Unified Wire For Linux Page 215

XVI. Traffic Management

maximum-rate is the bit rate (Kbps) for this TCP stream. The lower limit is 300 Mbps and

600 Mbps for T4 and T5 adapters respectively.

Packet size is the TCP mss size in bytes; for example – for an MTU of 1500, use a packet size of 1460.

4.2.

Configuring Traffic Management

4.2.1. For Non-offloaded connections

Traffic Management of non-offloaded connections is a 2-step process. In the first step bind connections to indicated NIC TX queue using tc utility from iproute2-3.9.0 package. In the second step bind the indicated NIC TX queue to the specified TC Scheduler class using the

cxgbtool utility.

1. Bring up the interface:

[root@host]# ifconfig ethX up

2. Bind connections to queues:

[root@host]# tc qdisc add dev ethX root handle 1: multiq

[root@host]# tc filter add dev ethX parent 1: protocol ip prio 1 u32 match ip dst <IP address of destination> action skbedit queue_mapping <queue>

Note

For additional binding options, run [root@host]# man tc

3. Now, bind the NIC TX queue with traffic class:

[root@host]# cxgbtool ethX sched-queue <queue> <class>

Here,

ethX is the Chelsio interface

queue is the NIC TX queue

class is the TX scheduler class

Chelsio T5/T4 Unified Wire For Linux Page 216

XVI. Traffic Management

Note

If the TX queue is all, * or any negative value, the binding will apply to all of the TX queues associated with the interface. If the class is unbind, clear or any negative value, the TX queue(s) will be unbound from any current TX Scheduler Class binding.

4.2.2. For Offloaded connections

Traffic Management of offloaded connections can be configured either by applying COP policies that associate offloaded connections to classes or by modifying the application.

Both the methods have been described below:

Applying COP policy

1. Bring up the interface:

[root@host]# ifconfig ethX up

2. Create a new policy file (say new_policy_file) and add the following line to associate connections with the given scheduling class.

E.g.: src host 102.1.1.1 => offload class 0

The above example will associate all connections originating from IP address 102.1.1.1 with scheduling class 0

3. Compile the policy file using COP

[root@host]# cop -d –o <output_policy_file> <new_policy_file>

4. Apply the COP policy:

[root@host]# cxgbtool ethX policy <output_policy_file>

Where,

Chelsio T5/T4 Unified Wire For Linux Page 217

XVI. Traffic Management

ethX: Chelsio interface

Note

For more information on additional parameters, refer cop manual by running the man cop command.

Modifying the application

The application can also be modified in order to associate connections to scheduling classes.

Follow the steps mentioned below:

1. Determine the TCP socket file descriptor in the application through which data is sent.

2. Declare and initialize a variable in the application:

int cl=1;

Here,

cl is the TCP traffic class(scheduler-class-index) that the user wishes to assign the data stream to. This value needs to be in the range of 0 to 7.

The application will function according to the parameters set for that traffic class.

3. Add socket option definitions:

In order to use setsockopt() to set the options to the TCP socket, the following two definitions need to be made:

SOL_SCHEDCLASS used for setting TCP traffic class, which has the value 290.

IPPROTO_TCP used for setting the type of IP Protocol.

# define SOL_SCHEDCLASS 290

# define IPPROTO_TCP 6

4. Use the setsockopt() function to set socket options:

The setsockopt() call must be mentioned after the connect() call.

//Get the TCP socket descriptor variable setsockopt (sockfd , IPPROTO_TCP, SOL_SCHEDCLASS, &cl, sizeof(cl));

Chelsio T5/T4 Unified Wire For Linux Page 218

XVI. Traffic Management

Here:

sockfd : The file descriptor of the TCP socket

&cl : Pointer to the class variables

sizeof(cl) : The size of the variable

5. Now, compile the application.

Chelsio T5/T4 Unified Wire For Linux Page 219

XVI. Traffic Management

5. Usage

5.1.

Non-Offloaded Connections

The following example demonstrates the method to rate limit all TCP connections on class 0 to a rate of 300 Mbps for Non-offload connections:

1. Load the Network driver in NIC mode.

[root@host]# modprobe cxgb4

2. Bind connections with destination IP address 192.168.5.3 to NIC TX queue 3

[root@host]# tc qdisc add dev eth0 root handle 1: multiq

[root@host]# tc filter add dev eth0 parent 1: protocol ip prio 1 u32 match ip dst 192.168.5.3 action skbedit queue_mapping 3

3. Bind the NIC TX queue to class 0

[root@host]# cxgbtool eth0 sched-queue 3 0

4. Set the appropriate rule for class 0

[root@host]# cxgbtool eth0 sched-class params type packet level cl-rl mode class rate-unit bits rate-mode absolute channel 0 class 0 max-rate

300000 pkt-size 1460

5.2.

Offloaded Connections

The following example demonstrates the method to rate limit all TCP connections on class 0 to a rate of 300 Mbps for offloaded connections:

1. Load the Network driver in TOE mode

[root@host]# modprobe t4_tom

Chelsio T5/T4 Unified Wire For Linux Page 220

XVI. Traffic Management

2. Create a new policy file (say new_policy_file) and add the following line to associate connections with the given scheduling class.: src host 102.1.1.1 => offload class 0

3. Compile the policy file using COP

[root@host]# cop -d –o <output_policy_file> <new_policy_file>

4. Apply the COP policy:

[root@host]# cxgbtool eth0 policy <output_policy_file>

5. Set the appropriate rule for class 0

[root@host]# cxgbtool ethX sched-class params type packet level cl-rl mode class rate-unit bits rate-mode absolute channel 0 class 0 max-rate

300000 pkt-size 1460

5.3.

Offloaded Connections with Modified Application

The following example demonstrates the method to rate limit all TCP connections on class 0 to a rate of 300 Mbps for for offloaded connections with modified application.

1. Load the Network driver in TOE mode.

[root@host]# modprobe t4_tom

2. Modify the application as mentioned in the Configuring Traffic Management section.

3. Set the appropriate rule for class 0

[root@host]# cxgbtool ethX sched-class params type packet level cl-rl mode class rate-unit bits rate-mode absolute channel 0 class 0 max-rate

300000 pkt-size 1460

Chelsio T5/T4 Unified Wire For Linux Page 221

XVII. Unified Wire Manager (UM)

XVII. Unified Wire Manager (UM)

Chelsio T5/T4 Unified Wire For Linux Page 222

XVII. Unified Wire Manager (UM)

1. Introduction

Chelsio's Unified Wire Manager is a powerful management software tool, allowing you to view and configure different aspects of the system, including Chelsio hardware installed in the system. The software includes a command line interface (CLI) tool and a web management interface (Web GUI) to help you manage all Chelsio network adapters on the network across multiple operating systems.

Unified Wire Manager enables the management of all aspects of the client side of the iSCSI

SAN in two main areas. The ability to configure Chelsio adapter’s boot option ROM without entering each individual adapter’s configuration screen and manage group of iSCSI initiators remotely from a common user interface saves administrator’s time considerably. Unified Wire

Manager fully supports Linux’s Open-iSCSI initiator and Chelsio’s iSCSI Target. All supported

Chelsio FCoE initiators available on Linux and Windows operating systems can be managed.

Users can manage Option ROM (PXE and iSCSI for T3; PXE, FCoE and iSCSI for T5 and T4) capability for Chelsio cards using various tools available in the software.

Additionally, Unified Wire Manager allows Chelsio adapter NIC and TOE parameters to be centrally managed through the same easy-to-use user interface. It can bring interfaces up or down, team/bond interfaces together, tune parameters for optimal performance, and any number of administrative tasks normally done at each individual machine.

Chelsio’s Unified Wire Manager is an indispensable tool for saving administrator’s time for managing Network and SAN resources . Chelsio’s high performance network adapters with its

Unified Wire approach to networking can now be managed centrally in a simple and fast way!

1.1.

Features

Chelsio’s Unified Wire Manager is designed to provide the following features to the end-user:

Remotely manage Chelsio adapters and various related tasks like driver installation from a single application.

Single tool with CLI and Web interface that works across Storage, Networking and

Hardware.

Manage all Chelsio adapters installed on the system.

Tool for FAE to debug issues on the Customer front.

Freedom to choose various modes of management i.e. CLI or Web GUI.

Chelsio T5/T4 Unified Wire For Linux Page 223

XVII. Unified Wire Manager (UM)

1.2.

Reference Architecture

Chelsio’s Web GUI is a web-based management interface that lets you remotely manage several Chelsio CNAs from anywhere, at anytime on the network using a web browser. The

Web GUI provides a great amount of flexibility, efficiency and accessibility to system administrators in managing Network and SAN resources.The users have the freedom to access the interface using any of the major browsers available, based on individual preferences and corporate policy.

System performance degradation issues will not be observed when using the Web GUI, since it’s lightweight and utilizes very less system resources.

Figure 1.2 – Chelsio Unified Wire Manager with web interface (Web GUI)

1.3.

Unified Wire Manager Components

1.3.1. Management Agent

The Management Agent is a binary executable, which runs as a service on the system that has at least one Chelsio card installed. It is installed along with libraries that can manage various components of the system and enabled during system startup.

1.3.2. Management Client

The Management Client can be used to connect to agents and manage them. Once connected you can view and configure Chelsio HBAs and related networking, storage and hardware

Chelsio T5/T4 Unified Wire For Linux Page 224

XVII. Unified Wire Manager (UM) properties. You can use either the CLI or Web GUI client to manage agents based on your preference. It makes service requests based on the command issued by the user and returns the appropriate information.

CLI Client

The CLI Client (chelsio_uwcli) is an executable binary which allows you to manage and configure agents using the command-line interface. It is not a command shell with a prompt; it accepts all command parameters as arguments when launching it, making it script-friendly.

Web GUI Client

The Web Management Interface (Web GUI) client is a web-based management interface which allows you to securely manage agents from anywhere using a web browser. The management interface uses a secure 256-bit encrypted HTTP connection, ensuring that authentication and configuration data are protected during transmission from the web browser to the system and vice versa. Many agents can be accessed on single interface making it very efficient & user-friendly.

Currently supported browsers are Internet Explorer 9+, Mozilla Firefox 3.6.9+, Google

Chrome 5+ and Apple Safari 5+.

1.4.

Authentication and encryption

The Unified Wire Manager requires user authentication to manage a system. A user must have administrative privileges to manage a system. The authentication credentials, as well as all data exchanged between the CLI client or the Web GUI and the agent, are encrypted using SSL.

This ensures that the data cannot be accessed when it is being transmitted over the network.

Chelsio T5/T4 Unified Wire For Linux Page 225

XVII. Unified Wire Manager (UM)

2. Hardware and Software

2.1.

Supported Adapters

Following are the currently shipping Chelsio Adapters that are compatible with Chelsio Unified

Wire Manager:

T520-LL-CR

T520-CR

T580-LP-CR

T420-CR

T440-CR

T422-CR

T420-SO-CR

T404-BT

T420-BCH

T440-LP-CR

T420-BT

T420-LL-CR

T420-CX

S302E

S302E-C

S310E-CR

S310E-CR-C

S310E-CXAS310E-SR+

S310E-SR

S310E-BT

S320E-CR

S320E-LP-CRS320E-CXA

S320EM-BS

S320EM-BCH

N320E-G2-CR

N320E

N320E-CXA

N320E-BT

N310E

N310E-CXA

Chelsio T5/T4 Unified Wire For Linux Page 226

XVII. Unified Wire Manager (UM)

2.2.

Platform/Component Matrix

The table below lists the Linux distributions and the supported UM components.

Distribution

RHEL5.8, 2.6.18-308.el5 (64-bit)

Supported UM Components

Management Agent, Management Client

RHEL6.3, 2.6.32-279.el6 (64-bit)

Management Agent, Management Client

RHEL 6.4, 2.6.32-358.el6 (64-bit)

Management Agent, Management Client

SLES11SP2, 3.0.13-0.27 (64-bit)

Management Agent, Management Client

Note

This release of Chelsio Unified Wire doesn't support the Web-based management station (Web GUI) component. Please use the CLI client to manage and configure agents.

2.3.

Platform/Driver Matrix

The table below lists the Chelsio drivers and their supported versions:

NIC

TOE

Bypass

Bonding iSCSI Target

Chelsio driver Version

T3: 2.0.0.1(RHEL 5.8,6.3; inbox driver for SLES11SP2)

T4,T5: 2.6.0.2

T3 : 2.0.0.1 (RHEL 5.8,6.3)

T4,T5: 2.6.0.2

T4: 2.6.0.2

T3 : 2.0.0.1

T4: 3.7.1

T3: 5.2.0-0560

T4,T5: 2.6.0.2-675

T3,T4,T5: 2.0-872 Open iSCSI Initiator iWARP

T3,T4,T5: 2.6.0.2

Chelsio T5/T4 Unified Wire For Linux Page 227

XVII. Unified Wire Manager (UM)

FCoE Initiator

Chelsio driver

T4: 2.6.0.2

Version

Chelsio T5/T4 Unified Wire For Linux Page 228

XVII. Unified Wire Manager (UM)

3. Installing Unified Wire Manager

Chelsio Unified Wire has been designed to install Unified Wire Manager (UM) by default. All the three UM components, i.e. Management Agent, Client and Station, will be installed on selecting any of the Terminator 4/Terminator 5 configuration tuning options during installation. Hence, no separate installation is required.

Chelsio T5/T4 Unified Wire For Linux Page 229

XVII. Unified Wire Manager (UM)

4. Verifying UM components status

The following section explains how to verify status of various UM components.

4.1.

Verifying Management Agent

1. Execute the following query command :

[root@chelsio]# ps –eaf | grep UW

The above query should confirm that Management Agent is running by displaying a similar result: root 30531 1 0 09:27 ? 00:00:00 ./UWMgrServer root 30534 1 0 09:27 ? 00:00:00 ./UWMgrServer --run=slp root 30537 1 0 09:27 ? 00:00:00 ./UWMgrServer --run=logserver root 30581 28384 0 09:45 pts/1 00:00:00 grep UW

2. You can also execute the following comand to determine if Management Agent is running:

[root@chelsio]# /etc/init.d/chelsio-uwire_mgmtd status

The above command should display the following result:

Unified Wire Manager Agent : Running

Chelsio T5/T4 Unified Wire For Linux Page 230

XVII. Unified Wire Manager (UM)

4.2.

Verifying Management Client

Execute the following query command to determine if Management Client is installed:

[root@host]# chelsio_uwcli -V

The above query should confirm that Management Client is installed by displaying a similar result:

Unified Manager client CLI version : 2.x.yy

4.3.

Verifying Management Station

Execute the following query command to determine the status of Management Station:

[root@host]# /etc/init.d/chelsio-mgmtstd status

The above command will display one of the following messages:

UM Management Station: Running

UM Management Station: Stopped

Based on the status displayed, you can start, stop or restart Management Station by executing the following command:

[root@host]# /etc/init.d/chelsio-mgmtstd [start|stop|restart]

Note

While accessing the Web Management Interface, if an error “Management

Station is not running” is displayed at the log-in page, follow the above steps to restart Management Station and try again.

Chelsio T5/T4 Unified Wire For Linux Page 231

XVII. Unified Wire Manager (UM)

5. Management Agent

5.1.

Communication

The agent uses a TCP connection over IP to communicate with the client. After the connection is established, SSL (Secure Sockets Layer) encryption is enabled using the Open SSL libraries.

The agent listens on a TCP port for new incoming connections from clients. This port is set to

35001 by default. It may be changed in the configuration file for the agent. The agent needs to be restarted after the change.

5.2.

Configuration

The agent uses a configuration file uwmgr.conf, which is in the agent's installation directory in

/etc/chelsio-uwire. The only configurable parameter available for the agent is the TCP listening port, which can be specified with syntax similar to the example below: PORT 35001. After changing the port, please restart Management Agent for changes to take effect.

5.3.

Service configuration

The agent is installed as a service on the system, and enabled to start on boot. The following sections will describe the procedure to configure service startup manually:

5.3.1. Service startup configuration

The service name is chelsio-uwire_mgmtd and can be configured using the chkconfig utility. The service startup configuration can be viewed and modified as below:

Execute the following command to list the service configuration.

