HP IRF Configuration Guide

7/26/2019 HP IRF Configuration Guide

1/64

HP 5920 & 5900 Switch Series

IRF

Configuration Guide

Part number: 5998-5305

Software version: Release 2307

Document version: 6W100-20140108


2/64

Legal and notice information

Copyright 2014 Hewlett-Packard Development Company, L.P.

No part of this documentation may be reproduced or transmitted in any form or by any means withoutprior written consent of Hewlett-Packard Development Company, L.P.

The information contained herein is subject to change without notice.HEWLETT-PACKARD COMPANY MAKES NO WARRANTY OF ANY KIND WITH REGARD TO THISMATERIAL, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY

AND FITNESS FOR A PARTICULAR PURPOSE. Hewlett-Packard shall not be liable for errors containedherein or for incidental or consequential damages in connection with the furnishing, performance, oruse of this material.

The only warranties for HP products and services are set forth in the express warranty statementsaccompanying such products and services. Nothing herein should be construed as constituting anadditional warranty. HP shall not be liable for technical or editorial errors or omissions containedherein.


3/64

i

Contents

IRF overview 1

Hardware compatibility 1

IRF benefits 1

Application scenario 1Basic concepts 2

IRF member roles 2

IRF member ID 2IRF port 2Physical IRF port 3

IRF domain ID 3

IRF split 4IRF merge 4

Member priority 4

Interface naming conventions 4

File system naming conventions 5Configuration synchronization mechanism 6

Master election 6

IRF multi-active detection 7Multi-active handling procedure 7

LACP MAD 8BFD MAD 9

ARP MAD 10ND MAD 11

Configuring IRF13

General restrictions and configuration guidelines 13Software requirements 13

IRF physical port restrictions and binding requirements 13IRF link redundancy 14Multichassis link aggregation 14

MAD and IRF domain restrictions 14Other configuration guidelines 15

Setup and configuration task list 15Planning the IRF fabric setup 16

Assigning a member ID to each IRF member device 17Specifying a priority for each member device 17Connecting physical IRF ports 18

Binding physical ports to IRF ports 19Accessing the IRF fabric 21

Configuring a member device description 21Configuring IRF link load sharing mode 21

Configuring the global load sharing mode 22Configuring a port-specific load sharing mode 22

Configuring IRF bridge MAC persistence 23

Enabling software auto-update for software image synchronization 24

Configuration prerequisites 24Configuration procedure 24

Setting the IRF link down report delay 25

Configuring MAD 25


4/64

ii

Configuring LACP MAD 26Configuring BFD MAD 28

Configuring ARP MAD 29Configuring ND MAD 30

Excluding a port from the shutdown action upon detection of multi-active collision 31Recovering an IRF fabric 32

Displaying and maintaining an IRF fabric 33

Configuration examples 34

LACP MAD-enabled IRF configuration example 34BFD MAD-enabled IRF configuration example 38ARP MAD-enabled IRF configuration example 43

ND MAD-enabled IRF configuration example 48

Support and other resources 53Contacting HP 53

Subscription service 53Related information 53

Documents 53

Websites 53

Conventions 54

Index 56


5/64

1

IRF overview

HP Intelligent Resilient Framework (IRF) technology creates a large IRF fabric from multiple devices toprovide data center class availability and scalability. IRF virtualization technology offers processingpower, interaction, unified management, and uninterrupted maintenance of multiple devices.

This book describes IRF concepts and guides you through the IRF setup procedure.

Hardware compatibilityYou can establish an IRF fabric by using HP 5900 switches, HP 5920 switches, or switches from bothseries.

IRF benefitsIRF provides the following benefits:

Simplified topology and easy managementAn IRF fabric appears as one node and is accessibleat a single IP address on the network. You can use this IP address to log in at any member deviceto manage all the members of the IRF fabric. In addition, you do not need to run the spanning treefeature among the IRF members.

1:N redundancyIn an IRF fabric, one member works as the master to manage and control theentire IRF fabric, and all the other members process services while backing up the master. When themaster fails, all the other member devices elect a new master from among them to take over without

interrupting services.IRF link aggregationYou can assign several physical links between neighboring members to theirIRF ports to create a load-balanced aggregate IRF connection with redundancy.

Multichassis link aggregationYou can use the Ethernet link aggregation feature to aggregate thephysical links between the IRF fabric and its upstream or downstream devices across the IRFmembers.

Network scalability and resiliencyProcessing capacity of an IRF fabric equals the totalprocessing capacities of all the members. You can increase ports, network bandwidth, andprocessing capacity of an IRF fabric simply by adding member devices without changing thenetwork topology.

Application scenarioFigure 1shows an IRF fabric that has two devices, which appear as a single node to the upper and lowerlayer devices.


6/64

2

Figure 1IRF application scenario

Basic conceptsThis section describes the basic concepts you might encounter when you work with IRF.

IRF member roles

IRF uses two member roles: master and standby (called "subordinate" throughout the documentation).

When devices form an IRF fabric, they elect a master to manage and control the IRF fabric, and all theother devices back up the master. When the master device fails, the other devices elect a new masterautomatically. For more information about master election, see "Master election."

IRF member ID

An IRF fabric uses member IDs to uniquely identify and manage its members. This member ID informationis included as the first part of interface numbers and file paths to uniquely identify interfaces and files inan IRF fabric. For more information about interface and file path naming, see " Interface namingconventions" and "File system naming conventions."

If two devices have the same IRF member ID, they cannot form an IRF fabric. If the IRF member ID of adevice has been used in an IRF fabric, the device cannot join the fabric.

IRF port

An IRF port is a logical interface for the connection between IRF member devices. Every IRF-capabledevice supports two IRF ports. The IRF ports are named IRF-port n/1 and IRF-port n/2, where nis themember ID of the switch. The two IRF ports are referred to as "IRF-port 1" and "IRF-port 2" in this bookfor simplicity.


7/64

3

To use an IRF port, you must bind at least one physical port to it. The physical ports assigned to an IRFport form an aggregate IRF link automatically. An IRF port goes down only if all its physical IRF ports aredown.

Physical IRF port

Physical IRF ports connect IRF member devices and must be bound to an IRF port. They forward the IRFprotocol packets between IRF member devices and the data packets that must travel across IRF memberdevices.

For more information about physical ports that can be used for IRF links, see "IRF physical port restrictionsand binding requirements."

IRF domain ID

One IRF fabric forms one IRF domain. IRF uses IRF domain IDs to uniquely identify IRF fabrics and preventIRF fabrics from interfering with one another.

As shown in Figure 2, Device A and Device B form IRF fabric 1, and Device C and Device D form IRFfabric 2. Both fabrics use the LACP aggregate links between them for MAD. When a member device inone IRF fabric receives an extended LACP packet for MAD, it looks at the domain ID in the packet to seewhether the packet is from the local IRF fabric or from a different IRF fabric. Then, the device can handlethe packet correctly.

Figure 2A network that comprises two IRF domains

Device A Device BIRF 1 (domain 10)

IRF link

Core network

IRF 2 (domain 20)

IRF link

Device C Device D

Access network


8/64

4

IRF split

IRF split occurs when an IRF fabric breaks up into two or more IRF fabrics because of IRF link failures, asshown in Figure 3. The split IRF fabrics operate with the same IP address and cause routing andforwarding problems on the network. To quickly detect a multi-active collision, configure at least one

MAD mechanisms (see "IRF multi-active detection").Figure 3IRF split

IRF mergeIRF merge occurs when two split IRF fabrics reunite or when two independent IRF fabrics are united, asshown in Figure 4.

Figure 4IRF merge

Member priority

Member priority determines the possibility of a member device to be elected the master. A member withhigher priority is more likely to be elected the master.

The default member priority is 1. You can change the member priority of a device to affect the masterelection result.

Interface naming conventionsAn interface is named in the chassis-number/slot-number/port-indexformat.

chassis-numberRepresents the IRF member ID of the switch. This argument defaults to 1. The IRFmember ID always takes effect regardless of whether or not the switch is part of an IRF fabric.

slot-numberRepresents the slot number of the front panel. This argument is fixed at 0 on theswitch.

port-indexPort index depends on the number of ports available on the device. To identify theindex of a port, look at its port index mark on the chassis.

