Scheduled restart of IM and Presence server issue

[Bekzod Fakhriddinov]

Hi guys. We r switching over services from one site IM&P server to another site IM&P server as regular procedure every 6 month . I am having issue when I failed over from active IM&P (A server ) to Backup IM&P (B server) (from CUCM-Presence Redundancy Group Configuration-), after fail over status of them Idle (A)and Running in backup mode (B server)  . B is primary now. 

How to make status of both of them Normal and still keep B server as primary ? I tried to restart A server but still status not normal ... I attached picture.

Thank you

[Jonathan Schulenberg]

I believe you are confusing user assignments with the subcluster failover. A subcluster is active/active - both nodes are perfectly capable of having users assigned to it. This is separate from the failover and fallback procedures. If a cluster node fails either an administrator must initiate a fallback or the automatic fallback service parameter must be enabled (it's disabled by default since a fallback is disruptive to the users being moved back to their assigned node).

[Bekzod Fakhriddinov]

right now i have 300 users assigned on IM&P server A  but active on server B. Can I move assignments of users to server B and keep them active on server B ? 

[Jonathan Schulenberg]

Sure. You can reassign them under CUCM Admin > User Management > Assign Presence Users.

[Bekzod Fakhriddinov]

sorry looks like my mistake : I just noticed site A is publisher and site B is subscriber. Now I am thinking does it makes sense to do this ?

[Bekzod Fakhriddinov]

sorry looks like my mistake : I just noticed site A is publisher and site B is subscriber. Now I am thinking does it makes sense to do this ?  What do u think Jonathan can i have users reloaded and assigned on Sub IM&P from PUB  and active on it ?

[Jonathan Schulenberg]

I'm a little unclear what you're trying to accomplish here. If it's just a regularly scheduled restart the servers every six months because of an outdated Windows-centric, non-ECC memory SOP then just restart the IM&P node and fallback afterward either manually or automatically. Reassigning the users won't accomplish anything more than the failover and fallback would in that scenario.

Whether all users in a sub cluster are assigned to the same node or balanced between the two nodes is a simple decision of which is more important: WAN bandwidth or speed/blast radius of a failover. For example, if all of the users are assigned to the same node than the XMPP fabric does not need to forward IMs between IM&P cluster nodes. On the other hand, balancing the users between the two sub cluster nodes cuts the number of users that need to failover if a node fails in half. By the way, there is an Enterprise Parameter User Assignment Mode for Presence Server on IM&P Administration to set this behavior automatically when new users are synced in from LDAP.

[Bekzod Fakhriddinov]

it's not regularly scheduled restart the servers every six months , its regularly scheduled switch over  the servers every six months between 2 sites. it's common thing when u have active/backup servers on different sites to test links,reach-ability, servers/apps. 

I found out my site A is Pub and site B is Sub , so I think does it makes sense to fail over to site B IM&P ...

Another question is: will subscriber IM&P provide service If we loose PUB IM&P server ?

Thank you

[Jonathan Schulenberg]

will subscriber IM&P provide service If we loose PUB IM&P server ?

Yes, assuming it is in the same sub cluster as the publisher.

[Bekzod Fakhriddinov]

Yes, I think we have 1 cluster with distributed over IP WAN nodes.

I see with 10.5 version CUCM is a Pub for IM&P servers too, all servers added under Publisher. . Does it mean IM&P CUCM PUB  keeps database of IM&P servers  too ?  

But I also see site A IM&P server as DBPub when I do "show network cluster", a little bit confusing. 

Thank you