Summary:ASTERISK-18604: Constant Lockups throughout the day
Reporter:Nate Slinin (nslinin)Labels:
Date Opened:2011-09-21 19:41:03Date Closed:2011-09-22 06:48:17
Versions: Frequency of
Environment:Attachments:( 0) backtrace.txt
( 1) backtrace2.txt
Description:I recently upgraded to asterisk and applied the queuelog patch. Currently I am experiencing a number of lockups throughout the day. I finally recompiled with Don't optimize and debug threads so I could get a bt for the tracker. There was no core dump. The log shows this before the lockup:

[Sep 21 20:05:31] WARNING[32074] chan_sip.c: sip_xmit of 0x7f81e0164870 (len 638) to returned -2: Interrupted system call
[Sep 21 20:05:32] WARNING[32074] chan_sip.c: Unsupported SDP media type in offer: video 0 RTP/AVP 99
[Sep 21 20:05:32] VERBOSE[17470] app_dial.c:     -- SIP/100032-00000316 answered Local/184@from-queue-d695;2
[Sep 21 20:05:32] VERBOSE[17470] pbx.c:     -- Executing [s@macro-auto-blkvm:1] Set("SIP/100032-00000316", "__MACRO_RESULT=") in new stack
[Sep 21 20:05:32] VERBOSE[17470] pbx.c:     -- Executing [s@macro-auto-blkvm:2] NoOp("SIP/100032-00000316", "Deleting: BLKVM/0000005/SIP/Transbeam-00000314 TRUE") in new stack
[Sep 21 20:05:32] VERBOSE[17432] app_queue.c:     -- Local/184@from-queue-d695;1 answered SIP/Transbeam-00000314
[Sep 21 20:05:32] VERBOSE[17432] res_musiconhold.c:     -- Stopped music on hold on SIP/Transbeam-00000314
[Sep 21 20:05:32] VERBOSE[17479] app_mixmonitor.c:   == Begin MixMonitor Recording SIP/Transbeam-00000314
[Sep 21 20:05:32] WARNING[17470] dsp.c: Inband DTMF is not supported on codec g722. Use RFC2833

Also I just noticed that the timeout for parked calls is going into negative seconds.

CTG-PBX*CLI> parkedcalls show
Num                   Channel (Context         Extension    Pri ) Timeout
*** Parking lot: default (1)
10         Local/1002@from-queue-d5b8;2 (park-dial       SIP_100055   1   )    -23s
1 parked call in total.

then some seconds later

 == Timeout for Local/1002@from-queue-d5b8;2 parked on 10 (default). Returning to park-dial,SIP_100055,1
   -- Executing [SIP_100055@park-dial:1] Dial("Local/1002@from-queue-d5b8;2", "SIP/100055,30") in new stack
 == Using SIP RTP TOS bits 184
 == Using SIP RTP CoS mark 5
   -- Called SIP/100055
   -- SIP/100055-0000005a is ringing
   -- Local/1002@from-queue-d5b8;2 requested special control 20, passing it to SIP/100055-0000005a
   -- Local/1002@from-queue-d5b8;2 requested special control 20, passing it to SIP/100055-0000005a
Comments:By: Nate Slinin (nslinin) 2011-09-21 19:47:33.418-0500

This is a 2nd backtrace from a lockup that occurred slightly after restarting asterisk.

By: Leif Madsen (lmadsen) 2011-09-22 06:46:35.606-0500

It looks like you're using res_timing_pthread -- that is likely the problem you're having. If using or later, you can use res_timing_timerfd, otherwise, you'll need to use res_timing_dahdi.

By: Nate Slinin (nslinin) 2011-09-22 06:47:59.983-0500

we don't use dadhi on this machine.

By: Leif Madsen (lmadsen) 2011-09-22 06:48:17.237-0500

Closing this as a known issue.

By: Nate Slinin (nslinin) 2011-09-22 06:54:53.354-0500

That is not the issue. After doing a noload on res_timing_pthread, there is no effect. The system is locking up on reload every 5 minutes now.

By: Nate Slinin (nslinin) 2011-09-22 06:57:24.446-0500

Also queue agents are reporting receiving multiple calls at once.

By: Gregory Hinton Nietsky (irroot) 2011-09-22 10:10:58.566-0500

Can see a problem not sure its timing fd related but will need output of core show locks and do try 1.8.7-rc2