|Summary:||ASTERISK-22015: 'core restart now' from a secondary remote console resulted in unresponsive CLI and certain commands no longer functioning|
|Reporter:||Rusty Newton (rnewton)||Labels:|
|Date Opened:||2013-07-04 20:08:20||Date Closed:||2013-07-20 15:14:33|
|Environment:||SVN-trunk-r393490 Asterisk built with: DONT_OPTIMIZE, DEBUG_THREADS, LOADABLE_MODULES, BETTER_BACKTRACES, BUILD_NATIVE 3.5.0-24-generic #37~precise1-Ubuntu SMP Thu Feb 7 13:50:53 UTC 2013 i686 i686 i386 GNU/Linux||Attachments:||( 0) backtrace-threads.txt|
( 1) backtrace-threads.txt
|Description:||Connected to Asterisk CLI via remote console
While working, connected to same Asterisk instance via another remote console over a separate SSH connection.
Executed "core restart now" on the new console. Returned to a CLI prompt with no other output. The secondary console (the one I'm on) becomes unresponsive completely. Any key press fails to result in a response.
On the original remote console, the console responds to many commands, but some are missing.
A tab tab of "core show .." shows these commands as possible:
rnewton-office-lab*CLI> core show
calls channel channels channeltype channeltypes codec codecs config help locks
taskprocessors translation uptime
We are missing "core show settings" "core show version", etc.
A "core stop now" or "core restart now" on the original console results in
Ignoring asterisk shutdown request, already in progress.
, but Asterisk does not end.
|Comments:||By: Rusty Newton (rnewton) 2013-07-04 20:09:01.997-0500|
attaching backtrace (while running) and "core show locks" output
By: Rusty Newton (rnewton) 2013-07-04 20:17:04.441-0500
Can't reproduce... tried a second time and things restarted cleanly... Something must have got in a funky state before I tried the restart on the first instance of this freeze.
By: Rusty Newton (rnewton) 2013-07-05 15:30:30.660-0500
Attaching new trace from what appears to be a very similar freeze. Unfortunately this may not be useful as I had already recompiled without DEBUG_THREADS.
This time, a "core stop now" failed to end Asterisk. CLI shortly became unresponsive. Had to kill Asterisk's process to end it.
By: Matt Jordan (mjordan) 2013-07-08 08:47:48.142-0500
This looks similar to the other reports of things hanging - we do an unsubscribe and join from the CDR engine and that blocks something in Stasis.
It'd be good to get a 'core show locks' the next time this happens
By: Thomas Sevestre (to) 2013-07-26 03:56:59.634-0500
Can you point another hanging issue? I haven't been able to find them.
By: Rusty Newton (rnewton) 2013-07-26 14:04:31.496-0500
@Thomas this has been fixed in trunk. If you are testing with Trunk, update to the latest version.
By: Thomas Sevestre (to) 2013-07-26 17:06:16.429-0500
Actually I'm using recent 1.8 versions.
Do you have a pointer to the issue or to the commit?
By: Rusty Newton (rnewton) 2013-08-06 15:51:01.513-0500
Not sure.. It doesn't show up in the Subversion tab here.
I'd recommend testing with the latest 1.8 SVN branch and if that doesn't fix your problem then it's likely a new one..
I believe Matt Jordan made the fix. I'll ping him to see if he can point out which commit it was.
By: Rusty Newton (rnewton) 2013-08-06 17:46:06.826-0500
Turns out the fix wasn't related to anything in 1.8. If you have an issue in 1.8 and you think it is a bug, please file a new issue. https://wiki.asterisk.org/wiki/display/AST/Asterisk+Issue+Guidelines