Solved: Phones randomly go into registering, data drops out

gmgarrian · ‎08-16-2011

I have a site that is remote to the CUCM server. My client is reporting that all the phones onsite (8945s) will randomly (but all at the same time) go into a registering state and the data connections to their computers will drop as well. But, this does not happen if there is an active call.

I'll be on site tomorrow to troubleshoot further, but has anyone else run into this? I'm suspecting an issue with the POE switch with the computers losing connection as well.

I'm thinking maybe a power issue? The ups isn't powerful enough for the Cat 4900 series switch?

I've opened a TAC case and I'll get them the logs tomorrow but I can't see CUCM showing anything other than the phones disconnecting and re-registering.

Thoughts?

ryabenne · ‎08-16-2011

You can only get the alarm event from the phone after its registered. The event will be generated from the phone and sent to the CUCM for later review. You can pick it out of the RTMT events. Honestly they aren't that helpful and will just point you to a general network issue. The reason for this is because the phone is on one end and the CUCM on another, so they don't know any of the devices in between and don't know where the actual connection dropped out.

Quickest way to resolve the issue: Get a packet capture from the back of the phone and some other points in between right up to the CUCM server. The quicker you get this in place the quicker TAC will be able to help you in troubleshooting what the offending device is. Without those we won't be able to tell you exactly where you issue lies within the network. We would only be able to take detailed CCM traces for minutes before/during/after the issue and tell you that we didn't receive a keepalive on the CUCM so we dropped the TCP connection and un-registered the phone just like it should be done.

Ryan

View solution in original post

ryabenne · ‎08-16-2011

You can only get the alarm event from the phone after its registered. The event will be generated from the phone and sent to the CUCM for later review. You can pick it out of the RTMT events. Honestly they aren't that helpful and will just point you to a general network issue. The reason for this is because the phone is on one end and the CUCM on another, so they don't know any of the devices in between and don't know where the actual connection dropped out.

Quickest way to resolve the issue: Get a packet capture from the back of the phone and some other points in between right up to the CUCM server. The quicker you get this in place the quicker TAC will be able to help you in troubleshooting what the offending device is. Without those we won't be able to tell you exactly where you issue lies within the network. We would only be able to take detailed CCM traces for minutes before/during/after the issue and tell you that we didn't receive a keepalive on the CUCM so we dropped the TCP connection and un-registered the phone just like it should be done.

Ryan

gmgarrian · ‎08-19-2011

TAC confirmed this, it's a connection issue. I have no control over the infrastructure, just the voip hardware so I'll have to see where this goes.

Thanks for the info!