UNIStim endpoint fail over process in an N+1 and a geo-redundant model

Last Updated : Feb 03, 2020 |
Prolog information
Metadata
Keywords:
  1. The following are the two scenarios in which a UNIStim endpoint fails over:
    ADDITIONAL INFORMATION:
    • When the endpoint detects a registration failure, the UNIStim endpoint restarts and attempts to fail over.
    • A loss of service can occur during call-related signaling. For example, during an active call, the user performs an operation that modifies the media path. In this scenario, the UNIStim endpoint restart and fail over is triggered by the following events:
      • If the endpoint detects the registration failure, the endpoint restarts and fails over.
      • If the endpoint does not detect the registration failure, the endpoint restarts after Device Adapter does not respond to the service request by the endpoint.
  2. The following are the fail-over processes:
    ADDITIONAL INFORMATION:
    • In case of an N+1 deployment model:
      • If the N+1 cluster has enough capacity, the load balancer passes the endpoint registration request to the node that has the capacity to accept this registration. The registration succeeds.
      • If no capacity is available on the N+1 cluster and if an alternate geo-redundant cluster is not configured, the endpoint waits for the configured time interval and reattempts the endpoint registration.
    • In case of a geo-redundant model:
      • If the load balancer does not find the required capacity on S1, it passes the registration request to S2.
      • If S2 has the required capacity, it accepts the registration request.