Arp cache

[BCSCB]

A simple question Maybe !!!!

We are running quite a large network consiting of 6500's on the core and 3500's/3560's as distribution/edge.

We have recently encountered a problem with our arp cache.

The problem is when our server cluster fails and moves over to the secondary cluster we cannot access it until the arp cache has timeout which is around 30mins. We can phsically log on to the core switches and clear them but as we have a 24/7 site this means that an engineer has to be called out of normal working hours.

This could be a deeper problem than it looks but is there anyway of refresing the arp table through a command line that would decrease it to say 30 secs or any other input would be great.

 

[vipergg]

Arp timeout is adjustable with the "arp timeout"command which is adjustable from 1 to 14400 seconds . I would be careful with this though, if you make it as low as you would like this means everyone will have to arp everytime they try to transmit which means more broadcasts etc.. on your network . Default is 4 hours for the arp timer . there may be other options but without knowing your network topology its hard to speculate .