[root@host]# chkconfig --list chelsio-uwire_mgmtd

Execute the following command to enable/disable the service to start at system runlevel 5.

[root@host]# chkconfig --level 5 chelsio-uwire_mgmtd on/off

Chelsio T5/T4 Unified Wire For Linux Page 232

XVII. Unified Wire Manager (UM)

5.3.2. Service start/stop/restart

You can start, stop or restart the service by using the following command:

[root@host]#/etc/init.d/chelsio-uwire_mgmtd [start|stop|restart]

5.4.

Firewall

If the system has a firewall configured, such as iptables, it should be configured to allow traffic to the management agent TCP port configured above in the configuration section, or the default port that the management agent uses, 35001. Review the firewall documentation and configure it appropriately. If there is a firewall appliance / software protecting the network that the system is on, and you wish to connect to the system from a different network, using the client, the firewall appliance also needs to be configured appropriately.

Chelsio T5/T4 Unified Wire For Linux Page 233

XVII. Unified Wire Manager (UM)

6. CLI client

6.1.

CLI Help system

A detailed help and usage documentation is built into the CLI, and is accessible through its help system. The help can be invoked by the usual argument of /? or --help.

6.1.1. Viewing help

Use the chelsio_uwcli command to view the help file as shown below:

[root@host]# chelsio_uwcli /?

6.2.

Client conflict resolution

The CLI and Web GUI cannot manage the same system at the same time by default. This is to ensure that configuration changes being applied by one client are not interrupted by another client. Also, two different Web GUI or CLI clients cannot connect to a management agent at the same time. There is no mechanism to allow this scenario.

Chelsio T5/T4 Unified Wire For Linux Page 234

XVII. Unified Wire Manager (UM)

7. Web GUI client

7.1.

Management Station

In order to access the Web Management Interface, Apache HTTP server should be installed and running on a machine. Also, Cookies and Javascript must be enabled in the browser.

7.1.1. Running Management Station on RHEL 6.x

1. Start/Restart Apache httpd daemon:

[root@host]# service httpd [start|restart]

2. Start/Restart the Management Station:

[root@host]# /etc/init.d/chelsio-mgmtstd [start|restart]

7.1.2. Running Management Station on SLES11SP2

1. On SLES11SP2, Management Station needs to be configured before running. Hence, execute the following in command prompt and provide valid inputs.

[root@host]# cd /etc/apache2

[root@host]# openssl genrsa -des3 -out server.key 1024

[root@host]# openssl req -new -key server.key -out server.csr

[root@host]# cp server.key server.key.org

[root@host]# openssl rsa -in server.key.org -out server.key

[root@host]# openssl x509 -req -days 365 -in server.csr -signkey server.key

-out server.crt

[root@host]# cp server.crt ./ssl.crt

[root@host]# cp server.key ./ssl.key

2. Start/Restart Apache services

[root@host]# rcapache2 [start|restart]

Chelsio T5/T4 Unified Wire For Linux Page 235

XVII. Unified Wire Manager (UM)

3. Start/Restart the Management Station:

[root@host]# /etc/init.d/chelsio-mgmtstd [start/restart]

7.2.

Accessing Web Management Interface

1. To access the Web GUI, type in the URL https://<management station IP address> in a web browser.

2. The security certificate used by the web server is a generic one. It may cause the following types of prompts in different browsers. You will need to select the correct option to continue.

Figure 7.2 (a) -

Security Certificate prompt in Internet Explorer

Chelsio T5/T4 Unified Wire For Linux Page 236

XVII. Unified Wire Manager (UM)

Figure 7.2 (b) -

Security Certificate prompt in Mozilla Firefox

Figure 7.2 (c) -

Security Certificate prompt in Apple Safari

Figure 7.2 (d) -

Security Certificate prompt in Google Chrome

3. The web interface requires password authorization to be accessed. Enter the username and corresponding password that was set up on the management station system and click on the Login button.

Chelsio T5/T4 Unified Wire For Linux Page 237

XVII. Unified Wire Manager (UM)

Figure 7.2 (e) - Web GUI Login page

Note

Not performing any operation/action for 5 minutes will result in session timeout.

You will have to re-login and connect to the Agents again.

7.3.

Layout and Navigation

The Web Management Interface consists of the following:

Title bar displaying the username on the left, Unified Wire Manager logo and name in the centre; and a Logout button on the right.

Menu Bar consisting of the Home, Add System, Remove System, Refresh, Subscribe and Bulk Configuration buttons.

The Navigation Pane with a cascading tree of links to various configuration modules for a

UM Agent. You can navigate between connected agents and various sections of the managed agent’s interface. You can view and hide the configuration modules for each

Agent by clicking on the “+”and “-”links respectively

The Details Pane on the right displaying panels associated with the tree menu item selected in the Navigation Pane. The panels can be expanded and collapsed by clicking on the panel heading.

The Bottom bar has the About link on the right and copyright details on the left.

Chelsio T5/T4 Unified Wire For Linux Page 238

XVII. Unified Wire Manager (UM)

Figure 7.3

– Web Management Interface

7.4.

Home page

The home page is displayed by default on launching the Web GUI. It displays Bookmarks and

History, Service Discovery and Bulk Driver Installation modules. Options to go back to home page, add/remove system, refresh and configure email alerts are also available.

7.4.1. Home

This option will display the home page.

Bookmarks and History

A history of the last 128 systems that were managed from this system, by the current user, will be shown here in a list. Each system's management IP address, TCP port, and Login details are also stored. This may be edited and saved. Any systems that are not required in the list may be deleted.

Important

Storing login passwords for the managed systems is inherently

insecure. The passwords are encrypted, but it is still advisable to store passwords only if the system you are running the GUI client on, is secure.

Chelsio T5/T4 Unified Wire For Linux Page 239

XVII. Unified Wire Manager (UM)

Figure 7.4.1 (a) - Bookmarks and history module

Connecting to a system

Select the system from the Bookmark list and click Connect. Once successfully connected, the system will appear on the left pane with different related modules on the right to view and manage.

Removing a system

Select the system from the Bookmark list and click Delete system to remove it.

Note

Once removed, the system will no longer appear in the Bookmarks and History module. If you wish to manage that system again, you will have to use the Add

system option.

Chelsio T5/T4 Unified Wire For Linux Page 240

XVII. Unified Wire Manager (UM)

Service Discovery

Using this module, all the Unified Wire Manager agents connected in the same or different subnet can be discovered. One can choose to discover agents based on OS type (Linux,

Windows, FreeBSD or All) or search for a particular agent if the agent's IP is known by using the

Choose the HostIP option. The Add Agents button adds the selected system to the list of discovered agents in the Bookmarks and History module. The Clear Agents button resets the list of discovered agents.

Figure 7.4.1 (b) - Service Discovery module

Bulk Driver Installation

This module allows you to install drivers for multiple systems simultaneously. Drivers available for installation for a particular system may differ depending on the network card (T3 or T4) selected and the operating system (Linux, Windows, ESX, Macintosh, Solaris, FreeBSD) selected.

Installing Driver

1. In the Choose the card fields, select T3 or T4 depending on the chip revision of the network card.

2. Select the operating system for which drivers are to be installed in the Choose the OS Type field. All the systems with the selected operating system will be displayed in the list below.

Chelsio T5/T4 Unified Wire For Linux Page 241

XVII. Unified Wire Manager (UM)

3. Select a system or systems from the list and choose the driver to be installed in the Driver

Installation section.

4. Download the appropriate driver from Chelsio’s Download Center, service.chelsio.com.

5. Locate the driver package.

6. Click Install button to install the driver.

Figure 7.4.1 (c) – Bulk Driver Installation module for Linux Agent

Note

Agents that report errors or whose passwords are not stored, will be

automatically skipped during the driver installation.

Chelsio T5/T4 Unified Wire For Linux Page 242

XVII. Unified Wire Manager (UM)

7.4.2. Add System

Use this option to connect to new Agents using their IP or Hostname. You can enter the TCP port for connection or leave it at its default value (35001). You will have to provide correct user credentials for the agent in order to connect successfully.

After connecting to the Agent, the menu bar on the left will display the connected system and its related modules.

If you deselect the ‘Remember Password’ option, you will be asked to enter password every time you try to connect to the system.

Figure 7.4.2 - Adding a UM Agent

7.4.3. Remove System

Use this option to disconnect an Agent. To remove an agent, click on the name of the system in the tree menu in the left and click Remove System. Then click Yes to confirm.

Figure 7.4.3 - Removing a UM Agent

Chelsio T5/T4 Unified Wire For Linux Page 243

XVII. Unified Wire Manager (UM)

7.4.4. Refresh

This option can be used to reload the Web GUI or UM Agent.

To reload the Web GUI, navigate to the Home page (by clicking on the “Home” button and click

Refresh. You can use this option to refresh Home page panels (Bookmarks and History,

Service Discovery and Bulk Driver Installation).

To reload an Agent, click on the name of the system in the tree menu in the left and click

“Refresh”. You can use this option to update any changes made to system settings like load/unload drivers.

7.4.5. Subscribe (Email Alerts)

This feature is available only on the Web Management Interface.

Using this option, you can receive email alerts regarding the link status of a Chelsio Network

Interface Card. This feature sends email notifications regarding the port and the card, on which the link up/down event is occurring. Not only can you configure multiple email addresses to receive notifications, but also customize the email id of the sender for troubleshooting purposes.

To subscribe to Email Alerts, enter the sender’s email address in the Email address field. It can be anything in the format of <name>@<domain>.<extension>. You can enter multiple email addresses for the Recipients field separated a comma. Enter Mail server details and ensure that the “Enable email Alerts” field is enabled. Select the Agent(s), for which you want to receive alerts and Click on Save.

Chelsio T5/T4 Unified Wire For Linux Page 244

XVII. Unified Wire Manager (UM)

Figure 7.4.5 - Subscribing to Email Alerts

7.4.6. Bulk Configuration

The Bulk Configuration page allows you to execute common configuration changes to multiple agents and their network adapters simultaneously. You can conveniently perform bulk operations like installing option ROM, setting MTU and VLAN ID, changing adapter and port parameters on various devices, without having to access multiple modules and thus saving considerable amount of administration time.

Various configurable parameters have been categorized into several modules like Boot

Configuration module to install and erase option ROM, Network Configuration module to set

MTU and VLAN ID, Card Configuration module to change driver parameters, etc.

Chelsio T5/T4 Unified Wire For Linux Page 245

XVII. Unified Wire Manager (UM)

Before accessing these modules, you will have to create groups and then add members to that group. Once done, you can select the group in the modules and the new setting will be applied to all members of that particular group.

Manage Groups

This is where you can add, delete and manage groups. Use the Create a Group section to create a group by specifying a gent’s platform and group type. There are various types of groups to choose from depending on the type of configuration setting you want to change. For example, to change the MTU size of a network interface (in the Network Configuration module), create a group with group type Network. To install or erase option ROM on a Chelsio T4 adapter (in the

Boot Configuration module), create a group with group type t4adapter.

Here is a list of available configuration modules and corresponding group type:

Boot Configuration: t3adapter,t4adapter

Network Configuration: Network

Card Configuration: t3adapter,t4adapter

Port Configuration: t3port, t4port

Bypass Configuration: Network

After the group has been created, add members to that group using the Add a member row button. Depending on the group type selected, you may be asked to provide additional details for the new member. Use the Fetch Details button to do so and finally click Save a member button to add the member to the group.

To delete a group, select it from the Select a Group drop-down list, and click Delete Group. To delete a member from a group, select the group to which the member belongs, select the radio button corresponding to the member to be deleted in the SNO field and finally click Delete

Member(s).

Figure 7.4.6 (a) – Creating a group

Chelsio T5/T4 Unified Wire For Linux Page 246

XVII. Unified Wire Manager (UM)

Figure 7.4.6 (b) - Managing a group

Boot Configuration

Using this module, you can install option ROM or erase option ROM on Chelsio network devices. The Set Default Boot Settings button will reset the adapter to factory boot settings.

Figure 7.4.6 (c) – Boot Configuration module

Chelsio T5/T4 Unified Wire For Linux Page 247

XVII. Unified Wire Manager (UM)

Network Configuration

In the Network Configuration module, you can set Maximum Transfer Unit (MTU),Virtual LAN

(VLAN) ID and change the IP address type for the members (network interfaces) of the Network group. MTU can be set between 1500-9000 bytes. VLAN id can be set for an adapter within the range 0-4094 (enter 0 to disable it).

Figure 7.4.6 (d) – Network Configuration module

Chelsio T5/T4 Unified Wire For Linux Page 248

XVII. Unified Wire Manager (UM)

Card Configuration

The Card Configuration module allows you to set various adapter settings including TCP

Offload. Offload settings are only available when using the TOE capable drivers (t4_tom and

toecore for T4 adapters; t3_tom and toecore for T3 adapters).

Figure 7.4.6 (e) – Card Configuration module

Chelsio T5/T4 Unified Wire For Linux Page 249

XVII. Unified Wire Manager (UM)

Port Configuration

In the Port Configuration module, you can set various port settings like enabling Tx checksum and TCP segmentation offload, setting Link speed and link duplex mode, etc. The settings depend on the device driver installed.

Figure 7.4.6 (f) - Port Configuration module

Chelsio T5/T4 Unified Wire For Linux Page 250

XVII. Unified Wire Manager (UM)

Bypass Configuration

Use the Bypass Configuration module to configure Chelsio’s bypass adapters like B420-SR and B404-BT. For more information on different bypass modes and configurational parameters,

see Bypass Driver chapter.

Figure 7.4.6 (g) - Bypass Configuration module

Chelsio T5/T4 Unified Wire For Linux Page 251

XVII. Unified Wire Manager (UM)

7.5.

System page

The system page is displayed, when the system hostname / IP address is selected in the tree menu on the left. On adding a system, this item is automatically selected, and this page is displayed. The system page contains generic system and support modules which are discussed below:

7.5.1. System Summary

This module lists the system Hostname, Operating System, platform and management connection details, and also gives the count of the Chelsio cards found.

Figure 7.5.1 - System Summary module

Chelsio T5/T4 Unified Wire For Linux Page 252

XVII. Unified Wire Manager (UM)

7.5.2. Drivers Installation

Using this module, one can install various Chelsio drivers for different operating systems. You can choose the configuration file type (Linux Agents only).

Figure 7.5.2 (a) - Drivers Installation module connected to Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 253

XVII. Unified Wire Manager (UM)

Figure 7.5.2 (b) - Drivers Installation module connected to FreeBSD Agent

7.5.3. Unified Manager Upgrade

Using this module you can update Management Agent. Use the “Select” button to locate the

Chelsio Unified Wire Package and the click “Install”. After the Agent is successfully upgraded, the system will be disconnected. Re-connect to the system again and verify the Agent version in the Unified Wire Manager Component Versions. The Server field will display the installed

Agent version.

Figure 7.5.3 - Unified Manager Upgrade module

Chelsio T5/T4 Unified Wire For Linux Page 254

XVII. Unified Wire Manager (UM)

7.5.4. Driver Details

A list of the Chelsio device drivers installed on the system is shown in this module, with version information and the current status, whether the driver is loaded and active or not.

Figure 7.5.4 - Driver Details module for a Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 255

XVII. Unified Wire Manager (UM)

7.5.5. System Diagnostics

Using this module, you can run various diagnostic tests on Chelsio HBAs to troubleshoot adapter related issues. Select the card(s) from the list for which you want to run the test, select the operation (type of test; you can run more than one test at a time) and click “Run Test”. After the tests are completed, the results will be displayed in a tabular format.

Figure 7.5.5 - System Diagnostics module for a T4 CNA

Chelsio T5/T4 Unified Wire For Linux Page 256

XVII. Unified Wire Manager (UM)

7.5.6. Unified Wire Manager Component Versions

A list of the Unified Wire Manager agent components installed on the managed system is shown in this module. The versions of the components are useful in case of reporting an issue to support.

Figure 7.5.6 - Unified Wire Manager Component Versions module

7.5.7. KVM Configuration (Linux)

This module allows you to enable or disable KVM related operations. Once enabled, two related modules, VM Configurations and VF Configurations, will be available and you can manage

VMs and related settings. While enabling, you will have to provide the number of Virtual

Functions (VFs) you want to create for the physical ports.

