Fail-over scenarios for UNIStim endpoints during the upgrade

Last Updated : Feb 27, 2020 |
Prolog information

UNIStim endpoint fail over when a registration failure is detected

UNIStim endpoints monitor the registration by using the “time to live” monitoring, but health checks in the registration allow the endpoints to detect a registration failure faster. When an endpoint detects a registration failure, the endpoint restarts and attempts a re-registration.
A UNIStim endpoint may detect registration failure. In an event when a registration failure occurs and the call is still active, Device Adapter and the UNIStim endpoint maintain the media path to avoid loss of call. But, a user operation during the call, such as call modification or call release, might result in the call being lost.
If the user is on a call, Device Adapter and the UNIStim endpoint maintain the media path. However:
  • If the user tries to modify the media path during an active call and if the endpoint detects the registration failure, the endpoint restarts and attempts a re-registration.
  • If the user tries to modify the media path during an active call and if the endpoint has still not detected the registration failure, the endpoint restarts only after Device Adapter fails to respond to a service request made by the endpoint. The endpoint then attempts a re-registration.
Depending on the deployment model, if the endpoint cannot reregister to the same node or cluster, the endpoint tries to fail over to an alternate node or cluster that has the required capacity.