Jabber on iPhone/macbook HA failover takes a long time

[Gert Plessers]

We're currently testing UC11.0 and noticed that IM&P failover takes a (very) long time, setup is one high-availability (sub)cluster. When disconnecting the active IM&P server it takes up to 7 - 8 minutes on the macbook client and up to 12 minutes on the iphone jabber client before they re-register and become available on the other server in the subcluster. In the CUCM menu we can see all the clients registering to the one active IM&P server ('Presence Redundancy Group"). We changed the 'Client re-login lower/upper limit' timers to 60 resp 120 - and that fixed the issue for jabber on windows (below 2 min), not for the other clients though.

Are there other parameters in use for the non-windows clients ? I verified the deployment guide and other docs - but can't find any special pointers for these. 

Any pointers appreciated.

Thanks,

GP

[Mark Swanson · Accepted solution]

The re-login thresholds should follow these guidelines;

http://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/im_presence/configAdminGuide/10_5_2/CUP0_BK_CEB3E82E_00_config-admin-guide-imp-1052/CUP0_BK_CEB3E82E_00_config-admin-guide-imp-1052_appendix_010100.pdf

These are Cisco's recommended thresholds... there's nothing wrong with changing these thresholds but if you lower these thresholds before all of the services have a chance to start, then users would likely receive 'false' error messages when they attempt to login.

These thresholds apply to all clients; Jabber for Windows, Jabber for Mac, Jabber for iPhone, etc.

To me, it sounds like the Firewall is blocking traffic for a short period of time; 5 or 10 minutes and then, the Firewall allows the connections. Did you check the Firewall yet?

[Gert Plessers]

Thanks Mark - apologize for the delay but only now we did another disaster recovery / failover test. I fiddled around a bit and it's a bit better. +5 :)

GP