Summary: | ASTERISK-22994: pjsip/incoming_calls_without_auth: test bounces | ||||
Reporter: | Matt Jordan (mjordan) | Labels: | |||
Date Opened: | 2013-12-15 17:58:30.000-0600 | Date Closed: | 2014-03-20 10:51:17 | ||
Priority: | Major | Regression? | |||
Status: | Closed/Complete | Components: | Channels/chan_pjsip Tests/testsuite | ||
Versions: | 12.0.0-beta2 | Frequency of Occurrence | |||
Related Issues: |
| ||||
Environment: | Attachments: | ( 0) backtrace_4944.txt ( 1) full.txt.3 ( 2) incoming_calls_without_auth.log | |||
Description: | This test continues to bounce as well:
https://bamboo.asterisk.org/bamboo/browse/AST-ATTSCD4-C632TE-145/test/case/1415088 Note that due to the high number of SIPp scenarios in this test, it is incredibly difficult to isolate what failed and when. It may be advisable to break this test up into more digestible pieces. | ||||
Comments: | By: Matt Jordan (mjordan) 2014-02-13 16:01:17.602-0600 This is not fixed :-( The test still crashes. Backtrace attached. By: Matt Jordan (mjordan) 2014-02-13 16:03:35.574-0600 Crash and logs taken from test run https://bamboo.asterisk.org/bamboo/browse/AST-ATTSCD-C664TE-296 By: Kevin Harwell (kharwell) 2014-03-20 10:51:11.816-0500 This turned out to be caused by a race condition within pjproject. In the 'pjsip_tsx_terminate' function there was a check to see if the transaction had already been terminated. If it had been then simply return. However, sometimes it was possible for the transaction to be terminated by another thread after the check, thus it would attempt to terminate the transaction twice. I sent an email to Teluu with a potential patch. In the meantime I also patched our fork with a fix (see commit 477abd1). |