RTMP push stops working after WAN/Internet provider change until MistServer restart

[alexmher]

I am using MistServer to receive a stream via UDP multicast and push it via RTMP to another server. Everything works normally under stable network conditions.

However, when my Linux server’s Internet provider (WAN connection) changes — for example, when the SD-WAN rule switches to a different provider — all RTMP pushes stop working.

The MistServer interface still shows that pushes are active and there is network traffic, but the remote RTMP server receives nothing.

Restarting MistServer completely fixes the issue, but I cannot restart the service every time the provider changes.

Steps to Reproduce:

Configure MistServer to receive a stream via UDP multicast.

Set an RTMP push to another server.

Use SD-WAN or routing rules to switch the Internet provider (change default route).

Observe that RTMP pushes stop working.

Expected Behavior:

RTMP pushes should automatically recover or reconnect after the network provider changes.

Actual Behavior:

RTMP pushes remain in an “active” state but do not send any real traffic until MistServer is restarted.

MistServer version - 3.9

[Rokamun]

Heya thanks for the report!

Ah... When connection is lost like this the TCP connection itself doesn't even know it's timed out, so it probably times out using system defaults and I wouldn't be surprised this takes a full day.

Yeah, we'll have to make a few adjustments to fix this scenario, I'll see if we can add this in a future patch or 3.10 at latest and I'll give you a ping if the development builds have a fix available.