Signaling server garp

[a12ozbottle]

I have a node with 2 signaling servers and 3 VGMCs. When the leader fails the 2nd server takes over as the leader, but it will not register any ip sets until the Cisco 3750s arp table is cleared.
We have found a fix for the Cisco switch that sounds like it fits the problem. However it would require upgrading the switch's software. Before this takes place to avoid wasting time I would like to find out if the ISP1100 signaling server running 4.50.88 uses gratuitous arp to update the arp tables.

 

[DaddyOfThree]

I believe that Nortel uses gratuitous ARP (unsolicited ARP broadcasts meant to update the ARP table in the event of a failover of the virtual leader IP address) in a number of redundancy/failover scenarios. The Leader IP address is one and the CPUs (ELAN) in the Call Server itself is another.

Cheers!

 

[a12ozbottle]

Thank you, I finally received confirmation from Nortel and they also verified that gratuitous arp is used. The bug fix for the Cisco 3750 cluster resolved our issues.