Solved: What is the fastest HSRP timers allowed

jimmyc_2 · ‎04-04-2008

We have two L3 core switches, and one L2 access switch connected to each core. All switches are 4500s and all connections are Gig-etherchannel. The access switch will be will be used to push almost all displays to our NOC, using Sun's Thin-Client adapters. If I configure both the core switches as:

"standby 1 timers msec 500 msec 1500"

will I get a fail over in three seconds?

Should I reduce the timers to 200/600 msecs for less of an outage?

If so, should I use a standby-delay statement also?

Cisco states:

Some HSRP state flapping can occasionally occur if the holdtime is set to less than 250 milliseconds, and the processor is busy. It is recommended that holdtime values less than 250 milliseconds be used on Cisco 7200 platforms or better, and on Fast-Ethernet or FDDI interfaces or better. You can use the standby delay command to allow the interface to come up completely before HSRP initializes.

Thanks. jc

Joseph W. Doherty · ‎04-14-2008

Don't have first hand experience with 4500 series, however I would guess that a 200/600 hello shouldn't be much of a problem unless you have lots of gateway interfaces and/or were using a slower supervisor.

View solution in original post

smahbub · ‎04-10-2008

By default, the hold and the hello timers are set to 3 and 10 seconds, respectively, which means that a hello packet is sent between the HSRP standby group devices every 3 seconds, and the standby device becomes active when a hello packet has not been received for 10 seconds. You can lower these timer settings to speed up the failover or preemption, but, to avoid increased CPU usage and unnecessary standby state flapping, do not set the hello timer below one (1) second or the hold timer below 4 seconds.

The command syntax is :: "standby [group-number] timers hellotime holdtime"

example:: "standby 1 timers 1 3" will set the failover to happen in 3 seconds.

For related information on HSRP timers refer:

http://www.cisco.com/en/US/tech/tk648/tk362/technologies_tech_note09186a0080094afd.shtml#topic14

Joseph W. Doherty · ‎04-11-2008

"standby 1 timers msec 500 msec 1500"

will I get a fail over in three seconds?

Should switch in 1.5 seconds, i.e. when the hold time has expired. The first parameter determines how often to send HSRP announcements, the second, how long to wait without hearing an announcement. Normally they're configured so there are at least 3 announcements during the hold time, which implies that many announcements have been missed.

How low to go is dependent on what the equipment can support and what the application requires, ideally to keep from dropping its connection. You'll might need to test what the Sun Thin-Client adapters will tolerate and what might be nice. (E.g. they might tolerate a multisecond outage, but less would be nicer.) Once you determine the outage time, set that to your second value, set the first to one third or one fourth of it.

The purpose of the standby-delay timer allows the device that's recovering and going to preempt time to stabilize before taking back the gateway. If you don't preempt, shouldn't be much need to use it.

jimmyc_2 · ‎04-14-2008

Hi Joseph,

Thanks for advice. What is the worst that can happen if I go to 200 milliseconds hello and 600 milliseconds hold-down? I can live with excessive traffic, but not with a CPU panic. Any danger of that?

Regards,

Jimmyc

Joseph W. Doherty · ‎04-14-2008

Don't have first hand experience with 4500 series, however I would guess that a 200/600 hello shouldn't be much of a problem unless you have lots of gateway interfaces and/or were using a slower supervisor.