There are some possible reasons that this problem occurs:
There can be replication issues between Cisco CallManager databases. In order to check this, run the DBLHelper app to check good replication. This is ideally located in c:\program files\cisco\bin\.
If the phones unregister across an Internet connection, this behavior can be expected once in a while. IP telephony design recommends end-to-end QoS configuration, which is something that is out of your control if you use the Internet.
The Spanning Tree Protocol (STP) is not configured on the switch so there are loops.
Cisco CallManager misses keepalives, therefore phones unregister.
The phone loads.
In order to resolve this issue, try to increase the keepalive timer on the Cisco CallManager in order to rule out a keepalive timeout issue. Go to the ccmadmin web page and under Service > Service Parameters, choose the server(s) that the IP phones are registered with. In Service*, choose Cisco CallManager. Change the value of Station Keep Alive Interval (sec)* to 45.
