[asterisk-bugs] [Asterisk 0012775]: Manager lock (?)

[noreply at bugs.digium.com]

The following issue has been CLOSED 
====================================================================== 
http://bugs.digium.com/view.php?id=12775 
====================================================================== 
Reported By:                apsaras
Assigned To:                
====================================================================== 
Project:                    Asterisk
Issue ID:                   12775
Category:                   Core/ManagerInterface
Reproducibility:            unable to reproduce
Severity:                   minor
Priority:                   normal
Status:                     closed
Asterisk Version:           1.6.0-beta9 
SVN Branch (only for SVN checkouts, not tarball releases): N/A 
SVN Revision (number only!):  
Disclaimer on File?:        N/A 
Request Review:              
Resolution:                 unable to reproduce
Fixed in Version:           
====================================================================== 
Date Submitted:             06-02-2008 17:39 CDT
Last Modified:              06-19-2008 12:35 CDT
====================================================================== 
Summary:                    Manager lock (?)
Description: 
Our app recently experienced a strange problem while communicating with an
Asterisk 1.4.19.2 (via a manager session). Apparently there was a long
pause (106”) in the manager session. After we issued a redirect action, a
response event was sent back. After exhaustive debugging (using detailed
log information produced by our app and a memory dump) we figured out that
the first line of the response (Response: Success) came in instantly, but
the event text was completed after 106sec (when reception is complete we
set a timestamp in the receiving structure). This is the actual response
event text:

Response: Success
ActionID: 18af6ce4-e873-4923-9602-c149ef9f08b8
Message: Redirect successful

The redirect command was executed asynchronously (hence the actionid).
Network is not an issue here (we received a lot of FastAGI sessions from
the same asterisk server in the meantime). The manager connection was not
dropped. In fact, after the response event was completed, all subsequent
events were received immediately, and the session resumed flawlessly
afterwards. 

Our conclusion is that the delay occurred while reading the lines of the
response event. We suspect that there was a lock in astman_append
(manager.c) causing the delay. Since the problem is not reproducible, we do
not have a way to provide more info (to our knowledge, the error has
occurred only once among many installations of the same environment). Also,
since the manager connection is not responding while the problem occurs, it
is not possible to automatically execute a core show locks command.
====================================================================== 

---------------------------------------------------------------------- 
 Corydon76 - 06-19-08 12:35  
---------------------------------------------------------------------- 
Without more specific debugging information, we are unable to proceed.  The
fact that you're also unable to reproduce this problem makes me wonder if
perhaps a network component failed for 106 seconds, then when services were
restored, the TCP connection retransmitted what it was unable to send
before. 

Issue History 
Date Modified   Username       Field                    Change               
====================================================================== 
06-19-08 12:35  Corydon76      Note Added: 0088949                          
06-19-08 12:35  Corydon76      Status                   new => closed       
06-19-08 12:35  Corydon76      Resolution               open => unable to
reproduce
======================================================================