Closed CyBHFal closed 3 years ago
Hmm, no, I'll check it out. Thanks for letting me know!
-Max
On Tue., Jun. 30, 2020, 11:02 a.m. CyBHFal, notifications@github.com wrote:
Hi,
In the following use case, the N+1 (RTCP port) is never closed. I try to correct the problem but i do not find where socket close is (or should be) done.
Version: 2.1.0 on Linux x64
To reproduce:
rtpproxy -fF -d DBUG -T 300
echo "U test0 1.2.3.4 1111 from-tag0" | socat - UNIX:/var/run/rtpproxy.sock 61092
netstat -anu Proto Recv-Q Send-Q Adresse locale Adresse distante Etat udp 0 0 0.0.0.0:61092 0.0.0.0: udp 0 0 0.0.0.0:61093 0.0.0.0:
echo "D test0 from-tag0" | socat - UNIX:/var/run/rtpproxy.sock 0
netstat -anu Proto Recv-Q Send-Q Adresse locale Adresse distante Etat *udp 0 0 0.0.0.0:61093 http://0.0.0.0:61093 0.0.0.0 http://0.0.0.0:** <--- SHOULD BE CLOSED
Doing this 10 times give 10 udp ports not release, each time the N+1 port.
Is this a known problem ?
Regards.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sippy/rtpproxy/issues/111, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABVMJW62RL2ZCQ5I4DYWPTRZISDJANCNFSM4OMQK4RA .
Hi CyBHFal,
Thanks for reporting, after further investigation I have figured out it's some minor issue already fixed in master branch. I call it a minor issue, since it's not really a leak, but rather some misbehavior in quite rare condition that would keep RTCP ports open indefinitely as long as proxy has no fully connected open RTP sessions. As long as one such session appears, all such ports that have been piled up would be freed at once. For tat reason nobody really stumbled upon such issue before, since it requires proxy to stay basically idle.
In any case I've merged relevant fix from master into the rtpp_2_1, so it should do the right thing even in this somewhat unusual scenario.
I am closing the ticket now, please check and feel free to let me know if the issue persists.
Thanks again and have a good weekend!
Hi,
In the following use case, the N+1 (RTCP port) is never closed. I try to correct the problem but i do not find where socket close is (or should be) done.
Version: 2.1.0 on Linux x64
To reproduce:
rtpproxy -fF -d DBUG -T 300
Doing this 10 times give 10 udp ports not release, each time the N+1 port.
Is this a known problem ?
Regards.