If a node fails, stable calls (those in conversation state) are preserved but unstable calls may be dropped.
Two parties will not be able to gracefully release a call established via node 1 after a switchover to node 2. The call context remains "hang-up", because the BYE message is not redirected to node 2.
If the failing node is the last active node in the system, all ongoing calls and their related billing data are lost.
If node 2 cannot reach the remote access connection of node 1, node 2 consults the survival authority and receives the takeover command back to trigger a failover of the IP addresses of node 1 to node 2.
Parent topic: Cluster Redundancy