Open marcbrevoort-cyberhive opened 7 months ago
When this occurs, the following can be seen in the logs
121625 2024-03-07 08:52:05.834850 192.168.0.100 18.132.207.60 STUN 198 Channel-Bind Request XOR-PEER-ADDRESS: 90.206.45.61:18025 ChannelNumber=0x4001 user: 1709887601:74134842-07f3-4d16-a5c3-33ea4c52ab03 realm: eturnal.net with nonce
121626 2024-03-07 08:52:05.857553 18.132.207.60 192.168.0.100 STUN 126 Channel-Bind Error Response realm: eturnal.net with nonce error-code: 438 (Stale Nonce) Stale Nonce
and there is no effort made to recover.
The client should retry, using the NONCE provided in the response (as per RFC5389).
As it says in the title, channels appear to time out after some time of inactivity (~15 minutes). Which is probably fine.
However, when traffic occurs again later on, this always occurs using SendIndication/DataIndication (i.e. plain permissions based packets) and another channel is never re-established.
While data can still flow between two peers without a channel, it is less efficient.