Slow SIP endpoints failover after one of the VCS peers fails in a cluster

[chrhsu]

With  VCS cluster implementation, there are four options for endpoints to  connect to a cluster: SIP Outbound (preferred), DNS SRV (2nd choice),  DNS Round-robin, and Static IP (lease preferred). I’m using option 1.  With option 1, an EP register to the peers in the cluster at the same  time. However, when I shut one of the VCSes down, the video EP’s take  about 3 minutes to be able to make/receive calls to/from other video  EP’s. Is this the expected behavior or there are some parameters needed  to be tuned?

Thanks!

[Tomonori Taniguchi]

If SIP UAs registered on all cluster VCS by using SIP Outbound, should able to receive the call without downtime (initiate the call is depend on UA implementation which SIP server to be select as outgoing call proxy).

What is “Outbound registration refresh minimum” and “Outbound registration refresh maximum” configuration on VCS?
If you change “Outbound registration refresh minimum (seconds)” to small number (for example, 30 sec.), you should see difference…

[chrhsu]

right now they are the defaults:

Outbound registration refresh minimum” : 300 seconds

“Outbound registration refresh maximum : 3600 seconds

What would be good parameters to use without causing any potential issues yet provide quick time to initiate the calls?

Thanks!

[Tomonori Taniguchi · Accepted solution]

This is very depend on deployment scenario and have no single “best” parameter it into all deployment scenarios.

Most of SIP UAs will try to update registration in half-of-refresh time, so you can calculate how often VCS receive registration request from all SIP UAs.

[chrhsu]

Thanks again!