[asterisk-bugs] [JIRA] (ASTERISK-27115) Deadlock? db_sync_thread at db.c

[Joshua Colp (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-27115?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=237668#comment-237668 ] 

Joshua Colp commented on ASTERISK-27115:
----------------------------------------

Yes, your backtrace shows the same:

{noformat}
Thread 119 (Thread 0x7f11da83d700 (LWP 1157)):
#0  0x000000324380aa86 in pthread_rwlock_rdlock () from /lib64/libpthread.so.0
#1  0x000000000056ebcb in ast_rtp_codecs_payload_code (codecs=0x7f121b118808, asterisk_format=1, format=0x15b97e8, code=0) at rtp_engine.c:937
#2  0x00007f11dd5db932 in bridge_p2p_rtp_write (instance=0x7f11b73075e0, rtcp=<value optimized out>) at res_rtp_asterisk.c:4455
#3  ast_rtp_read (instance=0x7f11b73075e0, rtcp=<value optimized out>) at res_rtp_asterisk.c:4711
#4  0x00007f11dcdb6429 in chan_pjsip_read (ast=0x7f11afdbaa40) at chan_pjsip.c:718
#5  0x00000000004b854d in __ast_read (chan=0x7f11afdbaa40, dropaudio=0) at channel.c:3952
#6  0x000000000047a1cb in bridge_handle_trip (bridge_channel=0x7f11bf348540) at bridge_channel.c:2416
{noformat}

That holds a channel lock and cascades to cause problems for other threads.

> Deadlock? db_sync_thread at db.c
> --------------------------------
>
>                 Key: ASTERISK-27115
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-27115
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>    Affects Versions: 13.15.0
>         Environment: Cettos 6.8 Final
>            Reporter: Ryan Smith
>            Assignee: Unassigned
>         Attachments: core-asterisk-running-2017-07-07T15-44-43+0100-brief.txt, core-asterisk-running-2017-07-07T15-44-43+0100-full.txt, core-asterisk-running-2017-07-07T15-44-43+0100-locks.txt, core-asterisk-running-2017-07-07T15-44-43+0100-thread1.txt
>
>
> Asterisk/PJSIP simply stops processing any further SIP packets. Incoming options "ping" packets from proxy see a task-processor limit error related to pjsip/distributor (live clients have been redirected by fail over by time of requesting a core dump) Unfortunately debug log files have been deleted until the next occurrence.
> Asterisk will often seq fault on requesting the core dump.
> {quote}
> Processor                                      Processed   In Queue  Max Depth  Low water High water
> pjsip/distributor-00000cda                        256958        292          8        450        500
> pjsip/distributor-00000cdb                        254784          0          7        450        500
> pjsip/distributor-00000cdc                        271315        251          6        450        500
> pjsip/distributor-00000cdd                        252324        551          8        450        500
> pjsip/distributor-00000cde                        252965        162          4        450        500
> pjsip/distributor-00000cdf                        265077        214          8        450        500
> pjsip/distributor-00000ce0                        265003        329          6        450        500
> pjsip/distributor-00000ce1                        250271        346          8        450        500
> pjsip/distributor-00000ce2                        251438        481          5        450        500
> pjsip/distributor-00000ce3                        251036         28          8        450        500
> {quote}



--
This message was sent by Atlassian JIRA
(v6.2#6252)