The KVM Configuation module will be available only when Chelsio Network driver (cxgb4) and

KVM modules (kvm and kvm_intel/kvm_amd) are loaded. If not done already, access Agent CLI and follow these steps to do so: i. If any or all the aforementioned drivers are already loaded, unload them before proceeding:

[root@host]# rmmod <kvm_intel/kvm_amd>

[root@host]# rmmod kvm

[root@host]# rmmod cxgb4

Chelsio T5/T4 Unified Wire For Linux Page 257

XVII. Unified Wire Manager (UM) ii. Next, reload them using modprobe:

[root@host]# modprobe kvm allow_unsafe_assigned_interrupts=1

[root@host]# modprobe <kvm_intel/kvm_amd>

[root@host]# modprobe cxgb4

Loading the kvm module with allow_unsafe_assigned_interrupts=1 option enables use of device assignment without interrupt remapping support. This is required in order to asssign VFs to

VMs. iii. Finally, access WebGUI. Remove the Agent using the “Remove System” button and connect to it again from the Bookmarks and History module. You should now be able to see the KVM Configuration module.

Figure 7.5.7 (a)

– KVM Configuration module

Figure 7.5.7 (b) – KVM Configuration module: Adding Virtual Functions

7.5.8. VM Configurations (Linux)

The VM Configurations module allows you to view UUID and domain state of Virtual Machines.

You can perform various system power options like start or resume (if VM is paused), turn off, restart or suspend (pause) a VM.

Chelsio T5/T4 Unified Wire For Linux Page 258

XVII. Unified Wire Manager (UM)

You can perform similar actions on multiple virtual machines. To do so, click on the machine names in the list. The properties box will display the domain state of the machines selected.

Now, click on any of the system power actions provided at the bottom.

Figure 7.5.8 (a) – VM Configurations module

Figure 7.5.8 (b) – VM Configurations module: viewing properties for multiple VMs

7.5.9. VF Configurations (Linux)

The VF Configurations module lists all the VMs, Virtual Functions mapped to each Virtual

Machine and all the available VFs. You can also add and remove VFs for a particular VM.

Chelsio T5/T4 Unified Wire For Linux Page 259

XVII. Unified Wire Manager (UM)

Figure 7.5.9 - VF Configurations module

7.5.10. Xen Configurations

The Xen Configurations module allows you to view UUID, power state of Virtual Machines and

Virtual Functions assigned to them. You can perform various system power options like start, resume (if VM is paused), turn off, restart or suspend (pause) a VM.

You can perform similar actions on multiple virtual machines. To do so, click on the machine names in the list. The properties box will display the power state of the machines selected.

Now, click on any of the system power actions provided at the bottom.

Figure 7.5.10 (a) – Xen Configurations module

Chelsio T5/T4 Unified Wire For Linux Page 260

XVII. Unified Wire Manager (UM)

Figure 7.5.10 (b)

– Xen Configurations module: viewing properties for multiple VMs

7.5.11. Xen VF Properties

Here you can view the list of virtual machines and list of available VFs. To assign a VF to a VM, select the guest name on the left and select the VF to be assigned on the right. You can assign more than one VF at a time. F inally, click “Assign VF” to add the selected VFs to the hosts.

To enable SR-IOV support, IOMMU must be enabled . To do this, select “Enable” for IOMMU and then click “Set IOMMU”. Reboot the host machine for changes to take effect.

Figure 7.5.11 (a) – Xen VF Properties module

Chelsio T5/T4 Unified Wire For Linux Page 261

XVII. Unified Wire Manager (UM)

Important

Please ensure that Chelsio Network driver (cxgb4) with Virtual Functions is enabled in the Driver Details module before attempting to assign VF. Next, you will have to provide the number of Virtual Functions (VFs) you want to create for the physical ports. If VFs are not listed, unload the driver, reload it and create VFs again.

Figure 7.5.11 (b) – Xen VF Properties module: Adding Virtual Functions

7.5.12. Managed system application logs

The management agent logs its activities and any errors that occur, in /var/log/chelsio in Linux and FreeBSD and in the Event log, in Windows. This log can be obtained in this module. Only

20 entries can be obtained and viewed at a time. Logs can be viewed by either choosing from a list of fixed range or by specifying a custom starting point.

Chelsio T5/T4 Unified Wire For Linux Page 262

XVII. Unified Wire Manager (UM)

Figure 7.5.12 (a) - Managed System Application Logs module for Windows Agent

7.6.

Network page

7.6.1. Network summary

The Network Summary module provides the total number of Chelsio adapters present, including the number of T5, T4 and T3 adapters. It also provides the total number of Network interfaces including corporate and Chelsio interfaces and VLANs.

Figure 7.6.1 (a) – Network Summary module

Chelsio T5/T4 Unified Wire For Linux Page 263

XVII. Unified Wire Manager (UM)

7.6.2. Chelsio card page

When a Chelsio card is selected in the tree menu on the left, this page is displayed. It provides details of the card and associated settings. It also displays any card specific statistics that the hardware provides. The modules available on this page are as below:

Card summary

This module provides PCI, firmware and other details of the card. The card's serial number and factory MAC address is also provided for inventory purposes.

Figure 7.6.2 (a) - Card Summary module for a Windows Agent

Chelsio T5/T4 Unified Wire For Linux Page 264

XVII. Unified Wire Manager (UM)

TCP Offload settings (Linux & FreeBSD)

The TCP offload settings applicable to the card are shown here. These settings are only available when using the TOE capable drivers (t3_tom and toecore for T3 adapters; t4_tom and

toecore for T4 and T5 adapters). On changing the settings, the changed settings may not reflect immediately on refreshing the data. Highlight the system item in the tree menu on the left, and click Refresh, to refresh data from the system, in case the updated settings are not being shown.

Figure 7.6.2 (b) – TCP Offload Settings module for a Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 265

XVII. Unified Wire Manager (UM)

Figure 7.6.2 (c) - TCP Offload Settings module for a FreeBSD Agent

Device Driver settings (Windows)

The device driver settings applicable to the card are shown here. For Chelsio T5/T4 adapters, only the MaxVMQueues field will be displayed. On changing the settings, the changed settings may not reflect immediately on refreshing the data. Highlight the system item in the tree menu on the left, and click Refresh, to refresh data from the system, in case the updated settings are not being shown.

Figure 7.6.2 (d) – Device Driver Settings module for a Windows Agent

Chelsio T5/T4 Unified Wire For Linux Page 266

XVII. Unified Wire Manager (UM)

Card statistics

Certain statistics are maintained on a per card basis (instead of a per port basis), since the card has a TCP/IP offload capability. The statistics are for TCP and IP protocol processing done in the card's hardware. These statistics may only be applicable if the card is TOE enabled.

Figure 7.6.2 (e) - Card Statistics module for a T3 HBA

Chelsio T5/T4 Unified Wire For Linux Page 267

XVII. Unified Wire Manager (UM)

7.6.2.1. Chelsio card's port

The port page is displayed on selecting a port of a Chelsio card listed in the tree menu on the left. It provides details of the port and port settings. It also displays any port specific statistics that the hardware provides. The modules available on this page are as below:

Port summary

The port details such as the Ethernet adapter name, connector type and link details are shown in this module.

Figure 7.6.2.1 (a) - Port Summary of T5/T4 CNA on Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 268

XVII. Unified Wire Manager (UM)

Port settings

Port settings such as MTU, Link speed and others can be set in this module. The settings depend on the device driver installed.

Figure 7.6.2.1 (b) - Port Settings of T4/T5 CNA on Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 269

XVII. Unified Wire Manager (UM)

Port statistics

Ethernet statistics and additional hardware statistics for the port are displayed in this module.

Figure 7.6.2.1 (c) - Port Statistics of T4/T5 CNA on Linux Agent

7.6.3. Networking Management page

The system networking and teaming / bonding configurations are shown on this page. IP addresses, MTU, VLAN Ids, DNS and default gateway settings can be viewed and modified here. Network adapters can also be enabled or disabled as required. The modules available on this page are as below:

System Network configuration

The list of network adapters on the system is displayed in a list on the left. The icon for the adapter indicates whether it is administratively enabled and if it is connected to the network.

Network teams are also indicated with an appropriate icon. The primary IP address (IPv4) can be set for the adapter, when it is selected. There is an option to add/modify/delete additional IP

Chelsio T5/T4 Unified Wire For Linux Page 270

XVII. Unified Wire Manager (UM) addresses or aliases for the specified adapter. Use the option to add additional IP addresses with caution, since multiple IP addresses configured on the same adapter, for the same network, may result in unpredictable behavior of the system's networking stack. Maximum

Transfer Unit (MTU) can be set between 1500-9000 bytes. VLAN id can also be set for an adapter within the range 0-4094 (enter 0 to disable it).

You can use the View/Set IP addresses option to add, modify or delete IP aliases.

Figure 7.6.3 (a) - System network configuration module

Figure 7.6.3 (b) – Managing IP aliases

Chelsio T5/T4 Unified Wire For Linux Page 271

XVII. Unified Wire Manager (UM)

System network statistics

Using this module, one can generate reports based on Throughput pkts/sec and Throughput

Mbs (Receive, Transmit, Bi-direction) in Table and Graph format for a network adapter. A report for hardware statistics can be generated based on different parameters, only in the Table view in the Advanced NIC characteristics. The polling time field sets the average time (in seconds) based on which the table/graph updates the report.

Figure 7.6.3 (c) - System network statistics module

Chelsio T5/T4 Unified Wire For Linux Page 272

XVII. Unified Wire Manager (UM)

Figure 7.6.3 (d) - Network Throughput Vs Time instant Graph

Figure 7.6.3 (e) - Network Throughput Vs Time instant Table

Chelsio T5/T4 Unified Wire For Linux Page 273

XVII. Unified Wire Manager (UM)

Default Gateway and DNS configuration

The DNS servers list can be set here. The default gateway for remote networks and the Internet can also be set here. On Linux and FreeBSD, only one default gateway is allowed. On

Windows, you may set multiple default gateways. Use the option to set multiple default gateways with caution, since it may cause the system to stop communicating with external networks.

Figure 7.6.3 (f) - Default gateway and DNS configuration module for a FreeBSD Agent

Chelsio T5/T4 Unified Wire For Linux Page 274

XVII. Unified Wire Manager (UM)

Create a network team/bond device (Linux and FreeBSD)

A list of regular network adapters is provided here, to create a Network Team / Bond device.

The available modes for the team depend on the OS teaming / bonding driver in use. On Linux the team may be created with a DHCP or Static IP address. On Windows, only DHCP is allowed when creating the team, although both DHCP and Static IP addressing is supported for the team adapter, after it is created successfully. Please check with the driver documentation for the supported modes for creating a team / bond, with offload enabled Chelsio cards. All modes may not be available with all configurations / combinations. Also, the team members can only be 2 ports of a single offload-enabled card, and not across Chelsio cards. Do not mix third party cards and offload-enabled Chelsio cards in a single team.

Figure 7.6.3 (g) - Create a network team/bond device module

Chelsio T5/T4 Unified Wire For Linux Page 275

XVII. Unified Wire Manager (UM)

Network troubleshooting

This module allows detecting and troubleshooting various network connectivity issues. The Ping utility helps to contact a system by specifying IP address, Number of ICMP packets to send and packet timeout. The result of the ping can be viewed by clicking on the Ping Result button.

Using TraceRoute one can determine the route taken by packets across an IP network.

Figure 7.6.3 (h) - Ping Utility

Figure 7.6.3 (i) - TraceRoute Utility

Chelsio T5/T4 Unified Wire For Linux Page 276

XVII. Unified Wire Manager (UM)

Figure 7.6.3 (j) - GetConnections Utility

7.6.3.1. Hypervisor

Xen Bridge Configuration

The Xen Bridge Configuration module allows you to view and manage network bridges, virtual nterfaces (vifs) and virtual machines to which those virtual interfaces are assigned. The left pane displays a list of different bridges created. Clicking on a bridge name will display related properties on the right.

If a virtual interface is attached to a particular bridge, a “+” link appears next to the bridge name.

Expanding the “+” link will display the virtual interface and expanding the “+” for that virtual interface displays the virtual machines to which it is assigned. Click on the virtual machine names to view their properties on the right.

To delete a bridge, click on the bridge name and then click “Delete Bridge”. This will also delete the virtual interface attached to that bridge and the VM(s) to which the interface was assigned.

Chelsio T5/T4 Unified Wire For Linux Page 277

XVII. Unified Wire Manager (UM)

Figure 7.6.3.1 (a) – Xen Bridge Configuration module

Bridge Configuration (Linux)

The Bridge Configuration module allows you to view and manage network bridges, virtual network interface (vnets) and virtual machines to which those virtual network interfaces are assigned. The left pane displays a list of different bridges created. Clicking on a bridge name will display related properties on the right.

If a virtual network interface is attached to a particular bridge, a “+” link appears for that bridge name. Expanding the “+” link will display the virtual network interface and expanding the “+” link for that virtual network interface displays the virtual machine to which it is assigned. Click on the virtual machine names to view their properties on the right. Only bridge properties are editable.

To detach a virtual network interface from a bridge, select it and click “Delete VIF/VNET“ Restart the guest machine for changes to take effect.

To delete a bridge, click on the bridge name and then click “Delete Bridge”. If there are virtual network interfaces attached, you will have to detach them first.

Chelsio T5/T4 Unified Wire For Linux Page 278

XVII. Unified Wire Manager (UM)

Figure 7.6.3.1 (b) – Bridge Configuration module (Linux)

Virtual Network Configuration (Linux)

Using the Virtual Network Configuration module, you can create network bridges and attach them to virtual machines. You can also assign physical interfaces on the host to bridges.

To create a bridge, enter a name and click “Create” in the “Create Bridge” section. All other parameters are optional. If not specified, the bridge will be created with default values. Except

STP and Priority, you can change all of the other parameters in the Bridge Configuration

module once the bridge is created.

Use the “Add Bridge to VM” section to attach a bridge to a virtual machine.

To assign interfaces on the host to a bridge, specify the bridge name and the host interface in the “Add Interface to Bridge” section. Click “Add”. You can add multiple interfaces to the same bridge using the aforementioned method. The Cost and Priority parameters are optional.

Chelsio T5/T4 Unified Wire For Linux Page 279

XVII. Unified Wire Manager (UM)

Figure 7.6.3.1 (c) – Creating Bridge

Figure 7.6.3.1 (d) – Adding Bridge to VM

Figure 7.6.3.1 (e) – Adding interfafe to Bridge

Chelsio T5/T4 Unified Wire For Linux Page 280

XVII. Unified Wire Manager (UM)

Virtual Network Configuration (Xen)

Using the Virtual Network Configuration module, you can create network bridges. You can also create and attach virtual interfaces to them.

To create a bridge, enter a label for the bridge and click “Create”. The MTU and Name

Description fields are optional. The bridge name will be generated automatically by the operating system. Once created, it will appear in the Xen Bridge Configuration module. You will have to use the label provided earlier to identify the bridge.

You can attach upto 64 virtual interfaces to a particular bridge. Device Number, Bridge Name and VM Name are the mandatory fields while creating a virtual interface.

Figure 7.6.3.1 (f) – Creating Bridge

Figure 7.6.3.1 (g)

– Creating and attaching virtual interface

Chelsio T5/T4 Unified Wire For Linux Page 281

XVII. Unified Wire Manager (UM)

Virtual Network Manager (Windows)

This module allows you to view and manage virtual networks. The left pane displays a list of different virtual networks created. Clicking on a virtual network name will display related properties on the right.

If a virtual network is added to a virtual machine , a “+” link appears next to the virtual network name. Expanding the “+” link will display the virtual machines to which the network is attached.

Click on the virtual machine names to view their properties on the right.

To delete a virtual network, click on the network name and then click “Delete Switch”. If it is attached to a virtual machine, you will have to detach the virtual machine first. To do so, click on the virtual machine and click “Detach”. Similarly, detach all the virtual machines and then use the “Delete Switch” to delete the virtual network.

