hsrp convergence time

[john.castanares]

Hello!

We have implemented an hsrp network using two(2) 3750(hsrp1 & hsrp2). I have 6 vlans in my LAN. THe 3 vlans are active state in the hsrp1 and others are active in hsrp2.. Our set-up is typically an example of a best practices of hsrp as it said by the cisco. Now, upon our testing if our hsrp works, we used the laptop to check the connectivity by "continuous pinging"...and patch it to the switch_3(vlan3). We ping the other site(wan) because most of the server was there... Now for the testing, I shut the interface from hsrp1 which is connected to the switch_3. THe pinging got only one timeout... BUt when i put it back, the convergence is very slow and i think we got a timeout of more than 10... and then, it goes back to normal.

Any ideas of the convergence time if it is normal that there is a slow converge when the active goes back?

By the way, in spanning tree, all of my switches are in a default bridge priority. And it seems that not all of the root bridge(per vlan) are there in the two 3750(hsrp1 & hsrp2)... Do we need to force one of them to be the root bridge of all vlans?

Pls. advise... thanks...

-john-

[lgijssel]

The spanning tree root must be at the node that is hsrp-active. The alternative root should be the second hsrp-switch. You must configure this to get hsrp working as desired.

Besides, you have to setup your network in a way that enables quick recovery from spanning-tree topology changes. To achieve this you may use rapid-spanning tree (spanning-tree proto rapid) or uplink-fast/backbone-fast with 802.1d. I believe there are still some unresolved issues with rstp but you may still give it a try. If it works for your network, this is the easiest way to configure your spanning tree. The two links below point to the corresponding config-guides.

http://www.cisco.com/en/US/products/hw/switches/ps5023/products_configuration_guide_chapter09186a008047627f.html#wp1166029

http://www.cisco.com/en/US/products/hw/switches/ps5023/products_configuration_guide_chapter09186a0080476271.html

Regards,

Leo

[Richard Burts]

I believe that Leo is on the right track in believing that the cause of the slow convergence for HSRP is spanning tree. In addition to the things that he has suggested you might want to be sure that switch ports that connect to user end stations are configured with port fast.

HTH

Rick

[dean holroyd]

hsrp hello interval is 3.3 seconds

dead time is 10 seconds (3x hello)

so it could take upto 10 seconds for HSRP to converge

if you want it to converger faster, change the timers

standby ??? timers 1 3

where 1 = hello and 3 = dead