IRF link

Device A Device BDevice A Device B

IRF 1 IRF 2 IRF

+ =


9/64

5

For one example, on the standalone switch Sysname, Ten-GigabitEthernet 1/0/1 represents the first porton the device. Set its link type to trunk, as follows:

syst em- vi ew

[ Sysname] i nt er f ace t en- gi gabi t ether net 1/ 0/ 1

[ Sysname- Ten- Gi gabi t Et hernet 1/ 0/ 1] por t l i nk- t ype t r unk

For another example, on the IRF fabric Master, Ten-GigabitEthernet 3/0/1 represents the first fixed porton member device 3. Set its link type to trunk, as follows:

syst em- vi ew

[ Mast er ] i nt er f ace t en- gi gabi t et her net 3/ 0/ 1

[ Mast er - Ten- Gi gabi t Et her net 3/ 0/ 1] por t l i nk-t ype tr unk

File system naming conventionsOn a standalone device, you can use its storage device name to access its file system. For moreinformation about storage device naming conventions, see Fundamentals Configuration Guide.

On an IRF fabric, you can use the storage device name to access the file system of the master. To accessthe file system of any other member device, use the name in the slotmember-ID#storage-device-nameformat. For example:

To access the testfolder under the root directory of the flash memory on the master switch:

mkdi r t est

Creati ng di rect ory f l ash: / t est. . . Done.

di r

Di rect ory of f l ash:

0 - r w- 43548660 J an 01 2011 08: 21: 29 syst em. i pe

1 drw- - J an 01 2011 00: 00: 30 di agf i l e

2 - r w- 567 J an 02 2011 01: 41: 54 dsakey3 - r w- 735 J an 02 2011 01: 42: 03 host key

4 - r w- 36 J an 01 2011 00: 07: 52 i f i ndex. dat

5 - r w- 0 J an 01 2011 00: 53: 09 l aut h. dat

6 drw- - J an 01 2011 06: 33: 55 l auu

7 drw- - J an 02 2000 00: 00: 07 l ogf i l e

8 - r w- 23724032 J an 01 2011 00: 49: 47 sys t em- cmw710. bi n

9 drw- - J an 01 2000 00: 00: 07 secl og

10 - r w- 591 J an 02 2011 01: 42: 03 ser verkey

11 - r w- 4609 J an 01 2011 00: 07: 53 st art up. cf g

12 - r w- 3626 J an 01 2011 01: 51: 56 st art up. cf g_bak

13 - r w- 78833 J an 01 2011 00: 07: 53 st ar t up. mdb

14 drw- - J an 01 2011 00: 15: 48 t est

25 drw- - J an 01 2011 04: 16: 53 ver si onI nf o

524288 KB t ot al ( 365292 KB f r ee)

To create and access the testfolder under the root directory of the flash memory on member device 3:

mkdi r s l ot3#f l ash: / t est

Creat i ng di r ectory sl ot 3#f l ash: / t est . . . Done.

cd sl ot3#f l ash: / t est

pwd


10/64

6

sl ot 3#f l ash: / test

Or:

cd sl ot3#f l ash: /

mkdi r t est

Creat i ng di r ectory sl ot 3#f l ash: / t est . . . Done.

To copy the file test.ipeon the master to the root directory of the flash memory on member device 3:

# Display the current working path. In this example, the current working path is the root directory of theflash on member device 3.

pwd

sl ot 3#f l ash:

# Change the current working path to the root directory of the flash memory on the master device.

cd f l ash: /

pwd

f l ash:

# Copy the file to member device 3.

copy t est. i pe sl ot3#f l ash: /

Copy f l ash: / t est. i pe to sl ot3#f l ash: / t est. i pe?[Y/N] : y

Copyi ng f i l e f l ash: / t est. i pe t o sl ot 3#f l ash: / t est. i pe. . . Done.

Configuration synchronization mechanismIRF uses a strict running-configuration synchronization mechanism so all chassis in an IRF fabric can workas a single node, and after the master fails, other members can operate correctly.

In an IRF fabric, all chassis get and run the running configuration of the master. Any configuration you

have made is propagated to all members.For more information about configuration management, see Fundamentals Configuration Guide.

Master electionMaster election is held each time the IRF fabric topology changes, for example, when the IRF fabric isestablished, the master device fails or is removed, the IRF fabric splits, or IRF fabrics merge. Masterelection does not occur when two split IRF fabrics merge.

Master election uses the following rules in descending order:

1. Current master, even if a new member has higher priority.When an IRF fabric is being formed, all members consider themselves as the master, and this ruleis skipped.

2. Member with higher priority. If all members have the same priority, this rule is skipped.

3. Member with the longest system uptime.

Two members are considered starting up at the same time if the difference between their startuptimes is equal to or less than 10 minutes. For these members, the next tiebreaker applies.

4. Member with the lowest CPU MAC address.

The IRF fabric is formed on election of the master.


11/64

7

During an IRF merge, the members of the IRF fabric that fails the master election reboot automatically torejoin the IRF fabric that wins the election.

After a master election, all subordinate members reboot with the configuration on the master. Theconfiguration files of the subordinate members are still retained, but these files do not take effect in theIRF fabric.

IRF multi-active detectionAn IRF link failure causes an IRF fabric to split in two IRF fabrics operating with the same Layer 3configurations, including the same IP address. To avoid IP address collision and network problems, IRFuses multi-active detection (MAD) mechanisms to detect the presence of multiple identical IRF fabrics,handle collisions, and recover from faults.

Multi-active handling procedure

The multi-active handling procedure includes detection, collision handling, and failure recovery.Detection

The device's MAD implementation detects active IRF fabrics with the same Layer 3 global configurationby extending the LACP, BFD, ARP, or IPv6 ND protocol.

These MAD mechanisms identify each IRF fabric with a domain ID and an active ID (the member ID ofthe master). If multiple active IDs are detected in a domain, MAD determines that an IRF collision or splithas occurred.

You can use at least one of these mechanisms in an IRF fabric, depending on your network topology.

IMPORTANT:

LACP MAD handles collisions in a different way than BFD MAD, ARP MAD, and ND MAD. To avoidconflicts, do not enable LACP MAD together with any of those mechanisms in an IRF fabric. However, youcan use BFD MAD, ARP MAD, and ND MAD together.

For a comparison of these MAD mechanisms, see "Configuring MAD."

Collision handling

MAD mechanisms remove multi-active collisions by setting one IRF fabric to the Detect state and other IRFfabrics to the Recovery state. The Detect-state IRF fabric is active. The Recovery-state IRF fabrics areinactive. Only members in the Detect-state fabric can continue to forward traffic.

LACP MAD handles a multi-active collision in the following procedure:1. Compares the number of members in each fabric.

2. Sets the fabric that has the most members to the Detect state and all other fabrics to the Recoverystate.

3. If all IRF fabrics have the same number of members, compares the member IDs of their masters.

4. Sets the IRF fabric that has the lowest numbered master to the Detect state and all other fabrics tothe Recovery (disabled) state.

5. Shuts down all physical network ports in the Recovery-state fabrics except their physical IRF portsand any ports you have specified with the mad exclude interface command.


12/64

8

In contrast, BFD MAD, ARP MAD, and ND MAD do not compare the number of members in fabrics. Theydirectly set the IRF fabric that has the lowest numbered master to the Detect state, set all other fabrics tothe Recovery state, and take the same action on the network ports in Recovery-state fabrics as LACP MADdoes.

Failure recovery

To merge two split IRF fabrics, first repair the failed IRF link and remove the IRF link failure.

If the IRF fabric in Recovery state fails before the failure is recovered, repair the failed IRF fabric and thefailed IRF link.

If the active IRF fabric fails before the failure is recovered, first enable the inactive IRF fabric to take overthe active IRF fabric and protect services from being affected. After that, recover the MAD failure.

LACP MAD

LACP MAD requires that every IRF member have a link with an intermediate device, and all these linksform a dynamic link aggregation group, as shown inFigure 5. The intermediate device must be a device

that supports extended LACP for MAD.

The IRF member devices send extended LACPDUs with TLVs that convey the domain ID and the active IDof the IRF fabric. The intermediate device transparently forwards the extended LACPDUs received fromone member device to all the other member switches.