Figure 7.6.3.1 (h) – Virtual Network Manager module

Add Virtual Network Configuration (Windows)

There are three kinds of virtual networks you can create using this module:

External network: Using this type, you can provide virtual machines access to external networks and vice versa via a physical network adapter in the host system. The virtual machines can also communicate with each other on the same virtual network.

Internal Network: This type allows communication between virtual machines in the same virtual network and also between the virtual machines and the host. This type of virtual network does is not bound to any physical network adapter and no access to external networks is provided.

Chelsio T5/T4 Unified Wire For Linux Page 282

XVII. Unified Wire Manager (UM)

Private Network: A Private Network is similar to Internal Network in that physical adapter is not required for setup and access to external networks is not provided. However, unlike Internal Network, guest operating systems can only communicate with guest operating systems in the same private network and not with the host. The host operating system cannot access the virtual machines on private network.

Once created, you can manage the virtual networks in the Virtual Network Manager module.

Figure 7.6.3.1 (i)

– Creating external virtual network

Figure 7.6.3.1 (j) – Creating internal virtual network

Figure 7.6.3.1 (k) – Creating private virtual network

Chelsio T5/T4 Unified Wire For Linux Page 283

XVII. Unified Wire Manager (UM)

Virtual Network Settings (Windows)

To attach a virtual network to a virtual machine, select the virtual network from the Virtual

Network list and the virtual machine from the VM list. Finally click Attach.

Figure 7.6.3.1 (l) – Attaching virtual network to VM (Windows)

7.6.4. iWARP iWARP Settings

On Linux Agents, iWARP parameter settings for Chelsio's RDMA capable NICs can be set using this module. These settings can be set only when iWARP driver (iw_cxgb4 for T4 and T5;

iw_cxgb3 for T3) is loaded. If you set any parameter for a T5 adapter, it applies for all the T5 adapters present. Same applies for T4 and T3 adapters.

On Windows Agents, only T3 HBAs are supported currently. Parameters can be set per port.

On FreeBSD Agents, only T4 CNAs are supported. iWARP parameter settings can be set only when iw_cxgbe driver is loaded.

Chelsio T5/T4 Unified Wire For Linux Page 284

XVII. Unified Wire Manager (UM)

Figure 7.6.4 (a) – iWARP settings for T4 CNA for FreeBSD Agent

Chelsio T5/T4 Unified Wire For Linux Page 285

XVII. Unified Wire Manager (UM)

Figure 7.6.4 (b) – iWARP settings for T4 CNA for Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 286

XVII. Unified Wire Manager (UM)

7.6.5. Wire Direct

WD-UDP Process Statistics & Attributes

The WD-UDP module lists the process ids (pid) of UDP traffic running on the agent and displays the corresponding statistics and attributes.

Note

Please ensure that WD-UDP traffic is running on the agent before accessing this module.

Figure 7.6.5 (a) – WD-UDP Process Statistics

Chelsio T5/T4 Unified Wire For Linux Page 287

XVII. Unified Wire Manager (UM)

Figure 7.6.5 (b) – WD-UDP Process Attributes

Chelsio T5/T4 Unified Wire For Linux Page 288

XVII. Unified Wire Manager (UM)

WD-TOE Process Statistics & Attributes

The WD-TOE module lists the process ids (pid) of TOE traffic running on the agent and displays the corresponding statistics and attributes.

Note

Please ensure that WD-TOE traffic is running on the agent before accessing this module.

Figure 7.6.5 (c) – WD-TOE Process Statistics

Figure 7.6.5 (d) – WD-TOE Process Attributes

Chelsio T5/T4 Unified Wire For Linux Page 289

XVII. Unified Wire Manager (UM)

7.7.

Storage Page

Storage Summary

The Storage module lists the status of configuration modules under Storage section, running on the agent.

Figure 7.7 – Storage Summary Module

7.7.1. FCoE Initiator (Linux, Windows, XenServer)

All supported Chelsio FCoE initiators available on the operating system can be managed from this page. FCoE support is extended on Linux, Windows and XenServer platforms. Please refer to Platform/Driver matrix section on the list of operating systems that are supported.

FCoE Initiator Summary

This module provides details about the driver installed; such as driver name and its version. The module also gives information about the number of FCoE enabled cards that are present on the machine.

Figure 7.7.1 (a)

– FCoE Initiator Summary module for Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 290

XVII. Unified Wire Manager (UM)

7.7.1.1. FCoE Initiator Card

FCoE Card Summary

Details pertaining to the card used such as model, firmware/hardware version etc, are provided in this module.

Figure 7.7.1.1 (a) – FCoE Card Summary module

FCoE Attributes

Information such as Interrupt modes (MSI/MSI-X/INTx), SCSI mode and the card state are provided in this module.

Figure 7.7.1.1 (b) – FCoE Attributes module

Chelsio T5/T4 Unified Wire For Linux Page 291

XVII. Unified Wire Manager (UM)

7.7.1.2. FCoE Port

This is an actual N_Port which communicates with the fabric and performs FIP and FCoE device discovery. This page lets the user to retrieve all the FCoE specific port information and also extend NPIV management support. It contains the following sections:

FCoE Port Summary

The SCSI adapter name and the underlying ENODE MAC address of the physical port can be found here.

Figure 7.7.1.2 (a) – FCoE Port Summary module for Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 292

XVII. Unified Wire Manager (UM)

FCoE Port Attributes

This module provides details about link status and port identifiers such as WWPN, WWNN, FC

ID and NPort MAC Address. The module also contains fabric information such as fabric name,

VLAN on which the FCoE service is currently running and the number of SCSI targets that are being discovered by this port. Port speed being mentioned in this section varies on the card type (10G/1G) being used. Note that only class 3 service is supported by the initiator for now and the frame size is fixed to 2128 bytes as per spec.

Figure 7.7.1.2 (b) – FCoE Port Attributes module for Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 293

XVII. Unified Wire Manager (UM)

FCoE NPIV management

NPIV is a fibre channel facility allowing multiple N_Port IDs to share a single physical N_Port.

This module allows the user to manage virtual ports on the corresponding FCoE Port.

To create a virtual port, select the option Create and the GUI allows two ways of creating a virtual port. i. Manual: Where the user can manually create a virtual port by providing a value to the

WWPN and WWNN fields. ii. Auto-generate: Where the FCoE function auto-generates a WWPN and WWNN for the virtual port.

To delete a virtual port, select the option Delete and select the virtual port WWPN which you want to delete and click on delete.

Figure 7.7.1.2 (c) – FCoE NPIV management module

Chelsio T5/T4 Unified Wire For Linux Page 294

XVII. Unified Wire Manager (UM)

7.7.1.3. FCoE Remote Port

Remote ports are the SCSI targets that are discovered by their respective N_port/virtual ports.

The GUI conveys the same via a tree structure so that the end user knows the initiator-target mapping.

FCoE Remote Port Attributes

This module provides details about the discovered target such as target’s FC ID, WWPN and

WWNN so that the user can identify the discovered target accordingly.

Figure 7.7.1.3 (a)

– FCoE Remote Port Attributes module

Chelsio T5/T4 Unified Wire For Linux Page 295

XVII. Unified Wire Manager (UM)

FCoE Remote Port Lun Details

This module provides the LUN information such as size of the LUN, SCSI address, and LUN address. For Linux, the SCSI address is displayed in H:C:T:L (Host:Channel:Target:Lun) format and for Windows, it is displayed in P:B:T:L(SCSI Port:Bus:Target:Lun) format.

Figure 7.7.1.3 (b) – FCoE Remote Port Lun Details module

Chelsio T5/T4 Unified Wire For Linux Page 296

XVII. Unified Wire Manager (UM)

7.7.1.4. FCoE Virtual Port

A virtual port allows multiple Fibre Channel initiators to occupy a single physical port, easing hardware requirements in SAN design, especially where virtual SANs are called for. The virtual ports appear under their respective N_Ports after creation and the GUI conveys it via a tree structure so that the end user knows the N_port-VN_Port mapping. It contains the following modules:

FCoE Virtual Port Summary

The SCSI adapter name and the underlying ENODE MAC address of the physical port can be found here.

Figure 7.7.1.4 (a) – FCoE Virtual Port Summary module for Linux Agent

Chelsio T5/T4 Unified Wire For Linux Page 297

XVII. Unified Wire Manager (UM)

FCoE Virtual Port Attributes

The module provides details about link status and port identifiers such as WWPN, WWNN, FC

ID and Virtual NPort MAC Address. The module also contains fabric information such as fabric name, VLAN on which the FCoE service is currently running and the number of SCSI targets that are being discovered by this virtual port. Port speed being mentioned in this section varies on the card type (10G/1G) being used. Note that only class 3 service is supported by the initiator for now and the frame size is fixed to 2128 bytes as per spec.

Figure 7.7.1.4 (b) – FCoE Virtual Port Attributes module

Chelsio T5/T4 Unified Wire For Linux Page 298

XVII. Unified Wire Manager (UM)

FCoE Remote Port Attributes

This module provides details about the discovered target for remote port associated with virtual port . Details such as target’s FC ID, WWPN and WWNN are provided so that the user can identify the discovered target accordingly.

Figure 7.7.1.4 (c) - FCoE Remort Port Attributes module

FCoE Remote Port Lun Details

This module provides LUN information for remote port associate with virtual port. Details such as size of the LUN, SCSI address, and LUN address are provided. For Linux, the SCSI address is displayed in H:C:T:L (Host:Channel:Target:Lun) format and for Windows, it is displayed in

P:B:T:L(SCSI Port:Bus:Target:Lun) format.

Figure 7.7.1.4 (d) - FCoE Remote Port Lun Details module

Chelsio T5/T4 Unified Wire For Linux Page 299

XVII. Unified Wire Manager (UM)

7.7.2. iSCSI initiator (Linux, Windows)

All supported iSCSI initiators can be managed from this page. The supported initiators on

Windows are Microsoft and Chelsio iSCSI initiator (T5/T4 adapters). On Linux, Open iSCSI initiator is supported. The modules available on this page are:

Initiator nodes

This module lists the initiator nodes / virtual adapters configured in the initiator stack. The node can be enabled or disabled (Chelsio node cannot be disabled in Windows), and its properties can be viewed and edited in this module. In the Chelsio Linux stack, new initiator nodes can be created too. Disabling the initiator causes it to log out of any iSCSI targets that it is connected to, thus removing any disks provided by the iSCSI targets that were connected. Use the Disable option with caution. The CHAP authentication secret should be between 12 and 16 characters in length, and the initiator's IQN name should start with "iqn.".

Figure 7.7.2 (a) - Open iSCSI initiator

Chelsio T5/T4 Unified Wire For Linux Page 300

XVII. Unified Wire Manager (UM)

Figure 7.7.2 (b) - Microsoft iSCSI initiator

Chelsio T5/T4 Unified Wire For Linux Page 301

XVII. Unified Wire Manager (UM)

Figure 7.7.2 (c) - Chelsio iSCSI initiator

Chelsio T5/T4 Unified Wire For Linux Page 302

XVII. Unified Wire Manager (UM)

Discover targets

iSCSI targets can be discovered by providing the IP address and TCP port (usually 3260) of the target. In Windows, you can specify the initiator HBA to use and its IP address. The discovery operation fetches the targets found at that Portal (combination of IP address and TCP port). The discovery operation also fetches all the other Portals that the target(s) are listening on. The discovered target can be deleted if required. Please note that all the Portals that the target sent are listed. The delete operation will not work on all the portals, only on the original discovery portal (the IP address and TCP Port specified when discovering the target).

Note

If there are any pre-existing iSCSI sessions established to the target, deletion of the target Portal from the discovered targets list will fail.

Figure 7.7.2 (d) - Discover targets module

Chelsio T5/T4 Unified Wire For Linux Page 303

XVII. Unified Wire Manager (UM)

Figure 7.7.2 (e) - Discover targets module

Chelsio T5/T4 Unified Wire For Linux Page 304

XVII. Unified Wire Manager (UM)

Targets

The iSCSI targets that have been discovered, or are currently connected, are listed here. You may login, logout and delete the target from the initiator's configuration. In Windows, for the

Microsoft iSCSI initiator, connections to an already established iSCSI session can be added or deleted. For the Microsoft iSCSI initiator or the Open iSCSI initiator, you may specify the authentication details and digest settings while logging in. For Chelsio Linux initiator, these settings should be set prior to attempting a login, in the Initiator nodes module. If a target is connected, the sessions and connections to the target, and the disks provided by the target will be listed.

In Windows, you can specify the initiator HBA to use and its IP address while logging in.

Figure 7.7.2 (f) - Targets module after logging in

Chelsio T5/T4 Unified Wire For Linux Page 305

XVII. Unified Wire Manager (UM)

7.7.3. FO iSCSI Initiator (Linux)

Full Offload iSCSI Hardware Information

PCI, firmware and other adapter related details are provided in this module. Select the Chelsio adapter for which you want to view properties from the Select a T4 Card drop-down list and the module will expand to display related properties. You can also view details like link id, status, enode mac, etc of all the ports of the selected adapter.

Figure 7.7.3 (a) - Full Offload iSCSI Hardware Information module

Chelsio T5/T4 Unified Wire For Linux Page 306

XVII. Unified Wire Manager (UM)

FO iSCSI Manage Ports

Here you can configure various port settings like VLAN id, Maximum Transmission Unit (MTU) and IP. Select a Chelsio adapter from Select a T4 Card drop-down list and then select the port for which you want set any of the aforementioned properties. MTU can be set between 1500-

9000 bytes. VLAN id can be set within the range 0-4094 (enter 0 to disable it). The IP type can be IPV4 (static) or DHCP.

The Port Up and Port Down buttons will enable and disable the selected port respectively. The

Clear IP button deletes values set for the IP Type, IP, Subnet Mask and Gateway properties and resets them.

Figure 7.7.3 (b) - FO iSCSI Manage Ports module

FO iSCSI Initiator Properties

In the FO iSCSI Initiator Properties module, you can configure FO iSCSI Initiator by setting different properties like enabling/disabling CHAP authentication, setting Header and Data digest, etc.

Chelsio T5/T4 Unified Wire For Linux Page 307

XVII. Unified Wire Manager (UM)

Figure 7.7.3 (c) - FO iSCSI Initiator Properties

FO iSCSI Manage Instances

The FO iSCSI Initiator service maintains multiple instances of a target depending on the discovery method. In this module, you can set upto 8 instances. Configurable parameters include initiator node name (IQN), alias (friendly) name, Initiator (CHAP) Username and password.

Chelsio T5/T4 Unified Wire For Linux Page 308

XVII. Unified Wire Manager (UM)

Figure 7.7.3 (d) - FO iSCSI Manage Instances module

7.7.4. iSCSI Target (Linux)

This page allows to create new Targets and manage them (add/delete portals, add/delete

LUNs, add/delete ACLs).It also provides information on Session details. Viewing and modifying

Target properties is also available. The modules available on this page are as below:

Target Stack Globals

This module displays various global properties of a currently connected iSCSI target.

Authentication priority between CHAP and ACL can be set here.

Chelsio T5/T4 Unified Wire For Linux Page 309

XVII. Unified Wire Manager (UM)

Figure 7.7.4 (a) - Target Stack Globals module

Chelsio T5/T4 Unified Wire For Linux Page 310

XVII. Unified Wire Manager (UM)

Target properties

Properties such as Target name and Alias, Max Data Receive Length, Authentication mode related to a specific iSCSI target can be viewed and modified here. iSCSI targets can be started/stopped or deleted.

Figure 7.7.4 (b) - Target properties module

Chelsio T5/T4 Unified Wire For Linux Page 311

XVII. Unified Wire Manager (UM)

Session details

Details including Session ID, Initiator IQN and Connections List of all discovered and currently connected iSCSI targets are listed here.

Figure 7.7.4 (c) - Session Details module

Chelsio T5/T4 Unified Wire For Linux Page 312

XVII. Unified Wire Manager (UM)

New Target Creation

New iSCSI target can be created here by specifying the Target IQN and Target Alias name.

