[Home]

Summary:ASTERISK-07929: RTP listener not getting cleaned up upon call teardown
Reporter:jerjer (jerjer)Labels:
Date Opened:2006-10-14 02:00:13Date Closed:2006-11-16 14:49:13.000-0600
Priority:MajorRegression?No
Status:Closed/CompleteComponents:Core/RTP
Versions:Frequency of
Occurrence
Related
Issues:
Environment:Attachments:
Description:This machine, running 1.2.12.1, has 18 active SIP channels but asterisk is listening on way more than 18 rtp ports:

http://pastebin.ca/202284

Blitzrage confirmed this behavior on his box, same Asterisk version:

http://pastebin.ca/202281

****** ADDITIONAL INFORMATION ******

Only recent relevant change I can find is


URL: http://svn.digium.com/view/asterisk?rev=44213&view=rev
Log:
Change the fd on the I/O context in case it changed during the reload, which is indeed possible. (issue ASTERISK-7733 reported by eclubb)


Does this fix this issue? If so, should a new release be made?

Comments:By: jerjer (jerjer) 2006-10-14 02:03:35

Juggie reported in IRC that his v1.2.9.1 does not experience this situation, with uptime of 4 weeks.

By: Serge Vecher (serge-v) 2006-10-23 08:19:03

well, that patch is in 1.2.13 ... Any luck with testing that release?

By: Olle Johansson (oej) 2006-10-29 13:57:53.000-0600

Any updates?

By: Olle Johansson (oej) 2006-10-29 14:03:46.000-0600

We need to listen on RTCP ports as well, you know.

Interesting is that RTP and RTCP is on different bindings. RTP on all IPs, RTCP only on the hostname. We must be binding differently. I see 13 RTP streams here - and you had 18 calls?

By: Olle Johansson (oej) 2006-10-29 14:22:06.000-0600

Fixed RTCP bindings in svn trunk. Did not see it as a bug that needed to be fixed in 1.2/1.4, but anyway...

By: Joshua C. Colp (jcolp) 2006-11-16 13:06:19.000-0600

If this is still an issue then giving oej or myself access to a box where it is happening will allow us to track it down better.

By: jerjer (jerjer) 2006-11-16 14:49:12.000-0600

Unable to duplicate this situation on 1.2.13 - assume situation resolved.