If the domain IDs and the active IDs in the extended LACPDUs sent by all the member devices arethe same, the IRF fabric is integrated.

If the extended LACPDUs convey the same domain ID but different active IDs, a split has occurred.LACP MAD handles this situation as described in "Collision handling."


13/64

9

Figure 5LACP MAD application scenario

BFD MADBFD MAD can work with or without intermediate devices. Figure 6shows a typical BFD MAD applicationscenario.

To use BFD MAD:

Set up dedicated BFD MAD link between each pair of IRF members or between each IRF memberand the intermediate device. Do not use the BFD MAD links for any other purpose.

Assign the ports connected by BFD MAD links to the same VLAN, create a VLAN interface for theVLAN, and assign a MAD IP address to each member on the VLAN interface.

The MAD addresses identify the member devices and must belong to the same subnet.With BFD MAD, the master tries to establish BFD sessions with other member devices by using its MADIP address as the source IP address:

If the IRF fabric is integrated, only the MAD IP address of the master is effective, and the mastercannot establish a BFD session with any other member. If you execute the display bfd sessioncommand, the state of the BFD sessions is Down.

When the IRF fabric splits, the IP addresses of the masters in the split IRF fabrics take effect, and thetwo masters can establish a BFD session. If you execute the display bfd sessioncommand, the stateof the BFD session between the two devices is Up.

Device

Master Subordinate

IRF

Internet

Customer

premise

network

IRF link

Common traffic path

LACP MAD traffic path

LACP-enabled dynamic

link aggregation

LACP-enabled dynamic

link aggregation


14/64

10

Figure 6BFD MAD application scenario

ARP MAD

ARP MAD detects multi-active collisions by using extended ARP packets that convey the IRF domain ID

and the active ID.

You can set up ARP MAD links between neighbor IRF member devices, or between each IRF memberdevice and an intermediate device (see Figure 7). If an intermediate device is used, you must also run thespanning tree feature between the IRF fabric and the intermediate device.

Device

Master Subordinate

IRF

Internet

Customer

premise

network

IRF link

Link aggregation

BFD MAD link

VLAN 2

192.168.1.2/24VLAN 2

192.168.1.3/24


15/64

11

Figure 7ARP MAD application scenario

Each IRF member compares the domain ID and the active ID in incoming extended ARP packets with its

domain ID and active ID:

If the domain IDs are different, the extended ARP packet is from a different IRF fabric. The devicedoes not continue to process the packet with the MAD mechanism.

If the domain IDs are the same, the device compares the active IDs:

If the active IDs are different, the IRF fabric has split.

If the active IDs are the same, the IRF fabric is integrated.

ND MAD

ND MAD detects multi-active collisions by using the ND protocol's NS packets to transmit the IRF domainID and the active ID.

You can set up ND MAD links between neighbor IRF member devices or between each IRF memberdevice and an intermediate device (see Figure 8). If an intermediate device is used, you must also run thespanning tree protocol between the IRF fabric and the intermediate device.

Device

Master Subordinate

IRF

Internet

Customer

premise

network

IRF link

Common traffic path

Extended ARP traffic path

STP domain (all devic es

must run the spanning

tree feature)


16/64

12

Figure 8ND MAD application scenario

Each IRF member device compares the domain ID and the active ID in incoming NS packets with its

domain ID and active ID:

If the domain IDs are different, the NS packet is from a different IRF fabric. The device does notcontinue to process the packet with the MAD mechanism.

If the domain IDs are the same, the device compares the active IDs:

If the active IDs are different, the IRF fabric has split.

If the active IDs are the same, the IRF fabric is integrated.

Device

Master Subordinate

IRF

Internet

Customer

premise

network

IRF link

Common traffic path

Extended ND traffic path

STP domain (all devic es

must run the spanning

tree feature)


17/64

13

Configuring IRF

To ensure a successful IRF setup, read the configuration restrictions and guidelines carefully before youconnect and set up an IRF fabric.

General restrictions and configuration guidelines

Software requirements

All IRF member devices must run the same software image version. Make sure the software auto-updatefunction is enabled on all member devices.

IRF physical port restrictions and binding requirementsThe following are physical ports that can be used for IRF connection:

HP 5900 switchAll 10-GE, SFP+, or QSFP+ ports.

HP 5920 switchAll SFP+ ports.

Selecting transceiver modules and cables

When you select transceiver modules and cables, follow these guidelines:

Use 10-GE twisted pairs to connect 10-GE ports in a short distance.

Use SFP+/QSFP+ cables to connect SFP+/QSFP+ ports in a short distance.

Use SFP+/QSFP+ transceiver modules and fibers to connect SFP+/QSFP+ ports in a long distance.

10-GE IRF physical ports and QSFP+ transceiver modules are available only for the HP 5900switches. For more information about transceiver modules, see the switch installation guide.

The transceiver modules at the two ends of an IRF link must be the same type.

10-GE port restrictions for the HP 5900AF-48XGT-4QSFP+ Switch switch

When you use the 10-GE ports on the HP 5900AF-48XGT-4QSFP+ Switch (JG336A) as IRF physicalports, follow these guidelines:

The 10-GE ports are grouped by port number in order, starting from 1. Each group contains fourports. If you use one port in a group for IRF connection, you must also use all the other ports in thegroup for IRF connection. However, you can bind them to different IRF ports.

Before you bind a 10-GE port to an IRF port or remove it from the IRF port, you must shut down allthe 10-GE ports in the same group.

Bring up the ports after you complete the operation.

QSFP+ port restrictions for the HP 5900 switch

You can use a QSFP+ port as an IRF physical port, or use the using tengigecommand to split a QSFP+port into four 10-GE interfaces.

The following restrictions apply to these 10-GE interfaces:


18/64

14

You must use all or none of the four 10-GE interfaces as IRF physical ports. The four interfaces canbe bound to different IRF ports.

Before you bind a 10-GE interface to an IRF port or remove it from the IRF port, you must shut downall the 10-GE interfaces of the 40-GE port. If any of the interfaces is in up state, the bind or removeaction will fail.

Bring up the interfaces after you complete the operation.SFP+ port restrictions for the HP 5900AF-48XG-4QSFP+ Switch, HP 5900AF-48XG-4QSFP+TAA-compliant Switch, HP 5920AF-24XG Switch, and HP 5920AF-24XG TAA-compliant Switch

When you use the SFP+ ports on the HP 5900AF-48XG-4QSFP+ Switch (JC772A), HP5900AF-48XG-4QSFP+ TAA-compliant Switch (JG554A), HP 5920AF-24XG Switch (JG296A), and HP5920AF-24XG TAA-compliant Switch (JG555A) as IRF physical ports, follow these guidelines:

The SFP+ ports are grouped by port number in order, starting from 1. Each group contains four ports.If you use one port in a group for IRF connection, you must also use all the other ports in the groupfor IRF connection. However, you can bind them to different IRF ports.

Before you bind an SFP+ port to an IRF port or remove it from the IRF port, you must shut down all

the SFP+ ports in the same group.Bring up the ports after you complete the operation.

SFP+ port restrictions for the HP 5900AF-48G-4XG-2QSFP+ Switch

When you use the SFP+ ports on the HP 5900AF-48G-4XG-2QSFP+ Switch (JG510A) as IRF physicalports, follow these guidelines:

If you use one SFP+ port for IRF connection, you must also use all the other SFP+ port for IRFconnection. However, you can bind them to different IRF ports.

Before you bind an SFP+ port to an IRF port or remove it from the IRF port, you must shut down allthe SFP+ ports.

Bring up the ports after you complete the operation.

Connecting IRF ports

When you connect two neighboring IRF members, you must connect IRF-port 1 on one member toIRF-port 2 on the other.

IRF link redundancy

You can bind up to four physical ports to an IRF port for link redundancy.

Multichassis link aggregationFor high availability, connect a downstream device to each IRF member device and assign these links toone link aggregation group.

MAD and IRF domain restrictions

When you configure an IRF fabric, follow these MAD and IRF domain restrictions and guidelines:

If LACP MAD, ARP MAD, or ND MAD runs between two IRF fabrics, assign each fabric a uniqueIRF domain ID. For BFD MAD, this task is optional.


