[asterisk-bugs] [JIRA] (ASTERISK-30445) stasis: Failure to allocate memory causes crashes

[Joshua C. Colp (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-30445?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=261789#comment-261789 ] 

Joshua C. Colp commented on ASTERISK-30445:
-------------------------------------------

The backtraces just show the trying to publish a message failed, due to the subscriber being NULL when it shouldn't be. It doesn't show how things got to that state and with the available information would most likely be impossible to determine. Providing all of the logging around the assertions MAY provide further insight.

> stasis: Failure to allocate memory causes crashes
> -------------------------------------------------
>
>                 Key: ASTERISK-30445
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-30445
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Resources/res_stasis
>    Affects Versions: 18.16.0, 20.1.0
>         Environment: CentOS 7, kernel 3.10.0-1160.83.1.el7.x86_64.
>            Reporter: Quality Unit
>            Severity: Major
>         Attachments: asterisk_dump_logs_2023-04-17-10_core.1913.zip, asterisk_dump_logs_2023-04-17_core.18037.zip, asterisk_dump_logs_2023-04-17_core.26817.zip, core_asterisk.zip, module_show_output&logs_before_incident.txt.zip
>
>
> This started to happen after upgrade to `18.16.0` from `16.29.0`. Asterisk starting to spam logs with similar messages like:
> {code:none}
> [2023-02-24 09:30:57.001] ERROR[11361] stasis.c: FRACK!, Failed assertion bad magic number 0x0 for object 0x7f7ec4031d10 (0)
> [2023-02-24 09:30:57.006] ERROR[11361] : Got 14 backtrace records
> # 0: /usr/sbin/asterisk() [0x459ae6]
> # 1: /usr/sbin/asterisk(__ao2_ref+0x79) [0x459fc9]
> # 2: /usr/sbin/asterisk(stasis_forward_all+0x83) [0x555283]
> # 3: /usr/lib64/asterisk/modules/res_stasis.so(+0x107e1) [0x7f7f62a7e7e1]
> # 4: /usr/lib64/asterisk/modules/res_stasis.so(+0x14e2a) [0x7f7f62a82e2a]
> # 5: /usr/lib64/asterisk/modules/res_stasis.so(+0x12966) [0x7f7f62a80966]
> # 6: /usr/lib64/asterisk/modules/res_stasis.so(+0x1554e) [0x7f7f62a8354e]
> # 7: /usr/lib64/asterisk/modules/res_stasis.so(stasis_app_exec+0x569) [0x7f7f62a79789]
> # 8: /usr/lib64/asterisk/modules/app_stasis.so(+0xbd7) [0x7f7f4d6f4bd7]
> # 9: /usr/sbin/asterisk(pbx_exec+0xb9) [0x50db89]
> #10: /usr/lib64/asterisk/modules/res_ari_channels.so(+0xa528) [0x7f7f48594528]
> #11: /usr/sbin/asterisk() [0x57e80d]
> #12: /lib64/libpthread.so.0(+0x7ea5) [0x7f7fb932fea5]
> #13: /lib64/libc.so.6(clone+0x6d) [0x7f7fb88ff98d]
> {code}
> After around ~3 minutes of these (Asterisk during that time is already unresponsive), we have final errors before service crash:
> {code:none}
> [2023-02-24 09:32:48.153] ERROR[11361] astmm.c: Memory Allocation Failure in function topic_add_subscription at line 1214 of stasis.c
> {code}
> And, from `/var/log/messages`:
> {code:none}
> Feb 24 09:32:48 localhost kernel: asterisk[11361]: segfault at 1 ip 0000000000552827 sp 00007f7ed3e0a6b0 error 4 in asterisk[400000+2a2000]
> {code}
> It happens randomly, once, twice a day, sometime doesn't happen at all.
> Though, it's always happening when Asterisk is processing calls, so it's not happening during night (when asterisk is almost idle)



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