Detection of lost listener
We have a configuration with listener/sender, caller/receiver.
Everything is connecting correctly, data flows through.
Issue: If we kill the listener, the caller does not appear to give any notification to the client code.
We see the trace from the rist_timeout_check
function (rist-common.c): "Listening peer 1 timed out after 2042 ms".
That is followed by: "************** Session Timeout after 2s of no data, deleting flow with id 2395157172 ****".
If we then destroy the caller context, at that point we get the status callback (with RIST_CONNECTION_TIMED_OUT
).
If we do not destroy the caller, but bring the listener back up, the connection appears to recover.
This is against master up to commit: fe67a68d
Maybe related: #133 (closed)