19/64

15

An IRF fabric has only one IRF domain ID. You can change the IRF domain ID by using the followingcommands: irf domain, mad enable, mad arp enable, or mad nd enable. The IRF domain IDsconfigured with these commands overwrite each other.

LACP MAD handles collisions in a different way than BFD MAD, ARP MAD, and ND MAD. Toavoid conflicts, do not enable LACP MAD together with any of those mechanisms in an IRF fabric.However, you can configure BFD MAD, ARP MAD, and ND MAD together in an IRF fabric for

prompt IRF split detection.

To exclude a port from the shutdown action that is executed when an IRF fabric transits to theRecovery state, use the mad exclude interfacecommand. To bring up a port after the IRF fabrictransits to the Recovery state, you must use the mad restorecommand instead of the undo shutdowncommand to activate the entire IRF fabric.

Other configuration guidelines

If a subordinate device uses the same next-startup configuration file name as the master device, thefile might be overwritten depending on your configuration file management settings. To continue to

use the configuration file after removing the switch from the IRF fabric, back up the file before settingup the IRF fabric.

If two IRF fabrics have the same bridge MAC address, they cannot merge unless you change thebridge MAC address of one IRF fabric.

Assign each member a unique IRF member ID to make sure they can merge. You must reboot themembers to validate the IRF member ID settings.

Assign the highest member priority to the device you want to use as the master.

Save any configuration you have made to the startup configuration file before you reboot the IRFmember devices.

Setup and configuration task listHP recommends the basic IRF setup procedure in Figure 9. Perform the tasks in this figure on eachmember device. After the IRF fabric is set up, you can access the IRF fabric to manage its member devicesas if they were one device.

Figure 9Basic IRF setup flow chart

HP recommends the following IRF fabric setup and configuration procedure:

Setup and configuration procedure Remarks

1. (Required.) Planning the IRF fabric setup N/A


20/64

16

Setup and configuration procedure Remarks

2. (Required.)Assigning a member ID to each IRF member devicePerform this task on each memberdevice.

3. (Required.) Specifying a priority for each member device Perform this task on each memberdevice.

4. (Required.) Connecting physical IRF ports N/A

5. (Required.) Binding physical ports to IRF ports

Perform this task on each memberswitch.

When you complete IRF portbinding and activation on all IRFmember devices, the IRF fabric isformed.

6. (Required.)Accessing the IRF fabric

When you log in to the IRF fabric,you are placed at the master's CLI,where you complete subsequent

IRF settings and configure otherfeatures for the member devices asif they were one device.

7. (Optional.) Configuring a member device description N/A

8. (Optional.) Configuring IRF link load sharing mode:

Configuring the global load sharing mode

Configuring a port-specific load sharing mode

N/A

9. (Optional.) Configuring IRF bridge MAC persistence N/A

10. (Optional.) Enabling software auto-update for software imagesynchronization

HP recommends enablingsoftware auto-update to make suresystem software imagesynchronization.

11. (Optional.) Setting the IRF link down report delay N/A

12. (Required.) Configuring MAD:

Configuring LACP MAD

Configuring BFD MAD

Configuring ARP MAD

Configuring ND MAD

Excluding a port from the shutdown action upon detection ofmulti-active collision

MAD mechanisms areindependent of one another. Youcan configure at least one MADmechanism for an IRF fabric.

13. (Optional.) Recovering an IRF fabric N/A

Planning the IRF fabric setupConsider the following items when you plan an IRF fabric:

Hardware compatibility and restrictions

IRF fabric size

Master switch

IRF physical ports


21/64

17

Member ID and priority assignment scheme

Fabric topology and cabling scheme

For more information about hardware and cabling, see the switch installation guide.

Assigning a member ID to each IRF member deviceCAUTION:

In an IRF fabric, changing IRF member IDs might cause undesirable configuration changes and even dataloss. Before you do that, back up the configuration and make sure you fully understand the impact on yournetwork. For example, all member switches in an IRF fabric are the same model. If you swapped the IDsof any two members, their interface settings would also be swapped.

By default, the member IDs of all switches are 1. To create an IRF fabric, you must assign a unique IRFmember ID to each member device.

Perform this task before the IRF fabric is formed. To prevent any undesirable configuration change or dataloss, avoid changing member IDs after the IRF fabric is formed.

The new member ID takes effect at a reboot. After the switch reboots, the settings on all memberID-related physical resources (including common physical network ports) are removed and requirereconfiguration, regardless of whether you have saved the configuration.

To assign a member ID to a device:

Step Command Remarks

1. Enter system view. system-view N/A

2. Assign a member ID to a

member device.

irf member member-id renumber

new-member-id The default IRF member ID is 1.

3. (Optional.) Save theconfiguration. save

If you have bound physical portsto IRF ports or assigned memberpriority, save the configurationbefore rebooting the device sothese settings can continue totake effect after the reboot.

4. Reboot the device. reboot [ slotslot-number ] [ force ] N/A

Specifying a priority for each member deviceIRF member priority represents the possibility for a device to be elected the master in an IRF fabric. Thehigher the priority, the higher the possibility.

A member priority change affects the election result at the next master election, but it does not causeimmediate master re-election.

To specify a priority for a member device:




22/64

18


2. Specify a priority for thedevice. irf member member-idpriority priority

The default IRF member priorityis 1.

Connecting physical IRF portsWhen you connect two neighboring IRF members, connect the physical ports of IRF-port 1 on onemember to the physical ports of IRF-port 2 on the other, as shown in Figure 10.

Suppose you have four chassis: A, B, C, and D. IRF-port 1 and IRF-port 2 are represented by A1 and A2on chassis A, represented by B1 and B2 on chassis B, and so on. To connect the four chassis into a ringtopology of A-B-C-D(A), the IRF link cabling scheme must be one of the following:

A1-B2, B1-C2, C1-D2, and D1-A2.

A2-B1, B2-C1, C2-D1, and D2-A1.

IMPORTANT:

No intermediate devices are allowed between neighboring members.

Figure 10Connecting IRF physical ports

Connect the devices into a daisy chain topology or a ring topology. A ring topology is more reliable(see Figure 11). In ring topology, the failure of one IRF link does not cause the IRF fabric to split as in daisychain topology. Rather, the IRF fabric changes to a daisy chain topology without interrupting networkservices.

To use the ring topology, you must have at least three member devices.


23/64

19

Figure 11Daisy chain topology vs. ring topology

Binding physical ports to IRF portsWhen you bind physical ports to IRF ports, follow these guidelines:

Follow the restrictions in "IRF physical port restrictions and binding requirements."

You must always shut down a physical port before binding it to an IRF port or removing the binding.Start the shutdown operation on the master and then the member device that has the fewest numberof hops from the master.

On a physical port bound to an IRF port, you can execute only the shutdown, description,

priority-flow-control, and flow-intervalcommands. For more information about these commands, seeLayer 2LAN SwitchingCommand Reference.

To bind physical ports to IRF ports:



2. Enter interface view orinterface range view.

Enter interface range view:

Method 1:interface range { interface-typeinterface-number[ to

interface-typeinterface-number] } &

Method 2:interface rangenamename[ interface{ interface-typeinterface-number[tointerface-typeinterface-number] } & ]

Enter interface view:interface interface-typeinterface-number

To shut down a range of physicalIRF ports, enter interface rangeview.

To shut down one physical IRFport, enter its interface view.

IRF

Ring topology

Subordinate Subordinate

Master

IRF-Port1 IRF-Port2

IRF-Port1

IRF-Port2IRF-Port1

IRF-Port2

Daisy chain topology

IRF

Master

Subordinate

Subordinate

IRF-Port2

IRF-Port2

IRF-Port1

IRF-Port1


24/64

20


3. Shut down the port or therange of ports. shutdown By default, all ports are up.

4. Return to system view. quit N/A

5. Enter IRF port view.irf-port member-id/port-number N/A

6. Bind each physical port tothe IRF port.

port group interface interface-typeinterface-number

By default, no physical ports arebound to any IRF port.

Repeat this step to assign multiplephysical ports to the IRF port forlink redundancy.

You can bind up to four physicalports to an IRF port. The bindingattempt will fail if you have boundfour physical ports to the IRF port.



