04-27-2011 10:49 AM - edited 03-19-2019 02:50 AM
We have a PUB/SUB environment for our Unity cluster (Unity Connection 8.x). Our vendor set up the cluster to have the Publisher as the Primary. We were having issues with MWI, so he made the Subscriber the Primary. This didn't clear up the MWI issues, but now when we log into Unity Connection Administration, we get a message that a failover has occurred.
What is the proper way to set up the Unity cluster in a PUB/SUB environment? Should the Publisher be primary or should the Subscriber be primary?
Thanks.
Solved! Go to Solution.
04-27-2011 11:20 AM
Hi Justin,
It should be the Publisher
Table 3-1 Server Status Functions in the Cisco Unity Connection Cluster
Cheers!
Rob
PS: if you let us know what type of MWI issues you are seeing we may be able to help.
04-27-2011 11:20 AM
Hi Justin,
It should be the Publisher
Table 3-1 Server Status Functions in the Cisco Unity Connection Cluster
Cheers!
Rob
PS: if you let us know what type of MWI issues you are seeing we may be able to help.
04-27-2011 11:42 AM
Awesome! Thanks so much for the help!
I was going to start a TAC case on the MWI, but if anyone knows the fix on that, here are the symptoms.
Basically, the MWI stays ON even when there are no messages. The MWI light is ON on the endpoint and in Unity Connection, it shows ON as well. So I have been having our end users check their voicemails periodically, as right now they don't know if they actually have messages waiting.
Thanks again for the help with the PUB/SUB question, much appreciation!
04-27-2011 12:02 PM
What version of Unity Connection are you running? If you are running 8.5 then I recently encountered a bug and several interrelated bugs that would cause MWI issues during some failure scenarious such as failover/failback.
A few things:
1) The Publisher should be the Primary during normal operations. It will be responsible for the DB and also should be primary for web connections.
2) The Subscriber should be configured (in CUCM) to answer most, if not all, of the phone calls to VM.
For MWI:
The Cisco Unity Connection Notifier service plays a role in MWI and it only runs on the server that is the PRIMARY role in the cluster. If you have been running the Subscriber as the Primary (or currently are) then manually failback to the Publisher as primary and restart the CU Notifier service from within the Unity Connection Serviceability page. You will be warned that restarting this service will cause a failover but it will not as long as you restart the service immediately after stopping it (you actually have about 2 minutes of grace period before this would trigger failover). If this resolves your issue, you may well be facing one of the bugs I recently came across and I can provide more information based on your actual symptoms.
Hailey
Please rate helpful posts!
04-29-2011 01:16 PM
Your suggestion on the MWI worked as well.
Much thanks!!
06-09-2014 01:46 PM
it just says "when the cluster is functioning normally, the publisher server has Primary status"
1) but why?
2) why the subscriber server wouldn't be the Primary in a cluster?
The above link just describes the server status functions.
I have tested both scenarios when both servers are running and both worked fine. Seems they can play both roles. I just want to clarify the primary and secondary role in the cluster management page.
Thanks.
06-09-2014 01:48 PM
it just says "when the cluster is functioning normally, the publisher server has Primary status"
1) but why?
2) why the subscriber server wouldn't be the Primary in a cluster?
The above link just describes the server status functions.
I have tested both scenarios when both servers are running and both worked fine. Seems they can play both roles. I just want to clarify the primary and secondary role in the cluster management page.
Thanks.
06-09-2014 06:01 PM
CUC leverages an HA active/active clustering model. So, aside from some critical level services - both servers are fully functional at all times. From there, either server can take over the critical level services in the event the other server goes down. The Publisher should be the Primary from a DB perspective. IOW, one node is primarily responsible for those critical services. The Subscriber should take the majority of calls (i.e. offloading the call processing from the DB/web processing node). If you can take a look at the CUC design guide or the SRND then it explains everything in detail.
Hailey
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide