UCCX HA over WAN

Unanswered Question
Oct 14th, 2007
User Badges:

Hi All


I am looking at deploying a High Availability standby server on a UCCX Enhanced box over a high speed WAN.

I know that it isn't recommended (or supported) according to the SRND but I am interested if it is actually possible?

Anyone done this before?


It maybe possible to keep the 2 boxes on the same VLAN if this helps but again the SRND says they must be in the same 'building', why i'm not sure??


Any help appreciated


Cheers

Andy

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 4 (2 ratings)
Loading.
Dennis Fogler Sun, 10/14/2007 - 13:59
User Badges:
  • Cisco Employee,

One of the biggest challenges with HA over the WAN is the "split master syndrome" where each UCCX server believes the other to have failed when they loose connectivity with each other during a WAN outage.


In this case each UCCX server beleives it is the primary or master UCCX server and both go into service. Once WAN connectivity is restored they have no mechanism in place to resync their data correctly

Riccardo Bua Sun, 10/14/2007 - 23:23
User Badges:
  • Cisco Employee,

Andy, if you lose Heartbeats to the Unified Call Manager on the master, the slave might not kick in in doubt if there is or not a master in service and do a master reelectioninstead causing the call in progress to be lost while it is busy doing so.


I could assure you it is unfortunately still something quite common in the event of a WAN connection even with some of the most reliable ones.


Regards,


Riccardo

jleon22 Mon, 03/15/2010 - 14:15
User Badges:
  • Bronze, 100 points or more

Hypothetically speaking, what if one of the servers is shutdown prior to the high speed WAN link being restored. Assuming nobody cares about the "new" data/calls on the shutdown server, would it the HA cluster resume normal operations again safely?

Aaron Harrison Mon, 03/15/2010 - 17:16
User Badges:
  • Super Bronze, 10000 points or more
  • Community Spotlight Award,

    Member's Choice, May 2015

Hi


I've had customers hit this kind of issue - usually a bounce of one server restores things, occasionally I believe they've had to reboot both to straighten things out.


All fixed in 8.0, I'm assured..


Aaron

Actions

This Discussion