Phones showing unregistered in different CUCM susbcribers.

[catkins101]

I have been receiving Solarwinds alerts that some servers in my CUCM cluster have unregistered phones over 5%. All devices should be registering with the main subscriber. However, there are a handful of devices that are registered correctly with the main subscriber but they also show entries in one or two of the other servers in the cluster. This leads me to think that these few devices are bouncing between servers when they should always be registered to the main subscriber. I checked the servers with show risdb query phone and see that these devices are unregistered. One of the devices shows unregistered in two different servers while maintaining a registered status with the main subscriber (where it should). Any ideas what would cause this and how to correct it? There are a couple of CSF/Jabber devices, a couple of 8831 devices and a 7841.

Thanks All!

[Andrew Skelly]

Have you verified the state of cluster replication?  You can check that via RTMT or you can go in to CLI and type in "utils dbreplication runtimestate" and see if the cluster is properly replicated.  If not you may need to issue the "utils dbreplication rebuild" command (suggested to perform after hours or during a maintenance window).

[catkins101]

I ran utils dbreplication status and checked the replication. No errors or Mismatches found. From what I have read the RISdb entries are purged of unregistered entries after 3 days. I am not sure if this is accurate, but if these devices are registered correctly with the main subscriber then at some point maybe they lost connection and registered briefly with one of the other servers. Is there a place to see how long the device was registered? I wonder if they will clear out in he coming days.

[Andrew Skelly]

That is correct.  A phone will show Unregistered for 72 hours.  After that it is purged from the database and will show a status of None.  I am curious where you are seeing Phone A registered and where you see it unregistered.  The publisher is the main database and replicates to the subscribers.  If you are seeing Phone A registered to Sub 1 in the publisher and sub 1, but when logging in to sub 2 it shows as unregistered it would seem to me that there is a database replication issue.  Sub 2 should not be showing anything different than the other subs or pub.

[catkins101]

Maybe I explained it incorrectly. All servers (Pub and 2 subs) show the devices as registered to the main Subscriber. They are correctly registered to the proper subscriber at this point in time. However, the alert from Solarwinds led me to look at why any devices would have been registered to any server other than the main subscriber. So I checked the publisher and subscriber that were showing the alerts to see which devices were registering with those. When I ran the show risdb command there were 7 devices in the Pub and 1 device in the sub (the sub that shouldn't have any). The device in the sub that shouldn't have any devices showed in both the Pub and that sub. So I assume that 6 of the devices registered with the Pub because they lost connectivity to the main sub. And the 7th device registered to the Pub and then to the other sub. All devices have since registered back to the main sub where they should be. I just don't know why those devices would have had to register to the Pub and other sub at all. Now that I know the database will clear them out in 3 days I can check it again in a few days to see if those devices are still in there. When I run show risdb query phone will those devices still show up?

[Andrew Skelly · Accepted solution]

Apologies, I missed where you said you were in CLI for each server.  Short answer is yes, these will disappear from the DB after 3 days....assuming you haven't change the default value.  You can confirm this in CM Administration under the service parameter for Cisco RIS Data Collector.  You want to look for RIS Unused Cisco CallManager Device Store Period

This parameter specifies the RIS database information storage period for any unregistered or rejected device information from the Cisco CallManager service. After the time specified in this parameter expires, Cisco CallManager removes the expired entries during the next RIS database cleanup time (specified in the RIS Cleanup Time of the Day parameter)
This is a required field.
Default: 3
Minimum: 1
Maximum: 30
Unit: day

As far as why those phones were bouncing around between servers (your publisher should not have the Call Manager service running on it), is there anything unique about them, such as all in the same device pool (which is different from the other phones), same site, subnet, etc.?

[catkins101]

Thank you for your assistance. I pulled this query again and see some of the phones have now cleared out, but there is one that was making several attempts at one server (#regAttempts). I may try rebooting the device to see if it clears.