Figure 7.7.4 (d) - New Target Creation module

Chelsio T5/T4 Unified Wire For Linux Page 313

XVII. Unified Wire Manager (UM)

7.7.5. LUNs

Various Logical Units created in an iSCSI Target can be managed here. The modules available on this page are as below:

View/Edit iSCSI Target LUNs

This module displays various Logical Units created in an iSCSI Target. Selected LUNs can be deleted.

Figure 7.7.5 (a) - View/Edit iSCSI Target LUNs module

Chelsio T5/T4 Unified Wire For Linux Page 314

XVII. Unified Wire Manager (UM)

Add LUN

New LUNs can be added here by providing various parameters like Target Name, Target

Device and RAM Disk Size etc. RW (Read-Write) and RO (Read Only) are the two kinds of permissions that can be set. If Ram Disk is selected, then a minimum of 16 MB should be provided.

Figure 7.7.5 (b) - Adding a new LUN

Chelsio T5/T4 Unified Wire For Linux Page 315

XVII. Unified Wire Manager (UM)

7.7.6. Portal Groups

Portal details for currently connected iSCSI Targets can be viewed and added here. The modules available on this page are as below:

View/Edit iSCSI Target Portals

Portal List on the left displays details of the portal group on which an iSCSI target is listening and the related info is displayed on the right under Portal Details. Selected portals can be deleted.

Figure 7.7.6 (a) - View/Edit iSCSI Target Portals module

Chelsio T5/T4 Unified Wire For Linux Page 316

XVII. Unified Wire Manager (UM)

Add Portal

New Portals can be added here by choosing the specific target and Portal IP address. The Port number should be 3260.

Figure 7.7.6 (b) - Adding a new Portal

Chelsio T5/T4 Unified Wire For Linux Page 317

XVII. Unified Wire Manager (UM)

7.7.7. ACLs

ACLs configured for currently connected iSCSI Targets can be managed here. The modules available on this page are as below:

View/Edit iSCSI Target ACLs

This module displays details for all the ACLs configured for an iSCSI Target. Selected ACLs can be deleted.

Figure 7.7.7 (a) - Target ACL operations module

Chelsio T5/T4 Unified Wire For Linux Page 318

XVII. Unified Wire Manager (UM)

Add ACL

New ACLs can be configured by specifying Target name, initiator IQN name, IP address and permission type.

Figure 7.7.7 (b) - Adding new ACL

Chelsio T5/T4 Unified Wire For Linux Page 319

XVII. Unified Wire Manager (UM)

7.8.

Hardware Features

The Hardware module lists the status of configuration modules under Hardware Features section, running on the agent.

Figure 7.8 – Hardware module for a Linux Agent

7.8.1. Filtering (Linux)

Filtering feature enhances network security by controlling incoming traffic as they pass through network interface based on source and destination addresses, protocol, source and receiving ports, or the value of some status bits in the packet. The modules available on this page are as below:

T3 Filtering configuration

T3 Filtering options can be set only when offload driver (t3_tom) is not loaded.

This module lists various parameters which can be set while determining filtering options for a system IP. You can set the maximum number of filters and also add/delete filters. A filter with default values (the Action field set to pass; the Protocol field set to any) is created at the time of configuring the filtering module. To remove the default filter, enter 0 in the Set Maximum

Filters field and click on Set Filters. The fields lfName and FilterId are mandatory. The Action field is set to pass and the Protocol field is set to any by default. Other possible values for the

Protocol field are tcp, udp and frag. The Priority field can be used to determine the priority of a filter when Vlan ids are same. Insert at position features allows user to add a filter at a specified position.

Chelsio T5/T4 Unified Wire For Linux Page 320

XVII. Unified Wire Manager (UM)

Figure 7.8.1(a) – T3 Filtering Configuration module

Note

Results for actions like adding a new filter or setting maximum filters make some time to reflect. Highlight the system item in the tree menu on the left, and click

"Refresh system", to refresh data from the system, in case the updated settings are not being shown.

T4 Filtering configuration

T4 Filtering options can be set only when offload driver (t4_tom) is not loaded.

A list of pre-defined filter selection combinations is displayed. The combination fragmentation,

mpshittype, protocol, vlan, port, fcoe is active by default. To select a different combination, highlight it in the Combinations list by clicking and click “Set Active Combination”.

You can create filter rules for any combination in the list. However, filter rule created only for the

Active Combination will apply. To create a new rule, select a combination and click “Add a filter rule”. The FILTERID and T4 CARD fields are mandatory. After providing appropriate values for the parameters click “Save Changes”.

Note

For a detailed explanation regarding different fields, please refer cxgbtool manual by running man cxgbtool command on Management Agent CLI.

Chelsio T5/T4 Unified Wire For Linux Page 321

XVII. Unified Wire Manager (UM)

Figure 7.8.1(b)

– T4 Filtering Configuration module

Chelsio T5/T4 Unified Wire For Linux Page 322

XVII. Unified Wire Manager (UM)

7.8.2. Traffic Management (Linux)

Using this page, one can add/delete/modify offload policies only in the presence of offload driver

(t3_tom for T3 adapters; t4_tom for T4 adapters).

Traffic Management configuration

The Chelsio Card section on the left displays all the cards available in the server and their corresponding policies on the right. Policies can be added and deleted. Policy Details displays the primitives (maximum 8) and actions which can be modified. For more details on creating policies, please refer to COP man pages.

Figure 7.8.2 - Traffic Management Configuration module

Chelsio T5/T4 Unified Wire For Linux Page 323

XVII. Unified Wire Manager (UM)

7.8.3. Boot

T4/T5 Save Config File (Linux)

This module displays the current T5/T4 configuration tuning option selected. You can also change the tuning option by selecting the config file for each option located in /ChelsioUwire-

x.x.x.x/src/network/firmware. For instance, to select Low latency Networking for T4 adapter, locate the file, t4-config.txt, in /ChelsioUwire-x.x.x.x/src/network/firmware/low_latency_config directory.

Figure 7.8.3 (a) – T4/T5 Save Config File module

Chelsio T5/T4 Unified Wire For Linux Page 324

XVII. Unified Wire Manager (UM)

T4/T5 Boot Option ROM management

This module allows managing the PXE and FCoE boot capability for Chelsio T5 and T4 adapters. The Option ROM (PXE and FCoE) may be installed to or erased from the card. The version of Option ROM flashed can be viewed here.

Figure 7.8.3 (b) – T4/T5 Option ROM Management module

Chelsio T5/T4 Unified Wire For Linux Page 325

XVII. Unified Wire Manager (UM)

T4/T5 Boot Configuration

This module can be used to view and configure PXE, FCoE and iSCSI Option ROM settings for

Chelsio T4 and T5 adapters.

PXE physical functions and order of ports for PXE boot can be selected using the PXE option.

You can also enable/disable PXE BIOS and set VLAN.

The FCoE option can be used to configure FCoE Option ROM settings. Using the Function parameter, you can set port order for target discovery and discovery timeout. The Boot parameter can be used to discover targets and view properties of LUNs assigned to the targets.

Clicking on the Discover Targets button will list all the discovered targets and clicking on target will list assigned LUNs on the right. Select a LUN from the list to view details. Show WWPN parameter will display the WWPNs of all the ports.

There are four configurable parameters available under the iSCSI option: Function, Initiator,

Network and Boot Devices. Using the Function parameter you can enable/disable BIOS, set port order for target discovery and discovery timeout. The Initiator parameter allows you to configer initiator properties like IQN name, header digest and data digest. You can also set

CHAP authentication method or disable it. The Network parameter allows you to configure various settings on the port like enabling/disabling IPv6 support, specifying initiator IP type, etc.

Using the Boot Devices option you can set various iSCSI target properties.

Note

Enable Option ROM only if you are planning to boot the system via PXE or install the operating system on discovered iSCSI or FCoE LUN.

Chelsio T5/T4 Unified Wire For Linux Page 326

XVII. Unified Wire Manager (UM)

Figure 7.8.3 (c) - PXE Boot configuration for T4 CNAs

Chelsio T5/T4 Unified Wire For Linux Page 327

XVII. Unified Wire Manager (UM)

Figure 7.8.3 (d) - FCoE Boot configuration for T4 CNAs: Function parameter

Figure 7.8.3 (e) - FCoE Boot configuration for T4 CNAs: Boot parameter

Chelsio T5/T4 Unified Wire For Linux Page 328

XVII. Unified Wire Manager (UM)

Figure 7.8.3 (f) - FCoE Boot configuration for T4 CNAs: Show WWPN parameter

Figure 7.8.3 (g) - iSCSI Boot configuration for T4 CNAs: Function parameter

Chelsio T5/T4 Unified Wire For Linux Page 329

XVII. Unified Wire Manager (UM)

Figure 7.8.3 (h) - iSCSI Boot configuration for T4 CNAs: Initiator parameter

Chelsio T5/T4 Unified Wire For Linux Page 330

XVII. Unified Wire Manager (UM)

Figure 7.8.3 (i) - iSCSI Boot configuration for T4 CNAs: Network parameter

Chelsio T5/T4 Unified Wire For Linux Page 331

XVII. Unified Wire Manager (UM)

Figure 7.8.3 (j) - iSCSI Boot configuration for T4 CNAs: Boot Devices parameter

7.8.4. Bypass

You can use the Bypass page to configure various settings for Chelsio’s bypass adapters like setting bypass operation mode, creating rules (filters), starting/stopping BA server, etc. There are two modules available: Bypass Configuration and Redirect Configuration.

Bypass Configuration

In the Bypass Configuration module, you can view the status and start/stop the BA server accordingly. The adapter will redirect packets using the mode speciffied in the Default bypass

mode field unless otherwise specified in the Current bypass mode.

The Watchdog timer is used to ensure that if there is a software failure, the switch will enter the default state. The Watchdog timeout value should be provided in milliseconds.

Chelsio T5/T4 Unified Wire For Linux Page 332

XVII. Unified Wire Manager (UM)

For more information on different bypass modes and configurational parameters, see Bypass

Driver chapter.

Figure 7.8.4 (a) - Bypass Configuration module

Redirect Configuration

In the Redirect Configuration module, you can set rules (filters), based on which the bypass adapter will redirect packets. You can group rules into tables. You can save the currently configured tables and rules for a bypass adapter into a shell script using the Download

Configuration button.

The Table Configuration tab displays BA server status and the number of tables created. You can create new tables or perform various actions on the existing ones.

Delete table: Delete the selected table and all the rules present in it.

Purge table: Delete all the rules present in the selected table. This action will not delete the table.

Activate table: Enable the selected table.

Deactivate table: Disable the selected table.

Chelsio T5/T4 Unified Wire For Linux Page 333

XVII. Unified Wire Manager (UM)

Create table: Create a new table. The new table created will be inactive by default. Use the Activate table option to enable it. You can cretate upto 5 tables.

In the Rules Configuration tab, you can add, delete and configure rules. Use the Add a Filter

row button to add a new rule by specifying the rule id in the INDEX field and providing the required parameters. Finally, click Save Changes.

To edit an existing rule, select the corresponding checkbox, change the desired paremeters and click Save Changes.

To delete a rule, select the corresponding checkbox and click Delete a Filter. Finally, click

Save Changes. You can delete multiple rules using this method.

For more information on Redirection, see Bypass Driver chapter.

Figure 7.8.4 (b) - Redirect Configuration module: Table configuration tab

Chelsio T5/T4 Unified Wire For Linux Page 334

XVII. Unified Wire Manager (UM)

Figure 7.8.4 (c) - Redirect Configuration module: Rules configuration tab

Chelsio T5/T4 Unified Wire For Linux Page 335

XVII. Unified Wire Manager (UM)

7.8.5. T4 Egress Class Schedulers

Schedulers can be set only when T5/T4 network driver (cxgb4) is loaded.

Egress Queue Map

Using this module, you can bind (map) NIC (non-offloaded) Tx queues to Tx Scheduler classes.

7.8.5 (a) – Egress Queue Map module

Egress Packet Scheduler

Using this module you can configure different scheduler hierarchy levels (i.e.Class Rate

Limiting, Class Weighted Round Robin and Channel Rate Limiting). Based on the parameters specified, different scheduler levels can be configured. To know more about the levels and

related parameters, please refer Traffic Management chapter ( Click here ).

7.8.5 (b) – Egress Packet Scheduler module

Chelsio T5/T4 Unified Wire For Linux Page 336

XVII. Unified Wire Manager (UM)

8. Uninstalling Unified Wire Manager

This section describes the method to uninstall components of Chelsio Unified Manger.

8.1.

Uninstalling Management Agent

1. Use the following query command to determine the name of the agent RPM:

[root@host]# rpm –qa | grep chelsio-uwire_mgmt-agent

2. Now, execute the following command with the result from the above query to uninstall

Management Agent:

E.g. for RHEL 6.3:

[root@host]# rpm –e chelsio-uwire_mgmt-agent-rhel6u3-2.2-xyz.x86_64

8.2.

Uninstalling Management Client

1. Use the following query command to determine the name of the client RPM:

[root@host]# rpm –qa | grep chelsio-uwire_mgmt-client

2. Now, execute the following command with the result from the above query to uninstall

Management Client:

E.g. for RHEL 6.3:

[root@host]# rpm –e chelsio-uwire_mgmt-client-rhel6u3-2.2-xyz.x86_64

Chelsio T5/T4 Unified Wire For Linux Page 337

XVII. Unified Wire Manager (UM)

8.3.

Uninstalling Management Station

1. Use the following query command to determine the name of the Management Station RPM:

[root@host]# rpm –qa | grep chelsio-uwire_mgmt-station

2. Now, execute the following command with the result from the above query to uninstall

Management Station:

E.g. for RHEL 6.3:

[root@host]# rpm –e chelsio-uwire_mgmt-station-rhel6u3-2.2-xyz.x86_64

Chelsio T5/T4 Unified Wire For Linux Page 338

Chapter XVIII. Unified Boot

XVIII. Unified Boot

Chelsio T5/T4 Unified Wire For Linux Page 339

Chapter XVIII. Unified Boot

1. Introduction

PXE is short for Preboot eXecution Environment and is used for booting computers over an ethernet network using a Network Interface Card (NIC).

FCoE SAN boot process involves installation of an operating system to an FC/FCoE disk and then booting from it. iSCSI SAN boot process involves installation of an operating system to an iSCSI disk and then booting from it.

This section of the guide explains how to configure and use Chelsio Unified Boot Option

ROM which flashes PXE and FCoE Option R OM onto Chelsio’s converged network adapters

(CNAs). It adds functionality like Network/FCoE SAN boot/ iSCSI SAN boot.

This section also describes the use and configuration of Chelsio’s DUD for OS installations via PXE server or on FC/FCoE LUN and iSCSI LUN. This solution can be used for installing operating systems over an ethernet network/SAN using Chelsio’s Converged Network Adapters

(CNAs). The Chelsio Unified Option ROM resides on the flash memory of the adapter from where it is loaded by an x86 system BIOS during power on initialization.

1.1.

Hardware Requirements

1.1.1. Supported Adapters

The following are the currently shipping Chelsio Adapters that are compatible with Chelsio

Unified Boot software:

T520-LL-CR

T520-SO-CR

T520-CR

T540-CR

T580-LP-CR

T420-CR

T440-CR

T422-CR

T404-BT

T420-BCH (Only PXE)

T420-SO-CR

T440-LP-CR

T420-LL-CR

T420-BT

Chelsio T5/T4 Unified Wire For Linux Page 340

Chapter XVIII. Unified Boot

1.2.

Software Requirements

Chelsio Unified Boot Option ROM software requires Disk Operating System to flash

PXE, FCoE ROM and iSCSI ROM onto Chelsio adapters.

The Chelsio Driver Update Disk driver has been developed to run on 64-bit Linux platforms. Following is the list of Drivers/Software and supported Linux distributions:

Linux Distribution

RHEL6.4, 2.6.32-358.el6

SLES11SP2, 3.0.13-0.27

SLES11SP3, 3.0.76-0.11

Driver/Software (DUDs)

PXE, FCoE, iSCSI

PXE, FCoE, iSCSI

PXE, FCoE, iSCSI

Chelsio T5/T4 Unified Wire For Linux Page 341

Chapter XVIII. Unified Boot

2. Flashing configuration file, firmware & option ROM

Chelsio legacy flash utility (cfut4.exe) is used to program the PXE Option ROM image onto the Chelsio HBAs.

