Running Edge 2024.6.1 on Dell R750 nodes in Tokyo. Primary WAN link flapped during ISP maintenance, kicking the cluster into local survivability mode while the dual NIC bond stays up. BIOS C-states are disabled to keep the CPU steady, but the screen recording agent still throws a MEDIA_ROUTING_FAILURE_403 when pushing the WebRTC stream to the Edge media proxy. Console logs show the recording process dying at handshake with ICE_CANDIDATE_EXPIRED right after the failover timer hits 12 seconds, and the local media port just won’t bind.
Check the Edge config for mediaProxyTimeout. The default 30s window usually expires during those R750 failovers. Bump it to 60s in the deployment settings. Also verify the STUN servers are reachable from the local survivability subnet.
"mediaProxy": {
"handshakeTimeoutMs": 60000
}