After a recent upgrade on my SME head end cluster from CUCM 126.96.36.19900-12 to 10.5.2.11005-1 I have been having an issue with my call center agents on a leaf cluster running 188.8.131.52900-7 (Cluster 1). This is both enterprise and express agents.
The call comes in on a leaf cluster running CUCM 184.108.40.20600-12 (Cluster 2)
Then goes to a head end SME CUCM 10.5.2.11005-1 cluster via a SIP trunk
Then the calls goes to a CUCM 220.127.116.11900-7 (Cluster 1) leaf cluster via an Inter-Cluster Trunk
The call center agents get put in reserved status but never get the call.
The callers just hear dead air.
The only changes is the upgrade to the head end cluster to 10.5.2.11005-1.
If I add a dial peer on the PSTN gateway CUCM 18.104.22.16800-12 (Cluster 2) leaf cluster so the call goes from gateway to CUCM 22.214.171.124900-7 (Cluster 1) the agent does not have any issues.
My initial thought is a possible change on the trunks from Cluster 2 to SME or SME to Cluster 1.
Anyone have any thoughts on this?
Not sure if this is related since this was a while back and it was on version 8.5 that I noticed this. If I recall there is was an issue with the SME cluster talking to the leaf cluster. If the SIP Trunk was talking to a specific node in the cluster, the intra cluster signaling had some problems. i.e. if you had Sub-A and Sub-B and if the SME Cluster sent the invite to Sub-A while the phone sat on Sub-B, the call wouldn't go through. It had to do with making sure that the Call Manager Group assigned to the device pool that was assigned to trunk had all CCM nodes in the cluster added to it. This became an irritation when I was working on it because we had more than 3 CCM Subscribers and there is a limit of 3 CCM nodes per Call Manager group and we had 4 CCM subs in the cluster which means we would never be able to send calls to a phone in the leaf cluster registered to the 4th node that wasn't part of the CCM Group for the device pool used by the SIP trunk between the Leaf and SME cluster.
You are dealing with 10.5 and you also mention that this was working prior, so I am not sure if this is the same, but I would certainly play around with the CMG for the device pool of the trunk and the CM that the phone registers to.
This turned out to be a networking issue. At the same time I was doing my CUCM upgrade the network team for my company had some network issues. A core switch in the data center that connected all the 10 gig man sites was consistently running at 90%. Once the traffic was rebalanced and the utilization of this switch returned to 15% I no longer had any reports of the issue.