Example # 1: This example assumes you are using a USB flash drive as a storage media for the necessary files. Follow the steps below: i. Create a DOS bootable USB flash drive. ii. Create CHELSIO folder on the USB flash drive. iii. If you haven’t done already, download ChelsioUwire-x.x.x.x.tar.gz from Chelsio Download

Center, service.chelsio.com

iv. Untar the downloaded package.

v. Change your working directory depending on the adapter type. For example, for T5 adapters, change your working directory to ChelsioUwire-x.x.x.x/Uboot/OptionROM/T5 vi. Copy all the files and place them in the CHELSIO folder created on the USB flash drive. vii. Plug in the USB flash drive in the system on which the Chelsio CNA is installed. viii. Reboot the system and go into the BIOS setup. ix. Make the USB flash drive as the primary boot device. Save the changes and exit. x. Once the system boots from the USB flash drive, navigate to the CHELSIO directory:

C:\>cd CHELSIO

Chelsio T5/T4 Unified Wire For Linux Page 342

Chapter XVIII. Unified Boot xi. Run the following command to list all Chelsio CNAs present in the system. The list displays a unique index for each CNA found.

C:\CHELSIO>cfut4 –l xii. Delete any previous version of Option ROM flashed on the CNA:

C:\CHELSIO>cfut4 –d <idx> -xb –xc

Here, idx is the CNA index found in step xi (0 in this case) xiii. Delete any previous firmware and configuration file using the following command:

C:\CHELSIO>cfut4 -d <idx> -xh –xf

Here, idx is the CNA index found in step xi (0 in this case)

Chelsio T5/T4 Unified Wire For Linux Page 343

Chapter XVIII. Unified Boot xiv. Run the following command to flash the appropriate configuration file (t4-config.txt for T4 adapters; t5-config.txt for T5 adapters).

C:\CHELSIO>cfut4 -d <idx> -uh <configuration_file>.txt

Here, idx is the CNA index found in step xi (0 in this case) and configuration_file.txt

is the configuration file present in the CHELSIO folder.

Chelsio T5/T4 Unified Wire For Linux Page 344

Chapter XVIII. Unified Boot xv. Run the following command to flash the appropriate firmware (t4fw-x.x.xx.x.bin for T4 adapters; t5fw-x.x.xx.x.bin for T5 adapters).

C:\CHELSIO>cfut4 -d <idx> -uf <firmware_file>.bin

Here, idx is the CNA index found in step xi (0 in this case) and firmware_file.bin

is the firmware image file present in the CHELSIO folder.

Chelsio T5/T4 Unified Wire For Linux Page 345

Chapter XVIII. Unified Boot xvi. Use the following command to flash the appropriate option ROM onto the CNA (cubt4.bin for T4 adapters; cuwlbt4.bin for T5 adapters):

C:\CHELSIO>cfut4 -d <idx> -ub <option_rom_file>.bin

Here, idx is the CNA index found in step xi (0 in this case) and option_rom_file.bin

is the unified option ROM image file.

Note

On HP DL160 systems with T4 adapters, please use cubt40.bin instead of cubt4.bin xvii. Reboot the system for changes to take effect.

Chelsio T5/T4 Unified Wire For Linux Page 346

Chapter XVIII. Unified Boot

3. Configuring PXE Server

The following components are required to configure a server as PXE Server:

DHCP Server

TFTP Server

PXE server configuration steps for Linux can be found on following links:

 http://linux-sxs.org/internet_serving/pxeboot.html

 http://www.howtoforge.com/ubuntu_pxe_install_server

PXE server configuration steps for Windows can be found on following links:

 http://technet.microsoft.com/en-us/library/cc771670%28WS.10%29.aspx

 http://tftpd32.jounin.net/ (Use port # 67, set PXE option and provide bootable file name in settings)

 http://unattended.sourceforge.net/pxe-win2k.html

Note

Chelsio Communications does not take any responsibility about contents given in above mentioned links. Those are given for example purposes only.

Chelsio T5/T4 Unified Wire For Linux Page 347

Chapter XVIII. Unified Boot

4. PXE boot process

Before proceeding, please ensure that the Chelsio HBA has been flashed with the provided configuration file, firmware and option ROM by following the procedure mentioned in section 2

( Click here ).

4.1.

For Legacy PXE boot

i. After configuring the PXE server, make sure the PXE server works. Then reboot the client machine. ii. Press [Alt+C] when the message Chelsio T4 Unified Boot Bios, Copyright (C) 2003-2012

Chelsio Communications Press <Alt-C> to Configure T4 Card(s). Press <Alt-S> to skip

BIOS" appears on the screen to enter the configuration utility. iii. The configuration utility will appear as below: iv. Choose the CNA on which you flashed the option ROM image. Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 348

Chapter XVIII. Unified Boot v. Enable the Adapter BIOS using arrow keys if not already enabled. Hit [ENTER].

Note Use the default values for Boot Mode, EDD and EBDA Relocation parameters,

unless instructed otherwise. vi. Choose PXE from the list to configure. Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 349

Chapter XVIII. Unified Boot vii. Use the arrow keys to highlight the appropriate function among the 4 supported NIC functions and hit [Enter] to select. viii. Enable NIC function bios if not already enabled.

Chelsio T5/T4 Unified Wire For Linux Page 350

Chapter XVIII. Unified Boot ix. Choose the boot port to try the PXE boot. It is recommended to only enable functions and ports which are going to be used. Please note that enabling NIC Func 00 will enable port 0 for PXE, enabling NIC Func 01 will enable port 1 and so on for NIC function. Please refer the table below:

NIC Function enabled Ports enabled

NIC Func00

NIC Func01

NIC Func02

NIC Func03

00

01

02

03

Note

VLAN id currently not supported in Legacy and UEFI environments. x. Hit [Esc] and then [Y] to save configuration changes. xi. Reboot the system. xii. Hit [F2] or [DEL] or any other key as mentioned during system startup to enter the system setup.

Chelsio T5/T4 Unified Wire For Linux Page 351

Chapter XVIII. Unified Boot xiii. Allow the Chelsio option ROM to initialize and setup PXE devices.

DO NOT PRESS ALT-S to skip Chelsio option ROM. xiv. In the system setup, choose any of the Chelsio PXE devices as the first boot device. xv. Reboot. DO NOT PRESS ALT-S to skip Chelsio option ROM, during POST. xvi. Hit [F12] key when prompted to start PXE boot.

Chelsio T5/T4 Unified Wire For Linux Page 352

Chapter XVIII. Unified Boot

4.2.

For uEFI PXE Boot

Important

Only uEFI 2.1 supported.

Any other uEFI version is NOT SUPPORTED and may render your system unusable.

Both Legacy and uEFI environments are supported on T4 adapters.

Whereas, only Legacy environment is currently supported on T5 adapters.

This section describes the method to configure and use Chelsio uEFI PXE interfaces. i. After configuring the PXE server, make sure the PXE server works. Then reboot the client machine. ii. Configure the system having CNA to boot in UEFI mode.

Chelsio T5/T4 Unified Wire For Linux Page 353

Chapter XVIII. Unified Boot iii. Boot to EFI Shell. iv. Issue command drivers to note down the handle# of Chelsio driver.

Chelsio T5/T4 Unified Wire For Linux Page 354

Chapter XVIII. Unified Boot v. Then issue command drvcfg -s <handle#> to invoke the configuration utility.

Chelsio T5/T4 Unified Wire For Linux Page 355

Chapter XVIII. Unified Boot vi. In the configuration utility, choose the CNA on which the option ROM image was flashed and Hit [Enter]. vii. Choose config utility if you want to configure adapter for PXE boot, else choose flash utility if you want to flash/erase firmware, option ROM image or configuration parameters.

Chelsio T5/T4 Unified Wire For Linux Page 356

Chapter XVIII. Unified Boot viii. If config utility is chosen, go to step (ix), else goto step (xviii). ix. Enable the Adapter BIOS if not already enabled. Hit [Enter]. x. Choose PXE as the function to use and configure.

Chelsio T5/T4 Unified Wire For Linux Page 357

Chapter XVIII. Unified Boot xi. Choose the appropriate function among the 4 NIC functions available. xii. Enable function Bios if not already enabled.

Note

VLAN is currently not supported.

Chelsio T5/T4 Unified Wire For Linux Page 358

Chapter XVIII. Unified Boot xiii. Choose the boot port to try the PXE boot. It is recommended to only enable functions and ports which are going to be used. Please note that enabling NIC Func 00 will enable port 0 for PXE, enabling NIC Func 01 will enable port 1 and so on for NIC function. Please refer the table below:

NIC Function enabled Ports enabled

NIC Func00

NIC Func01

NIC Func02

NIC Func03

00

01

02

03 xiv. Hit [Esc] and then [Y] to save the configuration. xv. Reboot the system. xvi. Boot to UEFI Boot Manager and choose any of the Chelsio PXE devices for PXE boot.

Chelsio T5/T4 Unified Wire For Linux Page 359

Chapter XVIII. Unified Boot xvii. To confirm if uEFI Chelsio PXE has loaded successfully, please boot to uEFI shell and issue the command drivers . Now, note that the type field should be B as shown below.

Chelsio T5/T4 Unified Wire For Linux Page 360

Chapter XVIII. Unified Boot xviii. If flash utility is chosen, choose any of available options and Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 361

Chapter XVIII. Unified Boot xix. If you have chosen to update option ROM image, then enter the full path to the file. E.g.:- If the pen drive is device fs0 and the option ROM image file is in fs0:\CHELSIO\cubt4.bin, then enter CHELSIO\cubt4.bin

.

Chelsio T5/T4 Unified Wire For Linux Page 362

Chapter XVIII. Unified Boot xx. Restart the system for changes to take place.

Chelsio T5/T4 Unified Wire For Linux Page 363

Chapter XVIII. Unified Boot

5. FCoE boot process

Before proceeding, please ensure that the Chelsio HBA has been flashed with the provided configuration file, firmware and option ROM by following the procedure mentioned in section 2

( Click here ).

Important

FCoE boot is currently not supported on T5 adapters.

5.1.

For Legacy FCoE boot

i. Reboot the system. ii. Press [Alt+C] when the message Chelsio T4 Unified Boot Bios, Copyright (C) 2003-2012

Chelsio Communications Press <Alt-C> to Configure T4 Card(s). Press <Alt-S> to skip

BIOS" appears on the screen to enter the configuration utility. iii. The configuration utility will appear as below: iv. Choose the CNA on which you flashed the option ROM image. Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 364

Chapter XVIII. Unified Boot v. Enable the Adapter BIOS if not already enabled. Hit [ENTER].

Note Use the default values for Boot Mode, EDD and EBDA Relocation parameters,

unless instructed otherwise. vi. Choose FCoE from the list to configure and hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 365

Chapter XVIII. Unified Boot vii. Choose the first option (function parameters) from the list of parameter type and hit [Enter]. viii. Enable FCoE BIOS if not already enabled.

Chelsio T5/T4 Unified Wire For Linux Page 366

Chapter XVIII. Unified Boot ix. Choose the order of the ports to discover FCoE targets. x. Set discovery timeout to a suitable value. Recommended value is >= 30.

Chelsio T5/T4 Unified Wire For Linux Page 367

Chapter XVIII. Unified Boot xi. Hit [Esc] and then [Y] to save the configuration. xii. Go back and choose boot parameters to configure.

Chelsio T5/T4 Unified Wire For Linux Page 368

Chapter XVIII. Unified Boot xiii. Select the first boot device and hit [Enter] to discover FC/FCoE targets connected to the switch. Wait till all targets reachable are discovered. xiv. List of discovered targets will be displayed. Highlight a target using the arrow keys and hit

[Enter] to select.

Chelsio T5/T4 Unified Wire For Linux Page 369

Chapter XVIII. Unified Boot xv. From the list of LUNs displayed for the selected target, choose one on which operating system has to be installed. Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 370

Chapter XVIII. Unified Boot xvi. Hit [Esc] and then [Y] to save the configuration. xvii. Reboot the machine. xviii. During POST, allow the Chelsio option ROM to discover FCoE targets.

Chelsio T5/T4 Unified Wire For Linux Page 371

Chapter XVIII. Unified Boot xix. Enter BIOS setup and choose FCoE disk discovered via Chelsio adapter as the first boot device. xx. Reboot and install the required OS to the FCoE disk.

Chelsio T5/T4 Unified Wire For Linux Page 372

Chapter XVIII. Unified Boot

5.2.

For uEFI FCoE Boot

Important

Only uEFI 2.1 supported.

Any other uEFI version is NOT SUPPORTED and may render your system unusable.

This section describes the method to perform FCoE boot on uEFI platforms. i. Configure the system having Chelsio CNA to boot in uEFI mode. ii. Boot to EFI Shell.

Chelsio T5/T4 Unified Wire For Linux Page 373

Chapter XVIII. Unified Boot iii. Issue command drivers to note down the handle# of Chelsio driver.

Chelsio T5/T4 Unified Wire For Linux Page 374

Chapter XVIII. Unified Boot iv. Then issue command drvcfg -s <handle#> to invoke the configuration utility.

Chelsio T5/T4 Unified Wire For Linux Page 375

Chapter XVIII. Unified Boot v. In the configuration utility, choose the CNA on which the option ROM image was flashed and Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 376

Chapter XVIII. Unified Boot vi. Choose config utility if you want to configure Chelsio adapter for PXE boot, else choose

flash utility if you want to flash/erase firmware, option ROM image or configuration parameters. vii. If config utility is chosen, go to step (viii), else goto step (xxii).

Chelsio T5/T4 Unified Wire For Linux Page 377

Chapter XVIII. Unified Boot viii. Enable the Adapter BIOS if not already enabled. Hit [Enter]. ix. Choose FCoE from the list to configure and hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 378

Chapter XVIII. Unified Boot x. Choose the first option (function parameters) from the list of parameter type and hit [Enter]. xi. Enable FCoE BIOS if not already enabled.

Chelsio T5/T4 Unified Wire For Linux Page 379

Chapter XVIII. Unified Boot xii. Choose the order of the ports to discover FCoE targets. xiii. Set discovery timeout to a suitable value. Recommended value is >= 30.

Chelsio T5/T4 Unified Wire For Linux Page 380

Chapter XVIII. Unified Boot xiv. Hit [Esc] and then [Y] to save the configuration. xv. Go back and choose boot parameters to configure.

Chelsio T5/T4 Unified Wire For Linux Page 381

Chapter XVIII. Unified Boot xvi. Select the first boot device and hit [Enter] to discover FC/FCoE targets connected to the switch. Wait till all targets reachable are discovered. xvii. List of discovered targets will be displayed. Highlight a target using the arrow keys and hit

[Enter] to select.

Chelsio T5/T4 Unified Wire For Linux Page 382

Chapter XVIII. Unified Boot xviii. From the list of LUNs displayed for the selected target, choose one on which operating system has to be installed. Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 383

Chapter XVIII. Unified Boot xix. Hit [Esc] and then [Y] to save the configuration. xx. Reboot the machine. xxi. Enter uEFI BIOS Manager, set DVD and discovered LUN as first and second boot device respectively. Save changes and start the OS installation. xxii. If flash utility is chosen, choose any of available options and Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 384

Chapter XVIII. Unified Boot xxiii. If you have chosen to update option ROM image, then enter the full path to the file. E.g.:- If the pen drive is device fs0 and the option ROM image file is in fs0:\CHELSIO\cubt4.bin, then enter CHELSIO\cubt4.bin

.

Chelsio T5/T4 Unified Wire For Linux Page 385

Chapter XVIII. Unified Boot xxiv. Restart the system for the changes to take place.

Chelsio T5/T4 Unified Wire For Linux Page 386

Chapter XVIII. Unified Boot

6. iSCSI boot process

Before proceeding, please ensure that the Chelsio HBA has been flashed with the provided configuration file, firmware and option ROM by following the procedure mentioned in section 2

( Click here ).

Important iSCSI boot is currently not supported on T5 adapters.

6.1.

