[asterisk-bugs] [JIRA] (ASTERISK-29609) Subsequent 'ael reload' will cause a lock up

[Mark Murawski (JIRA)]

     [ https://issues.asterisk.org/jira/browse/ASTERISK-29609?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Mark Murawski updated ASTERISK-29609:
-------------------------------------

    Description: 
Requirements for reproduction of bug:
- Have a 'decent amount' of ael.  In this case, I have 650 lines of AEL in variety of files.

Obviously not a 'normal' use case to issue 10 reloads at once, but when multiple scripts/systems might be asking Asterisk to reload at the same time, this could be triggered.

5 to 10 reloads should do it!
{code}
vbox-markm-x64 {/etc/asterisk/ael} root# for i in `seq 1 10`; do asterisk -rx "ael reload" &  done
[1] 12335
[2] 12336
[3] 12337
[4] 12338
[5] 12339
[6] 12340
[7] 12341
[8] 12342
[9] 12343
[10] 12344
vbox-markm-x64 {/etc/asterisk/ael} root# asterisk -rx "core show locks" | grep "Thread ID" | wc -l
12
{code}

Depending what else is running at the time (perhaps pjsip adding hints to dialplan).  This will escallate quickly and lead to a very locked up Asterisk

The fix is to address the missing reload mutex in pbx_ael.  

  was:
Requirements for reproduction of bug:
- Have a 'decent amount' of ael.  In this case, I have 650 lines of AEL in variety of files.

Obviously not a 'normal' use case to issue 10 reloads at once, but when multiple scripts/systems might be asking Asterisk to reload at the same time, this could be triggered.

5 to 10 reloads should do it!
vbox-markm-x64 {/etc/asterisk/ael} root# for i in `seq 1 10`; do asterisk -rx "ael reload" &  done
[1] 12335
[2] 12336
[3] 12337
[4] 12338
[5] 12339
[6] 12340
[7] 12341
[8] 12342
[9] 12343
[10] 12344
vbox-markm-x64 {/etc/asterisk/ael} root# asterisk -rx "core show locks" | grep "Thread ID" | wc -l
12

Depending what else is running at the time (perhaps pjsip adding hints to dialplan).  This will escallate quickly and lead to a very locked up Asterisk

The fix is to address the missing reload mutex in pbx_ael.  


> Subsequent 'ael reload' will cause a lock up
> --------------------------------------------
>
>                 Key: ASTERISK-29609
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-29609
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: PBX/pbx_ael
>    Affects Versions: GIT, 16.15.1, 17.9.3, 18.6.0
>            Reporter: Mark Murawski
>            Severity: Major
>
> Requirements for reproduction of bug:
> - Have a 'decent amount' of ael.  In this case, I have 650 lines of AEL in variety of files.
> Obviously not a 'normal' use case to issue 10 reloads at once, but when multiple scripts/systems might be asking Asterisk to reload at the same time, this could be triggered.
> 5 to 10 reloads should do it!
> {code}
> vbox-markm-x64 {/etc/asterisk/ael} root# for i in `seq 1 10`; do asterisk -rx "ael reload" &  done
> [1] 12335
> [2] 12336
> [3] 12337
> [4] 12338
> [5] 12339
> [6] 12340
> [7] 12341
> [8] 12342
> [9] 12343
> [10] 12344
> vbox-markm-x64 {/etc/asterisk/ael} root# asterisk -rx "core show locks" | grep "Thread ID" | wc -l
> 12
> {code}
> Depending what else is running at the time (perhaps pjsip adding hints to dialplan).  This will escallate quickly and lead to a very locked up Asterisk
> The fix is to address the missing reload mutex in pbx_ael.  



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