Enter interface range view: Method 1:

interface range { interface-typeinterface-number[ tointerface-typeinterface-number] } &



N/A

9. Bring up the port or therange of ports. undo shutdown N/A


11. Save the configuration. save

Activating IRF port configurationscan cause IRF merge and reboot.To avoid data loss, save therunning configuration to thestartup configuration file before

you perform the operation.

12. Activate the IRF portsettings. irf-port-configuration active

After this step is performed, thestate of the IRF port changes toUP, the member devices elect amaster automatically, and thesubordinate device rebootsautomatically.

After the IRF fabric is formed, youcan add more physical ports to anIRF port (in UP state) withoutperforming this step.


25/64

21

Accessing the IRF fabricThe IRF fabric appears as one device after it is formed. You configure and manage all IRF members atthe CLI of the master. All settings you have made are propagated to the IRF members automatically.

You can access an IRF fabric's CLI in one of the following ways:Local loginLog in through the console port of any member device.

Remote loginLog in at a Layer 3 interface on any member device by using methods includingTelnet and SNMP.

When you log in to an IRF fabric, you are placed at the CLI of the master, regardless of at which memberdevice you are logged in. After that, you can access the CLI of a subordinate device to execute a limitedset of maintenance commands.

For more information, see the chapter on login in Fundamentals Configuration Guide.

Configuring a member device descriptionYou can configure a description to describe the location or purpose of a member device.

To configure a description for a member device:



2. Configure a description for amember device. irf member member-iddescription text

By default, no member devicedescription is configured.

Configuring IRF link load sharing modeOn an IRF port, traffic is balanced across its physical links. By default, traffic is distributed automaticallybased on packet types, including Layer 2, IPv4, and IPv6. You can configure the IRF port to distributetraffic based on certain criteria, including IP addresses, MAC addresses, and any combination of thecriteria. If the device does not support a criterion combination, the system displays an error message.

Configure the IRF link load sharing mode for IRF links in system view or IRF port view:

In system view, the configuration is global and takes effect on all IRF ports.

In IRF port view, the configuration is port specific and takes effect only on the specified IRF port.An IRF port preferentially uses the port-specific load sharing mode. If no port-specific load sharing modeis available, an IRF port uses the global load sharing mode.

The IRF link load sharing mode takes effect on all types of packets, including unicast, multicast, andbroadcast.


26/64

22

NOTE:

To distribute traffic based on TCP/UDP ports, use one of the following methods:

Use the default setting for both global and port-specific IRF link load sharing modes.

Set the IRF link load sharing mode to be based on source IP, destination IP, or their combination (irf-portload-sharing mode{ destination-ip| source-ip } *), and set the global Ethernet link aggregation load

sharing mode to be based on source service port, destination service port, or their combination(link-aggregation global load-sharing mode{ destination-port| source-port} *). For moreinformation about Ethernet link aggregation load sharing, see Layer 2LAN Switching ConfigurationGuide.

Configuring the global load sharing mode



2. Configure the global IRF linkload sharing mode.

irf-port global load-sharing mode{ destination-ip| destination-mac|source-ip |source-mac } *

By default, packets aredistributed automatically acrossIRF member links based onpacket types.

If you execute this commandmultiple times, the most recentconfiguration takes effect.

Configuring a port-specific load sharing mode

Before you configure a port-specific load sharing mode, make sure you have bound at least one physicalport to the IRF port.

To configure a port-specific load sharing mode for an IRF port:



2. Enter IRF port view. irf-port member-id/port-number N/A

3. Configure the port-specificload sharing mode.

irf-port load-sharing mode{ destination-ip| destination-mac|source-ip |source-mac} *

By default, packets aredistributed automatically across

IRF member links based onpacket types.

If you execute this commandmultiple times, the most recentconfiguration takes effect.


27/64

23

Configuring IRF bridge MAC persistenceAn IRF fabric by default uses the bridge MAC address of the master device as its bridge MAC address.Layer 2 protocols, such as LACP, use this bridge MAC address to identify the IRF fabric. On a switchedLAN, the bridge MAC address must be unique.

To avoid duplicate bridge MAC addresses, an IRF fabric can change its bridge MAC addressautomatically after its master leaves, but the change can cause temporary traffic disruption.

Depending on your network condition, enable the IRF fabric to preserve or change its bridge MACaddress after the master leaves. Available options include:

irf mac-address persistent timerBridge MAC address of the IRF fabric is retained for 6 minutesafter the master leaves. If the master does not return before the timer expires, the IRF fabric uses thebridge MAC address of the new master as its bridge MAC address. This option avoids unnecessarybridge MAC address changes due to a device reboot, transient link failure, or purposeful linkdisconnection.

irf mac-address persistent alwaysBridge MAC address of the IRF fabric does not change afterthe master leaves.

undo irf mac-address persistentBridge MAC address of the new master replaces the original oneas soon as the old master leaves.

IMPORTANT:

If ARP MAD or ND MAD is used, configure the undo irf mac-address persistentcommand to enableimmediate bridge MAC address change after a master leaves.

If TRILL is configured, configure the irf mac-address persistent alwayscommand to retain the bridgeMAC address even if the master has changed.

If a daisy-chained IRF fabric has aggregate links with upstream or downstream devices, configure the irf

mac-address persistent alwayscommand. The setting prevents a bridge MAC change from causingtransmission delay or packet loss at a master/subordinate switchover.

If two IRF fabrics have the same bridge MAC address, they cannot merge.

To configure the IRF bridge MAC persistence setting:



2. Configure IRF bridge MACpersistence.

Retain the bridge MAC addresseven if the master has changed:

irf mac-address persistent always

Preserve the bridge MAC addressfor 6 minutes after the masterleaves:irf mac-address persistent timer

Change the bridge MAC addressas soon as the master leaves:undo irf mac-address persistent

By default, the IRF fabricpreserves its bridge MACaddress for 6 minutes after themaster leaves.


28/64

24

Enabling software auto-update for software imagesynchronization

IMPORTANT:To ensure a successful software auto-update in a multi-user environment, prevent the action of rebootingmember devices during the software auto-update process. You can configure the information center tooutput the software auto-update status to configuration terminals (see Network Management andMonitoring Configuration Guide).

The software auto-update function synchronizes the current software images of the master in an IRF fabricto all its members automatically.

To join an IRF fabric, a device must use the same set of software images as the master in the fabric.

When you add a device to the IRF fabric, the software auto-update function compares the startupsoftware images of the joining device with the current software images of the IRF master. If the two setsof images are different, the device performs the following tasks automatically:

1. Downloads the current software images of the master.

2. Sets the downloaded images as its main startup software images.

3. Reboots with the new software images to rejoin the IRF fabric.

If software auto-update is disabled, you must manually update the device with the software images of themaster before adding it to the IRF fabric.

Configuration prerequisites

Make sure the device you are adding to the IRF fabric has sufficient storage space for the new softwareimages.

If sufficient storage space is not available, the device deletes the current software images automatically.If the reclaimed space is still insufficient, the device cannot complete the auto-update, and you mustreboot the device and access the Boot menu to delete files.

Configuration procedure

To enable an IRF fabric to synchronize software images of the master automatically to the devices you areadding to the IRF fabric:



2. Enable softwareauto-update.

irf auto-update enableBy default, software auto-updateis enabled.


29/64

25

Setting the IRF link down report delayYou can avoid IRF link flapping causing frequent IRF splits and merges during a short time by configuringthe IRF ports to delay reporting link down events. An IRF port works as follows:

When the IRF link changes from up to down, the port does not immediately report the change to theIRF fabric. If the IRF link state is still down when the delay time is reached, the port reports thechange to the IRF fabric.

When the IRF link changes from down to up, the link layer immediately reports the event to the IRFfabric.

To set the IRF link down report delay:



2. Set the IRF link downreport delay. irf link-delayinterval

The default IRF link down report delay is 4

seconds.HP recommends setting the delay to 0 seconds inone of the following situation:

The IRF fabric requires a fastmaster/subordinate or IRF link switchover.

The BFD or GR feature is used.

The IRF fabric might run other protocols (forexample, CFD, VRRP, FCoE, and OSPF) that havea shorter protocol packet lifetime than the delayinterval. For stable protocol running, make surethe delay interval is shorter than the maximumlifetime of these protocol packets. You can adjusteither the IRF link down report delay or themaximum lifetime of the protocol packets.