For Legacy iSCSI boot

i. Reboot the system. ii. Press [Alt+C] when the message Chelsio T4 Unified Boot Bios, Copyright (C) 2003-2012

Chelsio Communications Press <Alt-C> to Configure T4 Card(s). Press <Alt-S> to skip

BIOS" appears on the screen to enter the configuration utility. iii. The configuration utility will appear as below: iv. Choose the CNA on which you flashed the option ROM image. Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 387

Chapter XVIII. Unified Boot v. Enable the Adapter BIOS if not already enabled. Hit [Enter].

Note Use the default values for Boot Mode, EDD and EBDA Relocation parameters,

unless instructed otherwise. vi. Choose iSCSI from the list to configure and hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 388

Chapter XVIII. Unified Boot vii. Choose the first option (Configure Function Parameters) from the list of parameter type and hit [Enter]. viii. Enable iSCSI BIOS if not already enabled. By default, CBFT (Chelsio Boot Firmware Table) will be used.

Chelsio T5/T4 Unified Wire For Linux Page 389

Chapter XVIII. Unified Boot ix. Choose the order of the ports to discover iSCSI targets. x. Set discovery timeout to a suitable value. Recommended value is >= 30.

Chelsio T5/T4 Unified Wire For Linux Page 390

Chapter XVIII. Unified Boot xi. Hit [Esc] and then [Y] to save the configuration. xii. Go back and choose Configure Initiator Parameters to configure initiator related properties.

Chelsio T5/T4 Unified Wire For Linux Page 391

Chapter XVIII. Unified Boot xiii. Inititiar properties like IQN, Header Digest, Data Digest, etc will be displayed. Change the values appropriately or continue with the default values. Hit [F10] to save. xiv. CHAP authentication is disabled by default. To enable and configure, go back and choose

Configure CHAP Parameters

Chelsio T5/T4 Unified Wire For Linux Page 392

Chapter XVIII. Unified Boot xv. Enable CHAP authentication by selecting ENABLED in the CHAP Policy field. Next, choose either one-way or mutual as the authentication method. Finally, provide Initiator and

Target CHAP credentials according to the authentication method selected. Hit [F10] to save. xvi. Go back and choose Configure Network Parameters to configure iSCSI Network related properties.

Chelsio T5/T4 Unified Wire For Linux Page 393

Chapter XVIII. Unified Boot xvii. Select the port using which you want to connect to the target. Hit [Enter]. xviii. Select Yes in the Enable DHCP field to configure port using DHCP or No to manually configure the port. Hit [F10] to save.

Chelsio T5/T4 Unified Wire For Linux Page 394

Chapter XVIII. Unified Boot xix. Go back and choose Configure Target Parameters to configure iSCSI target related properties. xx. If you want to discover target using DHCP, select Yes in the Discover Boot Target via

DHCP field. To discover target via static IP, select No and provide the target IP and Hit

[F10] to save. The default TCP port selected is 3260.

Chelsio T5/T4 Unified Wire For Linux Page 395

Chapter XVIII. Unified Boot xxi. Go back and choose Discover iSCSI Target (s) to connect to a target. xxii. Select the portal group on which iSCSI service is provided by the target

Chelsio T5/T4 Unified Wire For Linux Page 396

Chapter XVIII. Unified Boot xxiii. A list of available targets will be displayed. Select the target you wish to connect to and hit

[Enter]. xxiv. A list of LUNs configured on the selected target will be displayed. Select the LUN you

wish to connect to and hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 397

Chapter XVIII. Unified Boot xxv. Hit [Esc] and then [Y] to save the configuration. xxvi. Reboot the machine. xxvii. During POST, allow the Chelsio option ROM to discover iSCSI targets.

Chelsio T5/T4 Unified Wire For Linux Page 398

Chapter XVIII. Unified Boot xxviii. Enter BIOS setup and choose iSCSI target LUN discovered via Chelsio adapter as the

first boot device. xxi. Reboot and install the required OS onto the iSCSI target LUN.

Chelsio T5/T4 Unified Wire For Linux Page 399

Chapter XVIII. Unified Boot

6.2.

For uEFI iSCSI Boot

Important

Only uEFI 2.1 supported.

Any other uEFI version is NOT SUPPORTED and may render your system unusable.

This section describes the method to perform iSCSI boot on uEFI platforms. i. Configure the system having Chelsio CNA to boot in uEFI mode. ii. Boot to EFI Shell.

Chelsio T5/T4 Unified Wire For Linux Page 400

Chapter XVIII. Unified Boot iii. Issue command drivers to note down the handle# of Chelsio driver.

Chelsio T5/T4 Unified Wire For Linux Page 401

Chapter XVIII. Unified Boot iv. Then issue command drvcfg -s <handle#> to invoke the configuration utility.

Chelsio T5/T4 Unified Wire For Linux Page 402

Chapter XVIII. Unified Boot v. In the configuration utility, choose the CNA on which the option ROM image was flashed and Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 403

Chapter XVIII. Unified Boot vi. Choose config utility if you want to configure Chelsio adapter for PXE boot, else choose

flash utility if you want to flash/erase firmware, option ROM image or configuration parameters. vii. If config utility is chosen, go to step (viii), else goto step (xxii).

Chelsio T5/T4 Unified Wire For Linux Page 404

Chapter XVIII. Unified Boot viii. Enable the Adapter BIOS if not already enabled. Hit [Enter]. ix. Choose iSCSI from the list to configure and hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 405

Chapter XVIII. Unified Boot x. Choose the first option (Configure Function Parameters) from the list of parameter type and hit [Enter]. xi. Enable iSCSI BIOS if not already enabled. By default, CBFT (Chelsio Boot Firmware Table) will be used. Hit [F10] to save.

Chelsio T5/T4 Unified Wire For Linux Page 406

Chapter XVIII. Unified Boot xii. Choose the order of the ports to discover iSCSI targets. xiii. Set discovery timeout to a suitable value. Recommended value is >= 30.

Chelsio T5/T4 Unified Wire For Linux Page 407

Chapter XVIII. Unified Boot xiv. Hit [Esc] and then [Y] to save the configuration. xv. Go back and choose Configure Initiator Parameters to configure initiator related properties.

Chelsio T5/T4 Unified Wire For Linux Page 408

Chapter XVIII. Unified Boot xvi. Inititiar properties like IQN, Header Digest, Data Digest, etc will be displayed. Change the values appropriately or continue with the default values. Hit [F10] to save. xvii. CHAP authentication is disabled by default. To enable and configure, go back and choose

Configure CHAP Parameters

Chelsio T5/T4 Unified Wire For Linux Page 409

Chapter XVIII. Unified Boot xviii. Enable CHAP authentication by selecting ENABLED in the CHAP Policy field. Next, choose either one-way or mutual as the authentication method. Finally, provide Initiator and

Target CHAP credentials according to the authentication method selected. Hit [F10] to save. xix. Go back and choose Configure Network Parameters to configure iSCSI Network related properties.

Chelsio T5/T4 Unified Wire For Linux Page 410

Chapter XVIII. Unified Boot xx. Select the port using which you want to connect to the target. Hit [Enter]. xxi. Select Yes in the Enable DHCP field to configure port using DHCP or No to manually configure the port. Hit [F10] to save.

Chelsio T5/T4 Unified Wire For Linux Page 411

Chapter XVIII. Unified Boot xxii. Go back and choose Configure Target Parameters to configure iSCSI target related properties. xxiii. If you want to discover target using DHCP, select Yes in the Discover Boot Target via

DHCP field. To discover target via static IP, select No and provide the target IP and hit

[F10] to save. The default TCP port selected is 3260.

Chelsio T5/T4 Unified Wire For Linux Page 412

Chapter XVIII. Unified Boot xxiv. Go back and choose Discover iSCSI Target (s) to connect to a target. xxv. Select the portal group on which iSCSI service is provided by the target

Chelsio T5/T4 Unified Wire For Linux Page 413

Chapter XVIII. Unified Boot xxvi. A list of available targets will be displayed. Select the target you wish to connect to and hit [Enter]. xxvii. A list of LUNs configured on the selected target will be displayed. Select the LUN you

wish to connect to and hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 414

Chapter XVIII. Unified Boot xxviii. Hit [Esc] and then [Y] to save the configuration. xxix. Reboot the machine. xxx. During POST, allow the Chelsio option ROM to discover iSCSI targets.

Chelsio T5/T4 Unified Wire For Linux Page 415

Chapter XVIII. Unified Boot xxxi. Enter BIOS setup and choose iSCSI target LUN discovered via Chelsio adapter as the

first boot device. xxxii. Reboot the machine. xxxiii. Enter uEFI BIOS Manager, set DVD and discovered LUN as first and second boot

device respectively. Save changes and start the OS installation.

Chelsio T5/T4 Unified Wire For Linux Page 416

Chapter XVIII. Unified Boot xxxiv. If flash utility is chosen, choose any of available options and Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 417

Chapter XVIII. Unified Boot xxxv. If you have chosen to update option ROM image, then enter the full path to the file. E.g.:-

If the pen drive is device fs0 and the option ROM image file is in

fs0:\CHELSIO\cubt4.bin, then enter CHELSIO\cubt4.bin

.

Chelsio T5/T4 Unified Wire For Linux Page 418

Chapter XVIII. Unified Boot xxxvi. Restart the system for the changes to take place.

Chelsio T5/T4 Unified Wire For Linux Page 419

Chapter XVIII. Unified Boot

7. Creating Driver Update Disk (DUD)

7.1.

Creating DUD for RedHat Enterprise Linux

i.

If you haven’t done already, download ChelsioUwire-x.x.x.x.tar.gz from Chelsio Download

Center, service.chelsio.com

ii. Untar the package,

[root@host]# tar -zxvfm ChelsioUwire-x.x.x.x.tar.gz iii. Change your current working directory to LinuxDUD directory,

[root@host]# cd ChelsioUwire-x.x.x.x/Uboot/LinuxDUD iv. Insert a blank, formatted USB flash drive. v. Depending on the distribution to be installed, copy the corresponding image file to the USB drive. For example, execute the following command for RHEL 6.0:

[root@host]# cp Chelsio-DriverUpdateDisk-RHEL6.0-x86_64-x.x.x.x-y.img <path to USB drive>

7.2.

Creating DUD for Suse Enterprise Linux

i. If you haven’t done already, download ChelsioUwire-x.x.x.x.tar.gz from Chelsio Download

Center, service.chelsio.com

ii. Untar the package,

[root@host]# tar -zxvfm ChelsioUwire-x.x.x.x.tar.gz iii. Insert a blank USB flash drive.

Chelsio T5/T4 Unified Wire For Linux Page 420

Chapter XVIII. Unified Boot iv. Format the USB drive

[root@host]# mkfs.vfat /dev/sda v. Depending on the distribution to be installed, copy the corresponding image file to the USB stick. For example, execute the following command for SLES11sp1.

[root@host]# dd if=/root/ ChelsioUwire-x.x.x.x/Uboot/LinuxDUD/Chelsio-

DriverUpdateDisk-SLES11sp1-x86_64-x.x.x.x-y.img of=/dev/sda

Chelsio T5/T4 Unified Wire For Linux Page 421

Chapter XVIII. Unified Boot

8. OS Installation

8.1.

Installation using Chelsio NIC DUD (PXE only)

This is the recommended method for installing Linux OS using Chelsio PXE boot. The

Chelsio Driver Update Disk (DUD) has support for all the new adapters. Use Network Boot (PXE

Boot) media to install the OS, and provide the Driver Update Disk as per the detailed instructions for each OS.

The DUD supports installation of RHEL and SLES distributions using Chelsio adapters over Network. There may be built-in Chelsio driver in these distributions. The driver may or may not work with Chelsio adapters, depending on the adapter in use, and the version of the driver that shipped in that particular distribution. Please flash the firmware provided in the package.

Chelsio T5/T4 Unified Wire For Linux Page 422

Chapter XVIII. Unified Boot

8.1.1. RHEL 6.X installation using Chelsio DUD

i. PXE boot prompt

Please make sure that the USB drive with DUD image is inserted. Type dd at the boot prompt for the installation media. The dd option specifies that you will be providing a Driver Update Disk during the installation.

ii. Driver disk prompt:

The installer will load and prompt you for the driver update disk. Now select “Yes” and hit [Enter] to proceed.

Chelsio T5/T4 Unified Wire For Linux Page 423

Chapter XVIII. Unified Boot

iii. Driver disk source prompt:

You will be asked to select the Driver Update Disk device from a list. USB drives usually show up as SCSI disks in Linux. So if there are no other SCSI disks connected to the system, the

USB drive would assume the first drive letter “a”. Hence the drive name would be “sda”.

You can view the messages from the Linux kernel and drivers to determine the name of the

USB drive, by pressing [Alt] + [F3/F4] and [Alt] + [F1] to get back to the list. iv. Select the Driver Update File:

Select the Appropriate image file and Choose “OK”. Now the installer will search for the appropriate drivers from the driver disk and load them. This step may take some time. Check on the [Alt] + [F3] or [Alt] + [F4] screens for log messages.

Chelsio T5/T4 Unified Wire For Linux Page 424

Chapter XVIII. Unified Boot v. Load additional drivers prompt:

The installer will ask if you wish to load more drivers. Choose “Yes” to load if you have any other drivers to load. Otherwise choose “No”. vi. Choose language and Keyboard type:

Select the required language from the list.

Chelsio T5/T4 Unified Wire For Linux Page 425

Chapter XVIII. Unified Boot vii. Select Keyboard type

Select the type of keyboard you have from the list. viii. Select Installation method:

In this step, you can choose the source which contains the OS installation ISO image. In this case, select “NFS directory”.

Chelsio T5/T4 Unified Wire For Linux Page 426

Chapter XVIII. Unified Boot ix. Select Displayed Network Devices:

The Chelsio Network Devices will be displayed. Select the appropriate Chelsio NIC interface to proceed with installation. x. Configure TCP/IP settings:

Here you can specify if you want to configure your network interfaces using DHCP or manually using IPv4. IPv6 is currently not supported. Hence disable IPv6 before proceeding.

Chelsio T5/T4 Unified Wire For Linux Page 427

Chapter XVIII. Unified Boot xi. Provide NFS/FTP/HTTP Server Name/IP and Path:

Proceeding with the installation will get NFS/FTP/HTTP setup page. Here, provide NFS server details to proceed with the installation. Then the graphical Installation screens for RHEL will appear. Then proceed with the installation as usual.

8.1.2. SLES installation using Chelsio DUD

i. PXE boot prompt:

Please make sure that the USB drive with DUD image is inserted.

Type dd at the boot prompt for the installation media. The dd option specifies that you will be providing a Driver Update Disk during the installation. The DUD gets detected and the driver is automatically loaded by installer.

Chelsio T5/T4 Unified Wire For Linux Page 428

Chapter XVIII. Unified Boot ii. Start Installation

Select “Start Installation” and then “Start Installation or Update”.

Chelsio T5/T4 Unified Wire For Linux Page 429

Chapter XVIII. Unified Boot iii. Select method of install:

Select “Network” as the source of medium to install the SLES Operating System. iv. Select the Network protocol:

Select the desired Network protocol from the list presented.

Chelsio T5/T4 Unified Wire For Linux Page 430

Chapter XVIII. Unified Boot v. Select appropriate Chelsio network Interface:

Select the appropriate Chelsio interface from the list to proceed with installation. You can view the messages from the Linux kernel and drivers to determine the name of NIC interface by pressing [Alt] + [F3] or [Alt] + [F4]. Press [Alt] + [F1] to get back to the list. vi. Configure DHCP IP

Select “Yes” to configure the network interface selected in the previous step using DHCP.

Chelsio T5/T4 Unified Wire For Linux Page 431

Chapter XVIII. Unified Boot vii. Provide NFS/FTP/HTTP/TFTP Server Name/IP and Path:

Provide a valid NFS/FTP/HTTP/TFTP Server IP address to proceed. viii. Provide operating system Directory Path:

Provide a valid directory path to the operating system to be installed. When the graphical

Installation screen for SLES appears, proceed with the installation as usual.

8.2.

