Sun Network Switch
 

PLA-Support>Sun Systems>SAND>Network>SN-SND: Sun Network Interfaces
 

• Virtual Router Redundancy Protocol (VRRP) and
• Virtual Service Redundancy Protocol (VSRP)

may need a change in an environment with both,

• a large number of VirtualServices and RealServices (more than 100) and
• multiple VRRP routing domains.

The default value is 1 second for both. This default should be changed to avoid increased systemBoard CPU activity and unnecessary Master/Backup transitions.

How can I see if timer changes would be beneficial ?
1. The CLI command

sun> show switchServices chassis cpuLoad

shows a high CPU load (70 % and more), and

2. The VSRP node electionChanges can be seen continually increasing

sun> show redundancy vsrp node

Traffic forwarding will be unstable during these transitions.
The same switchovers can also be seen in the VRRP domains.

How can I change the timers ?
The timers can be changed in the Web GUI or via CLI commands.
This example shows how to set the VSRP Hello timer to 10 s and the VRRP Interval to 5 s on CLI.

- on master:
(config)# redundancy vsrp node 103 helloTime 10
- on slave:
(config)# redundancy vsrp node 102 helloTime 10
- on both, for each VRRP instance:
(config-vSwitch-prods01 vRouter-default vrrp)#
interface ifName vlan.prod01 vrid 1 ipAddresses 10.100.64.1 interval 5

For more information on switch configuration see Sun[TM] N2000 Series Release 2.0 System Configuration Guide (Part No. 817-7637-10) delivered with the switch or online on http://docs.sun.com .

In the 3.0 software release it will not be necessary to change these timer values.

For large configurations, there is a potential for the switch which should be VSRP Backup to thrash between Master and Backup.

The Backup switch will become VSRP Master for a few seconds (the Master switch will still continue to be Master also) because it will detect the VSRP session as down, and then transition back to Backup.

The VSRP node electionChanges can be seen continually increasing.

Traffic forwarding will be unstable during these transitions.

Reason: The switch downloads VirtualService and RealService configuration to its Network Processors, a process which can take multiple seconds.

In release 2.0, there is the potential for VSRP and VRRP traffic to be dropped during this download process. This results in VSRP, and occasionally VRRP, switchovers, which cause the config to be loaded to the Network Processors again, etc.

This problem has been resolved in software release 3.0.

Internal References



See Esc 1-13233370

PTS Product Page: http://rsd-software.sfbay/products/nauticus

This solution has no attachment