Configuring MADThe following MAD mechanisms are available for detecting multi-active collisions in different networkscenarios:

LACP MAD

BFD MAD

ARP MAD

ND MAD

LACP MAD handles collisions in a different way than BFD MAD, ARP MAD, and ND MAD. To avoidconflicts, do not enable LACP MAD together with any of those mechanisms in an IRF fabric. However,you can use BFD MAD, ARP MAD, and ND MAD together.

Table 1provides a reference for you to make a MAD mechanism selection decision.


30/64

26

Table 1A comparison of the MAD mechanisms

MADmechanism

Advantages Disadvantages Application scenario

LACP MAD

Detection speed is fast.

Requires no MAD-dedicatedphysical ports or interfaces.

Requires an intermediatedevice that supportsextended LACP for MAD.

Link aggregation is usedbetween the IRF fabric

and its upstream ordownstream device.

For information aboutLACP, see Layer 2LANSwitching ConfigurationGuide.

BFD MAD

Detection speed is fast. No intermediate device is

required.

Intermediate device, if used,can come from any vendor.

Requires MAD dedicatedphysical ports and Layer3 interfaces, whichcannot be used fortransmitting user traffic.

If no intermediate deviceis used, the IRF membersmust be fully meshed.

If an intermediate deviceis used, every IRFmember must connect tothe intermediate device.

No specialrequirements fornetwork scenarios.

If no intermediatedevice is used, this

mechanism is onlysuitable for IRF fabricsthat have a smallnumber of membersthat aregeographically closeto one another.

For information aboutBFD, see HighAvailabilityConfiguration Guide.

ARP MAD

No intermediate device isrequired.


Requires no MAD dedicatedports.

Detection speed is slowerthan BFD MAD and LACPMAD.

The spanning tree featuremust be enabled.

Spanning tree-enabled

non-link aggregation IPv4network scenario.

For information aboutARP, see Layer 3IPServices ConfigurationGuide.

ND MAD

No intermediate device isrequired.


Requires no MAD dedicated

ports.

Detection speed is slowerthan BFD MAD and LACPMAD.

The spanning tree feature

must be enabled.

Spanning tree-enablednon-link aggregation IPv6network scenario.

For information aboutND, see Layer 3IP

Services ConfigurationGuide.

Configuring LACP MAD

When you use LACP MAD, follow these guidelines:

The intermediate device must be a device that supports extended LACP for MAD.

If the intermediate device is also an IRF fabric, assign the two IRF fabrics different domain IDs forcorrect split detection.


31/64

27

Use dynamic link aggregation mode. MAD is LACP dependent. Even though LACP MAD can beconfigured on both static and dynamic aggregate interfaces, it takes effect only on dynamicaggregate interfaces.

Configure link aggregation settings on the intermediate device.

To configure LACP MAD:



2. Assign a domain ID to the IRFfabric. irf domain domain-id The default IRF domain ID is 0.

3. Create an aggregateinterface and enteraggregate interface view.

Enter Layer 2 aggregate interfaceview:interface bridge-aggregationinterface-number

Enter Layer 3 aggregate interfaceview:

interface route-aggregationinterface-number

Use either command.

Perform this step also on theintermediate device.

4. Configure the aggregationgroup to operate in dynamicaggregation mode.

link-aggregation mode dynamic

By default, an aggregationgroup operates in staticaggregation mode.

Perform this step also on theintermediate device.

5. Enable LACP MAD. mad enableBy default, LACP MAD isdisabled.


7. Enter Ethernet interface viewor interface range view.

Enter interface range view: Method 1:

interface range { interface-typeinterface-number[ tointerface-typeinterface-number] } &


Enter Ethernet interface view:interface interface-typeinterface-number

To assign a range of ports to theaggregation group, enterinterface range view.

To assign one port to theaggregation group, enterEthernet interface view.

8. Assign the Ethernet port orthe range of Ethernet ports tothe specified aggregationgroup.

port link-aggregation groupnumber

Multichassis link aggregation isallowed.

Perform this step on theintermediate device as well.


32/64

28

Configuring BFD MAD

When you use BFD MAD, follow these guidelines:

Category Restrictions and guidelines

BFD MAD VLAN

Do not enable BFD MAD on VLAN-interface 1.

If you are using an intermediate device, assign the ports of BFD MAD links tothe BFD MAD VLAN on the device.

The IRF fabrics in a network must use different BFD MAD VLANs.

For correct traffic forwarding, make sure the BFD MAD VLAN contains onlyports on the BFD MAD links. Exclude a port from the BFD MAD VLAN if theport is not on the BFD MAD link. For example, if you have assigned the portto all VLANs by using the port trunk permit vlan allcommand, use the undoport trunk permitcommand to exclude the port from the BFD MAD VLAN.

BFD MAD VLAN and featurecompatibility

Do not use the BFD MAD VLAN for any purpose other than configuring BFDMAD. Layer 2 or Layer 3 features, including ARP and LACP, cannot work on

the BFD MAD-enabled VLAN interface or any port in the VLAN. If youconfigure any other feature on the VLAN, neither the configured feature northe BFD MAD function will work correctly.

Disable the spanning tree feature on all Layer 2 Ethernet ports in the BFDMAD VLAN. The MAD function is mutually exclusive with the spanning treefeature.

Do not bind a BFD MAD-enabled VLAN interface to any VPN instance. TheMAD function is mutually exclusive with VPN.

MAD IP address

To avoid problems, only use the mad ip addresscommand to configure IPaddresses on the BFD MAD-enabled VLAN interface. Do not configure an IPaddress with the ip addresscommand or configure a VRRP virtual address onthe BFD MAD-enabled VLAN interface.

All MAD IP addresses on the BFD MAD-enabled VLAN interface must be onthe same subnet.

To configure BFD MAD:



2. Assign a domain ID to theIRF fabric.

irf domain domain-idBy default, the domain ID of anIRF fabric is 0.

3.

Create a VLAN dedicated toBFD MAD. vlanvlan-id The default VLAN on the deviceis VLAN 1.



33/64

29



Enter interface range view:

Method 1:interface range { interface-typeinterface-number[ to

interface-typeinterface-number] } &



To assign a range of ports to theBFD MAD VLAN, enter interfacerange view.

To assign one port to the BFDMAD VLAN, enter Ethernetinterface view.

6. Assign the port or the rangeof ports to the BFD MADVLAN.

Assign the port to the VLAN as an

access port:port access vlanvlan-id

Assign the port to the VLAN as atrunk port:port trunk permit vlan vlan-id

Assign the port to the VLAN as ahybrid port:port hybrid vlan vlan-id { tagged|untagged}

Choose one commanddepending on the port type.

BFD MAD has no requirementsfor link type of the detection port.

The default link type of a port isaccess.


8. Enter VLAN interface view.interface vlan-interface

vlan-interface-idN/A

9. Enable BFD MAD. mad bfd enableBy default, BFD MAD isdisabled.

10. Configure a MAD IP addressfor a member device on theVLAN interface.

mad ip addressip-address{ mask|mask-length} membermember-id

By default, no MAD IP address isconfigured on any VLANinterface.

Repeat this step to assign a MADIP address to each memberdevice on the VLAN interface.

Configuring ARP MADWhen you use ARP MAD, follow these guidelines:

Do not configure ARP MAD on VLAN-interface 1.

Do not use the VLAN configured for ARP MAD for any other purposes.

If an intermediate device is used, you can use common data links as ARP MAD links. If nointermediate device is used, set up dedicated ARP MAD links between IRF member devices.

If an intermediate device is used, make sure the following requirements are met:

Run the spanning tree feature between the IRF fabric and the intermediate device.


34/64

30

Enable the IRF fabric to change its bridge MAC address as soon as the master leaves.

Create an ARP MAD VLAN and assign the ports on the ARP MAD links to the VLAN.

If the intermediate device is also an IRF fabric, assign the two IRF fabrics different domain IDsfor correct split detection.

To configure ARP MAD:



2. Assign a domain ID to the IRFfabric.

irf domaindomain-id The default IRF domain ID is 0.