Installation on FCoE LUN

8.2.1. Using CD/DVD ROM

Please make sure that the USB drive with DUD image is inserted. Also, change the boot priority to boot from CD/DVD in the BIOS setup.

i. Insert the OS installation disc into your CD/DVD ROM.

ii. On the Grub menu, choose Install or upgrade an existing system option if not already selected.

iii. Press Tab and type dd at the boot prompt.

Chelsio T5/T4 Unified Wire For Linux Page 432

Chapter XVIII. Unified Boot iv. Driver disk prompt:

The installer will load a nd prompt you for the driver update disk. Now select “Yes” and hit

[Enter] to proceed.

Chelsio T5/T4 Unified Wire For Linux Page 433

Chapter XVIII. Unified Boot v. Driver disk source prompt:

You will be asked to select the Driver Update Disk device from a list. USB drives usually show up as SCSI disks in Linux. So if there are no other SCSI disks connected to the system, the

USB drive would assume the first drive letter “a”. Hence the drive name would be “sda”.

You can view the messages from the Linux kernel and drivers to determine the name of the

USB drive, by pressing [Alt] + [F3/F4] and [Alt] + [F1] to get back to the list. vi. Select the Driver Update File:

Select the Appropriate image file and Choose “OK”. Now the installer will search for the appropriate drivers from the driver disk and load them. This step may take some time. Check on the [Alt] + [F3] or [Alt] + [F4] screens for log messages.

Chelsio T5/T4 Unified Wire For Linux Page 434

Chapter XVIII. Unified Boot vii. Load additional drivers prompt:

The installer will ask if you wish to load more drivers. Choose “Yes” to load if you have any other drivers to load. Otherwise choose “No” and hit [Enter]. viii. Testing Installation media

If you want to test the installation media, choose OK or else Skip. Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 435

Chapter XVIII. Unified Boot

ix. Graphical Installer

Red Hat graphical installer will now start. Click Next.

Chelsio T5/T4 Unified Wire For Linux Page 436

Chapter XVIII. Unified Boot x. Select storage device

Select the FC/FCoE LUN which was saved as boot device in system BIOS and click Next.

Then proceed with the installation as usual.

8.3.

Installation on iSCSI LUN

8.3.1. Using CD/DVD ROM

Please make sure that the USB drive with DUD image is inserted. Also, change the boot priority to boot from CD/DVD in the BIOS setup.

i. Insert the OS installation disc into your CD/DVD ROM.

ii. On the Grub menu, choose Install or upgrade an existing system option if not already selected.

iii. Press Tab and type dd at the boot prompt.

Chelsio T5/T4 Unified Wire For Linux Page 437

Chapter XVIII. Unified Boot iv. Driver disk prompt:

The installer will load and prompt you for the driver update disk. Now select “Yes” and hit

[Enter] to proceed.

Chelsio T5/T4 Unified Wire For Linux Page 438

Chapter XVIII. Unified Boot v. Driver disk source prompt:

You will be asked to select the Driver Update Disk device from a list. USB drives usually show up as SCSI disks in Linux. So if there are no other SCSI disks connected to the system, the

USB drive would assume the first drive letter “a”. Hence the drive name would be “sda”.

You can view the messages from the Linux kernel and drivers to determine the name of the

USB drive, by pressing [Alt] + [F3/F4] and [Alt] + [F1] to get back to the list. vi. Select the Driver Update File:

Select the Appropriate image file and Choose “OK”. Now the installer will search for the appropriate drivers from the driver disk and load them. This step may take some time. Check on the [Alt] + [F3] or [Alt] + [F4] screens for log messages.

Chelsio T5/T4 Unified Wire For Linux Page 439

Chapter XVIII. Unified Boot vii. Load additional drivers prompt:

The installer will ask if you wish to load more drivers. Choose “Yes” to load if you have any other d rivers to load. Otherwise choose “No” and hit [Enter]. xi. Testing Installation media

If you want to test the installation media, choose OK or else Skip. Hit [Enter].

Chelsio T5/T4 Unified Wire For Linux Page 440

Chapter XVIII. Unified Boot

xii. Graphical Installer

Red Hat graphical installer will now start. Click Next.

Chelsio T5/T4 Unified Wire For Linux Page 441

Chapter XVIII. Unified Boot

xiii. Select Specialized Storage Devices radio button and click Next.

xiv. The discovered LUNs will appear in the Basic Devices tab and you can proceed with installation as usual. If not, follow the steps mentioned below to add target.

Chelsio T5/T4 Unified Wire For Linux Page 442

Chapter XVIII. Unified Boot

xv. Click +Add Advanced Target button and select the Add iSCSI target radio button. Click

+Add drive button.

xvi. Enter the target IP. Initiator IQN will be autogenerated by the installer. Click Start

Discovery.

Chelsio T5/T4 Unified Wire For Linux Page 443

Chapter XVIII. Unified Boot

xvii. On successful discovery, LUNs configured on the discovered target will be displaye

xviii. Select the checkbox corresponding to the LUN on which you wish to install operating system and click Login.

Note

Make sure the same LUN discovered at the Option ROM stage is selected for

OS installation.

Chelsio T5/T4 Unified Wire For Linux Page 444

Chapter XVIII. Unified Boot

xix. Next, choose the CHAP authentication method and click Login.

xx. On successful login, a dialog box will display the results along with the connected LUN name. Click OK.

Chelsio T5/T4 Unified Wire For Linux Page 445

Chapter XVIII. Unified Boot

xxi. Select the LUN under the Other SAN Devices tab, and click Next.

xxii. Proceed with the installation as usual.

Chelsio T5/T4 Unified Wire For Linux Page 446

Chapter XIX. Appendix

XIX. Appendix A

Chelsio T5/T4 Unified Wire For Linux Page 447

Chapter XIX. Appendix

1. Troubleshooting

Cannot bring up Chelsio interface

Make sure you have created the corresponding network-script configuration file as stated in

Section 5.2 of Cheslsio Unified Wire chapter ( Click here ). If the file does exist, make sure the

structure and contents are correct. A sample is given in Section 5.1 of Chelsio Unified Wire

chapter ( Click here ). Another reason may be that the IP address mentioned in the configuration

file is already in use on the network.

Cannot ping through Chelsio interface

First, make sure the interface was successfully brought up using ifup ethX (where ethX is your interface) and that it is linked to an IP address, either static or obtained through DHCP.

You then may want to check whether the destination host (i.e. the machine you are trying to ping) is up and running and accepts ICMP requests such as ping. If you get a return value of 0 when doing a cat /proc/sys/net/ipv4/icmp_echo_ignore_all on the remote host that means it is configured to reply to incoming pings. Change ipv4 to ipv6 in the path if you are using IPv6. Note that this is a Linux-only tip.

If you have more than one interface wired to the network, make sure you are using the right one for your outgoing ping requests. This can be done by using the -I option of the ping command, as shown in the following example:

[root@host]# ping -I eth1 10.192.167.1

Where 10.192.167.1 is the machine you want to ping.

Configuring firewall for your application

In many cases the firewall software on the systems may prevent the applications from working properly. Please refer to the appropriate documentation for the Linux distribution on how to configure or disable the firewall.

FCoE link not up

Always enable LLDP on the interfaces as FCoE link won’t come up until and unless a successful LLDP negotiation happens.

Chelsio T5/T4 Unified Wire For Linux Page 448

Chapter XIX. Appendix

priority-flow-control mode on the switch

On the switch, make sure priority-flow-control mode is always set to auto and flow control is disabled.

Configuring Ethernet interfaces on Cisco switch

Always configure Ethernet interfaces on Cisco switch in trunk mode.

Binding VFC to MAC

If you are binding the VFC to MAC address in case of Cisco Nexus switch, then make sure you make the Ethernet interface part of both Ethernet VLAN and FCoE VLAN.

Cisco nexus switch reporting “pauseRateLimitErrDisable”

If in any case the switch-port on the Cisco nexus switch is reporting

“pauseRateLimitErrDisable”, then perform an Ethernet port shut/no shut.

Chelsio T5/T4 Unified Wire For Linux Page 449

Chapter XIX. Appendix

2. Chelsio End-User License Agreement (EULA)

Installation and use of the driver/software implies acceptance of the terms in the Chelsio End-

User License Agreement (EULA).

IMPORTANT: PLEASE READ THIS SOFTWARE LICENSE CAREFULLY BEFORE DOWNLOADING OR OTHERWISE

USING THE SOFTWARE OR ANY ASSOCIATED DOCUMENTATION OR OTHER MATERIALS (COLLECTIVELY,

THE "SOFTWARE"). BY CLICKING ON THE "OK" OR "ACCEPT" BUTTON YOU AGREE TO BE BOUND BY

THE TERMS OF THIS AGREEMENT. IF YOU DO NOT AGREE TO THE TERMS OF THIS AGREEMENT,

CLICK THE "DO NOT ACCEPT" BUTTON TO TERMINATE THE INSTALLATION PROCESS.

1. License. Chelsio Communications, Inc. ("Chelsio") hereby grants you, the Licensee, and you hereby accept, a limited, non-exclusive, nontransferable license to:

(i) install and use the Software on a single computer system or on multiple workstations, systems and servers that incorporate a Chelsio network adapter and may be accessed by multiple users from multiple locations; and (ii) make one copy of the Software in machine readable form solely for back-up purposes, provided you reproduce Chelsio's copyright notice and any proprietary legends, as required by

Chelsio.

2. Restrictions. This license granted hereunder does not constitute a sale of the

Software or any copy thereof. Except as expressly permitted under this Agreement, you may not:

(i) reproduce, modify, adapt, translate, rent, lease, loan, resell for profit, distribute, or create derivative works of or based upon, the Software or any part thereof; or

(ii) make available the Software, or any portion thereof, in any form, on the

Internet. The Software contains trade secrets and, in order to protect them, you may not decompile, reverse engineer, disassemble, or otherwise reduce the

Software to a human-perceivable form. You assume full responsibility for the use of the Software and agree to use the Software legally and responsibly.

3. Ownership of Software. As Licensee, you own only the media upon which the Software is recorded or fixed, but Chelsio retains all right, title and interest in and to the

Software recorded on the original media and all subsequent copies of the Software, regardless of the form or media in or on which the Software may be embedded.

4. Confidentiality. You agree to maintain the Software in confidence and not to disclose the Software, or any information or materials related thereto, to any third party without the express written consent of Chelsio. You further agree to take all reasonable precautions to limit access of the Software only to those of your employees who reasonably require such access to perform their employment obligations and who are bound by confidentiality agreements with you.

5. Term. This license is effective in perpetuity, unless terminated earlier. You may terminate the license at any time by destroying the Software (including the related documentation), together with all copies or modifications in any form. Chelsio may terminate this license, and this license shall be deemed to have automatically terminated, if you fail to comply with any term or condition of this Agreement. Upon any termination, including termination by you, you must destroy the Software

Chelsio T5/T4 Unified Wire For Linux Page 450

Chapter XIX. Appendix

(including the related documentation), together with all copies or modifications in any form.

6. Limited Warranty. Chelsio warrants only that the media upon which the Software is furnished will be free from defects in material or workmanship under normal use and service for a period of thirty (30) days from the date of delivery to you. CHELSIO

DOES NOT AND CANNOT WARRANT THE PERFORMANCE OR RESULTS YOU MAY OBTAIN BY USING THE

SOFTWARE OR ANY PART THEREOF. EXCEPT FOR THE FOREGOING LIMITED WARRANTY, CHELSIO

MAKES NO OTHER WARRANTIES, EXPRESS OR IMPLIED, AND HEREBY DISCLAIMS ALL OTHER

WARRANTIES, INCLUDING, BUT NOT LIMITED, TO NON-INFRINGEMENT OF THIRD PARTY RIGHTS,

MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow the exclusion of implied warranties or limitations on how long an implied warranty may last, so the above limitations may not apply to you. This warranty gives you specific legal rights and you may also have other rights which vary from state to state.

7. Remedy for Breach of Warranty. The sole and exclusive liability of Chelsio and its distributors, and your sole and exclusive remedy, for a breach of the above warranty, shall be the replacement of any media not meeting the above limited warranty which is returned to Chelsio. If Chelsio or its distributor is unable to deliver replacement media which is free from defects in materials or workmanship, you may terminate this

Agreement by returning the Software.

8. Limitation of Liability. IN NO EVENT SHALL CHELSIO HAVE ANY LIABILITY TO YOU OR

ANY THIRD PARTY FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR PUNITIVE

DAMAGES, HOWEVER CAUSED, AND ON ANY THEORY OF LIABILITY, ARISING OUT OF OR RELATED TO

THE LICENSE OR USE OF THE SOFTWARE, INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR LOSS

OF ANTICIPATED PROFITS, EVEN IF CHELSIO HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH

DAMAGES. IN NO EVENT SHALL CHELSIO'S LIABILITY ARISING OUT OF OR RELATED TO THE

LICENSE OR USE OF THE SOFTWARE EXCEED THE AMOUNTS PAID BY YOU FOR THE LICENSE GRANTED

HEREUNDER. THESE LIMITATIONS SHALL APPLY NOTWITHSTANDING ANY FAILURE OF ESSENTIAL

PURPOSE OF ANY LIMITED REMEDY.

9. High Risk Activities. The Software is not fault-tolerant and is not designed, manufactured or intended for use or resale as online equipment control equipment in hazardous environments requiring fail-safe performance, such as in the operation of nuclear facilities, aircraft navigation or communication systems, air traffic control, direct life support machines, or weapons systems, in which the failure of the Software could lead directly to death, personal injury, or severe physical or environmental damage. Chelsio specifically disclaims any express or implied warranty of fitness for any high risk uses listed above.

10. Export. You acknowledge that the Software is of U.S. origin and subject to U.S. export jurisdiction. You acknowledge that the laws and regulations of the United

States and other countries may restrict the export and re-export of the Software. You agree that you will not export or re-export the Software or documentation in any form in violation of applicable United States and foreign law. You agree to comply with all applicable international and national laws that apply to the Software, including the U.S. Export Administration Regulations, as well as end-user, end-use, and destination restrictions issued by U.S. and other governments.

11. Government Restricted Rights. The Software is subject to restricted rights as follows. If the Software is acquired under the terms of a GSA contract: use, reproduction or disclosure is subject to the restrictions set forth in the applicable

ADP Schedule contract. If the Software is acquired under the terms of a DoD or civilian agency contract, use, duplication or disclosure by the Government is subject to the restrictions of this Agreement in accordance with 48 C.F.R. 12.212 of the

Chelsio T5/T4 Unified Wire For Linux Page 451

Chapter XIX. Appendix

Federal Acquisition Regulations and its successors and 49 C.F.R. 227.7202-1 of the DoD

FAR Supplement and its successors.

12. General. You acknowledge that you have read this Agreement, understand it, and that by using the Software you agree to be bound by its terms and conditions. You further agree that it is the complete and exclusive statement of the agreement between

Chelsio and you, and supersedes any proposal or prior agreement, oral or written, and any other communication between Chelsio and you relating to the subject matter of this

Agreement. No additional or any different terms will be enforceable against Chelsio unless Chelsio gives its express consent, including an express waiver of the terms of this Agreement, in writing signed by an officer of Chelsio. This Agreement shall be governed by California law, except as to copyright matters, which are covered by

Federal law. You hereby irrevocably submit to the personal jurisdiction of, and irrevocably waive objection to the laying of venue (including a waiver of any argument of forum non conveniens or other principles of like effect) in, the state and federal courts located in Santa Clara County, California, for the purposes of any litigation undertaken in connection with this Agreement. Should any provision of this Agreement be declared unenforceable in any jurisdiction, then such provision shall be deemed severable from this Agreement and shall not affect the remainder hereof. All rights in the Software not specifically granted in this Agreement are reserved by Chelsio.

Chelsio reserves the right to modify this license agreement at any time without notice, and any modified version of this agreement shall supercede any earlier versions.

Should you have any questions concerning this Agreement, you may contact Chelsio by writing to:

Chelsio Communications, Inc.

370 San Aleso Ave.

Sunnyvale, CA 94085

Chelsio T5/T4 Unified Wire For Linux Page 452

advertisement

Was this manual useful for you? Yes No
Thank you for your participation!

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

Related manuals

Download PDF

advertisement