3. Configure the IRF bridgeMAC address to change assoon as the master leaves.

undo irf mac-address persistentBy default, the IRF fabric'sbridge MAC address persists 6minutes after the master leaves.

4. Create a VLAN dedicated toARP MAD. vlanvlan-id

The default VLAN on the deviceis VLAN 1.


6. Enter Ethernet interface view.interface interface-typeinterface-number

N/A

7. Assign the port to the ARPMAD VLAN.

Assign the port to the VLAN as anaccess port:port access vlanvlan-id


Assign the port to the VLAN as ahybrid port:port hybrid vlan vlan-id { tagged|untagged}


ARP MAD detection has norequirements for link type of thedetection port.

The default link type of a port is

access.


9. Enter VLAN interface view.interface vlan-interfacevlan-interface-id

N/A

10. Assign the interface an IPaddress.

ip addressip-address{ mask|mask-length}

By default, no IP address isassigned to any VLAN interface.

11. Enable ARP MAD. mad arp enableBy default, ARP MAD isdisabled.

Configuring ND MAD

When you use ND MAD, follow these guidelines:

Enable the IRF fabric to change its bridge MAC address as soon as the master leaves.

If the intermediate device is also an IRF fabric, assign the two IRF fabrics different domain IDs forcorrect split detection.

If an intermediate device is used, you can use common data links as ND MAD links. If nointermediate device is used, set up dedicated ND MAD links between IRF member devices.

Use a VLAN dedicated to ND MAD.


35/64

31

Do not configure ND MAD on VLAN-interface 1.

On the intermediate device, create the ND MAD VLAN and assign the ports on the ND MAD linksto the VLAN.

To configure ND MAD:

Step Command Remarks1. Enter system view. system-view N/A

2. Assign a domain ID to the IRFfabric.

irf domaindomain-id The default IRF domain ID is 0.

3. Configure the IRF bridgeMAC address to change assoon as the master leaves.

undo irf mac-address persistentBy default, the IRF fabric'sbridge MAC address persists 6minutes after the master leaves.

4. Create a VLAN dedicated toND MAD. vlanvlan-id

The default VLAN on the deviceis VLAN 1.


6. Enter Ethernet interface view. interface interface-typeinterface-number

N/A

7. Assign the port to the NDMAD VLAN.

Assign the port to the VLAN as anaccess port:port access vlanvlan-id


Assign the port to the VLAN as ahybrid port:port hybrid vlan vlan-id { tagged|

untagged}


ND MAD detection has norequirements for link type of thedetection port.

The default link type of a port isaccess.


9. Enter VLAN interface view.interface vlan-interfacevlan-interface-id

N/A

10. Assign the interface an IPaddress.

ipv6 address{ipv6-address/pre-length |ipv6address pre-length }

By default, no IPv6 address isassigned to any VLAN interface.

11. Enable ND MAD. mad nd enable By default, ND MAD is disabled.

Excluding a port from the shutdown action upon detection ofmulti-active collision

By default, all ports except the console and physical IRF ports shut down automatically when the IRFfabric transits to the Recovery state.

You can exclude a network port from the shutdown action for management or other special purposes. Forexample:

Exclude a port from the shutdown action so you can Telnet to the port for managing the device.


36/64

32

Exclude a VLAN interface and its Layer 2 ports from the shutdown action so you can log in throughthe VLAN interface.

CAUTION:

Excluding a VLAN interface and its Layer 2 ports from the shutdown action introduces IP collision risksbecause the VLAN interface might be active on both the active and inactive IRF fabrics.

To configure a port to not shut down when the IRF fabric transits to the Recovery state:



2. Configure a network port tonot shut down when the IRFfabric transits to theRecovery state.

mad exclude interface interface-typeinterface-number

By default, all network ports on aRecovery-state IRF fabric are shutdown, except for the IRF physicalports and console port.

Recovering an IRF fabricWhen the failed IRF link between two split IRF fabrics is recovered, all member devices in the inactive IRFfabric (in Recovery state) join the active IRF fabric as subordinate members automatically, and thenetwork ports that have been shut down by MAD restore their original physical state automatically, asshown in Figure 12.

Figure 12Recovering the IRF fabric

If the active IRF fabric has failed, for example, because of device or link failures, before the IRF link is

recovered (see Figure 13), use the mad restorecommand on the inactive IRF fabric to change its MADstate to Detect and bring up all physical ports that were shut down by MAD. After you repair the IRF link,the two parts merge into a unified IRF fabric.


37/64

33

Figure 13Active IRF fabric fails before the IRF link is recovered

To manually recover an inactive IRF fabric:

Step Command

1. Enter system view. system-view

2. Recover the inactive IRF fabric. mad restore

After the IRF fabric is recovered, all ports that have been shut down by MAD come up automatically.

Displaying and maintaining an IRF fabric

Execute displaycommands in any view.

Task Command

Display information about all IRFmembers.

display irf

Display the IRF fabric topology. display irf topology

Display IRF link information. display irf link


38/64

34

Task Command

Display IRF configuration. display irf configuration

Display the load sharing mode for

IRF links. display irf-port load-sharing mode [ irf-port[ member-id/port-number ] ]

Display MAD configuration. display mad [verbose ]

Configuration examplesThis section provides IRF configuration examples for IRF fabrics that use different MAD mechanisms.

LACP MAD-enabled IRF configuration exampleNetwork requirements

Set up a four-chassis IRF fabric at the access layer of the enterprise network in Figure 14.

Configure LACP MAD on the multichassis aggregation to Device E, an HP device that supports extendedLACP.

Figure 14Network diagram


39/64

35


1. Configure Device A:

# Shut down the physical ports used for IRF connection. This example uses the SFP+ port group thatcontains Ten-GigabitEthernet 1/0/45 to Ten-GigabitEthernet 1/0/48 for IRF connection.

syst em- vi ew

[ Sysname] i nt er f ace r ange name i r f i nt erf ace ten- gi gabi t et her net 1/ 0/ 45 t ot en- gi gabi t et her net 1/ 0/ 48

[ Sysname- i f - r ange- i r f ] shut down

[ Sysname- i f - r ange- i rf ] qui t

# Bind Ten-GigabitEthernet 1/0/45 and Ten-GigabitEthernet 1/0/46 to IRF-port 1/1.

[ Sysname] i r f - por t 1/ 1

[ Sysname- i r f - por t 1/ 1] por t gr oup i nt er f ace ten- gi gabi t et her net 1/ 0/ 45


[ Sysname- i r f - por t 1/ 1] qui t


[ Sysname] i r f - por t 1/ 2[ Sysname- i r f - por t 1/ 2] por t gr oup i nt er f ace ten- gi gabi t et her net 1/ 0/ 47



# Bring up the SFP+ ports and save the configuration.

[ Sysname] i nter f ace r ange name i r f

[ Sysname- i f - r ange- i r f ] undo shut down


[ Sysname] save

# Activate the IRF port configuration.

[ Sysname] i r f - por t - conf i gur at i on act i ve

2. Configure Device B:

# Change the member ID of Device B to 2 and reboot the device to validate the change.

syst em- vi ew

[ Sysname] i r f member 1 renumber 2

Renumber i ng t he member I D may r esul t i n conf i gur at i on change or l oss. Cont i nue? [Y/ N] : y

[ Sysname] qui t

r eboot

# Connect Device B to Device A as shown in Figure 14, and log in to Device B. This example usesthe SFP+ group that contains Ten-GigabitEthernet 2/0/45 to Ten-GigabitEthernet 2/0/48 for IRFconnection.

# Shut down the physical ports.

syst em- vi ew

[ Sysname] i nt er f ace r ange name i r f i nt erf ace ten- gi gabi t et her net 2/ 0/ 45 t o

t en- gi gabi t et her net 2/ 0/ 48








40/64

36











[ Sysname] save



The two devices perform master election, and the one that has lost the election reboots to form anIRF fabric with the master.

3. Configure Device C:# Change the member ID of Device C to 3 and reboot the device to validate the change.

syst em- vi ew



[ Sysname] qui t

r eboot

# Connect Device C to Device A as shown in Figure 14, and log in to Device C. This example usesthe SFP+ port group that contains Ten-GigabitEthernet 3/0/45 to Ten-GigabitEthernet 3/0/48 forIRF connection.

# Shut down the physical ports. syst em- vi ew



















[ Sysname] save


41/64

37



Device C reboots to join the IRF fabric.

4. Configure Device D:

# Change the member ID of Device D to 4 and reboot the device to validate the change.

syst em- vi ew



[ Sysname] qui t

r eboot

# Connect Device D to Device B and Device C as shown in Figure 14, and log in to Device D. Thisexample uses the SFP+ port group that contains Ten-GigabitEthernet 4/0/45 toTen-GigabitEthernet 4/0/48 for IRF connection.


syst em- vi ew

[ Sysname] i nt er f ace r ange name i r f i nt erf ace ten- gi gabi t et her net 4/ 0/ 45 t ot en- gi gabi t et her net 4/ 0/ 48

















[ Sysname] save



Device D reboots to join the IRF fabric. A four-chassis IRF fabric is formed.

5. Configure LACP MAD on the IRF fabric:

# Set the domain ID of the IRF fabric to 1.

syst em- vi ew

[ Sysname] i r f domai n 1

# Create a dynamic aggregate interface and enable LACP MAD.

[ Sysname] i nter f ace bri dge- aggregati on 2

[ Sysname- Br i dge- Aggr egat i on2] l i nk- aggregat i on mode dynami c

[ Sysname- Br i dge- Aggr egat i on2] mad enabl e

You need t o ass i gn a domai n I D ( r ange: 0- 4294967295)


42/64

38

[ Cur r ent domai n i s: 1] :

The assi gned domai n I D i s: 1

I nf o: MAD LACP onl y enabl e on dynami c aggregat i on i nter f ace.

[ Sysname- Br i dge- Aggr egati on2] qui t

# Assign Ten-GigabitEthernet 1/0/2, Ten-GigabitEthernet 2/0/1, Ten-GigabitEthernet 3/0/2,and Ten-GigabitEthernet 4/0/1 to the aggregate interface.


[ Sysname- Ten- Gi gabi t Et hernet 1/ 0/ 2] por t l i nk- aggr egati on gr oup 2

[ Sysname- Ten- Gi gabi t Et hernet 1/ 0/ 2] qui t








[ Sysname- Ten- Gi gabi t Et hernet 4/ 0/ 1] por t l i nk- aggr egati on gr oup 2[ Sysname- Ten- Gi gabi t Et hernet 4/ 0/ 1] qui t

6. Configure Device E as the intermediate device:

CAUTION:

If the intermediate device is also an IRF fabric, assign the two IRF fabrics different domain IDs forcorrect split detection. False detection causes IRF split.

# Create a dynamic aggregate interface.

syst em- vi ew

[ Sysname] i nter f ace bri dge- aggregati on 2

[ Sysname- Br i dge- Aggr egat i on2] l i nk- aggregat i on mode dynami c

[ Sysname- Br i dge- Aggr egati on2] qui t

# Assign Ten-GigabitEthernet 1/0/1, Ten-GigabitEthernet 1/0/2, Ten-GigabitEthernet 1/0/3,and Ten-GigabitEthernet 1/0/4 to the aggregate interface.







[ Sysname] i nt er f ace t en- gi gabi t ether net 1/ 0/ 3[ Sysname- Ten- Gi gabi t Et hernet 1/ 0/ 3] por t l i nk- aggr egati on gr oup 2




BFD MAD-enabled IRF configuration example

Network requirements

Set up a four-chassis IRF fabric at the core layer of the enterprise network in Figure 15.


43/64

39

Configure BFD MAD in the IRF fabric and set up BFD MAD links between each member device and theintermediate device.

Disable the spanning tree feature on the ports used for BFD MAD, because the two features conflict witheach other.

Assign the highest member priority to Device A so it can be elected as the master.

Figure 15Network diagram


1. Configure Device A:

# Shut down the physical ports used for IRF connection. This example uses the SFP+ port group thatcontains Ten-GigabitEthernet 1/0/45 to Ten-GigabitEthernet 1/0/48 for IRF connection.

syst em- vi ew











Device A Device B

IRF

Device C Device D

IRFlink

BFD MAD link

Data link

XGE1/0/47

XGE1/0/48

(IRF-port 1/2)

XGE2/0/47XGE2/0/48

(IRF-port 2/1)

XGE2/0/45XGE2/0/46

(IRF-port 2/2)

XGE4/0/45

XGE4/0/46

(IRF-port 4/1)

XGE4/0/47

XGE4/0/48

(IRF-port 4/2)

XGE3/0/47

XGE3/0/48

(IRF-port 3/1)

XGE3/0/45

XGE3/0/46

(IRF-port 3/2)

XGE1/0/45XGE1/0/46

(IRF-port 1/1)

XGE1/0/1

XGE3/0/1 XGE4/0/1

XGE2/0/1

XGE1/0/1 XGE1/0/2XGE1/0/3

XGE1/0/4

Device E


44/64

40









[ Sysname] save



2. Configure Device B:

# Change the member ID of Device B to 2 and reboot the device to validate the change.

syst em- vi ew



[ Sysname] qui t

r eboot

# Connect Device B to Device A as shown in Figure 15, and log in to Device B. This example usesthe SFP+ group that contains Ten-GigabitEthernet 2/0/45 to Ten-GigabitEthernet 2/0/48 for IRFconnection.


syst em- vi ew



















[ Sysname] save



The two devices perform master election, and the one that has lost the election reboots to form anIRF fabric with the master.


45/64

41

3. Configure Device C:

# Change the member ID of Device C to 3 and reboot the device to validate the change.

syst em- vi ew



[ Sysname] qui t

r eboot

# Connect Device C to Device A as shown in Figure 15, and log in to Device C. This example usesthe SFP+ port group that contains Ten-GigabitEthernet 3/0/45 to Ten-GigabitEthernet 3/0/48 forIRF connection.


syst em- vi ew















# Bring up the SFP+ ports and save the configuration.[ Sysname] i nter f ace r ange name i r f



[ Sysname] save



Device C reboots to join the IRF fabric.

4. Configure Device D:

# Change the member ID of Device D to 4 and reboot the device to validate the change.

syst em- vi ew



[ Sysname] qui t

r eboot

# Connect Device D to Device B and Device C as shown in Figure 15, and log in to Device D. Thisexample uses the SFP+ port group that contains Ten-GigabitEthernet 4/0/45 toTen-GigabitEthernet 4/0/48 for IRF connection.


syst em- vi ew


46/64

42



















[ Sysname] save



Device D reboots to join the IRF fabric. A four-chassis IRF fabric is formed.

5. Configure BFD MAD on the IRF fabric:

# Create VLAN 3, and add Ten-GigabitEthernet 1/0/1, Ten-GigabitEthernet 2/0/1,Ten-GigabitEthernet 3/0/1, and Ten-GigabitEthernet 4/0/1 to VLAN 3.

[ Sysname] vl an 3

[ Sysname- vl an3] port t en- gi gabi t ether net 1/ 0/ 1 t en- gi gabi t ether net 2/ 0/ 1

t en- gi gabi t et her net 3/ 0/ 1 t en- gi gabi t et her net 4/ 0/ 1

[ Sysname- vl an3] qui t

# Create VLAN-interface 3, and configure a MAD IP address for each member device on theVLAN interface.

[ Sysname] i nt er f ace vl an- i nt er f ace 3

[ Sysname- Vl an- i nter f ace3] mad bf d enabl e

[ Sysname- Vl an- i nter f ace3] mad i p addr ess 192. 168. 2. 1 24 member 1



[ Sysname- Vl an- i nter f ace3] mad i p addr ess 192. 168. 2. 4 24 member 4[ Sysname- Vl an- i nt erf ace3] qui t

# Disable the spanning tree feature on Ten-GigabitEthernet 1/0/1, Ten-GigabitEthernet 2/0/1,Ten-GigabitEthernet 3/0/1, and Ten-GigabitEthernet 4/0/1.


[ Sysname- Ten- Gi gabi t Ethernet 1/ 0/ 1] undo st p enabl e

[ Sysname- Ten- Gi gabi t Et hernet 1/ 0

HP IRF Configuration Guide

Documents

Transcript of HP IRF Configuration Guide