ODBC Voicemail Forward Message Crash

Asterisk will crash (hang, but AMI still seems to work) if a voicemail is forwarded to another mailbox and that mailbox doesn’t have a corresponding filesystem folder (/var/spool/asterisk/voicemail/{context}/{mailbox}). We have multiple clusters, each has local ODBC voicemail. The voicemail_users table contains a list of every mailbox, even if that box is on a different system.

Server1 has mailbox 1111 and 2222 in voicemail_users, but only 1111 exists in /var/spool/asterisk/voicemail/from-internal-sip/
2222 is on a different server.

Log Example

[May 21 07:44:06] WARNING[15227][C-00000017] app_voicemail.c: Failed to open/create '/var/spool/asterisk/voicemail/from-internal-sip/13661/greet.txt'
[May 21 07:44:06] VERBOSE[15227][C-00000017] file.c: <SIP/14974-00000026> Playing 'digits/1.g722' (language 'en')
[May 21 07:44:06] VERBOSE[15227][C-00000017] file.c: <SIP/14974-00000026> Playing 'digits/3.g722' (language 'en')
[May 21 07:44:07] VERBOSE[15227][C-00000017] file.c: <SIP/14974-00000026> Playing 'digits/6.g722' (language 'en')
[May 21 07:44:08] VERBOSE[15227][C-00000017] file.c: <SIP/14974-00000026> Playing 'digits/6.g722' (language 'en')
[May 21 07:44:09] VERBOSE[15227][C-00000017] file.c: <SIP/14974-00000026> Playing 'digits/1.g722' (language 'en')

A message is forwarded from mailbox 1111 to 2222 on Server1. Asterisk finds the mailbox in ODBC, but when it attempts to forward, Asterisk crashes and stops accepting input in the CLI. Current calls stay active, but new calls receive a busy signal. I’ve had to kill the asterisk process and restart it.

My current live version = 13.12, but I compared app_voicemail.c to 13.26 also. I didn’t see any changes that would affect this.

I created a workaround by writing a program to create the “missing” voicemail boxes locally based on what’s in the database. I’ll need to write another program to move the forwarded messages to the proper cluster.

I assume most people centralize their voicemail, so it might not affect many people. That’s not an option for us. Wasn’t sure if it needed to be a bug report or if this will suffice.

This is a useful place to triage the problem and give guidance on the information needed for a useful bug report; you currently don’t have adequate information to progress a bug report.

You will need to collect deadlock debugging information, see https://wiki.asterisk.org/wiki/display/AST/Getting+a+Backtrace#GettingaBacktrace-GettingInformationForADeadlock

You are likely to find that thread debugging and therefore core show locks are also needed.

Thanks David, I’ve posted here before and even placed bug reports, but since I had a workaround, I vacillated over taking the extra time. A few of the admins, like yourself, frequent the forums, so I thought I’d just throw it out there. Like I said, it’s doubtful that it affects many people. It could be one of those “look at in your spare time” (ha) kinda things. If I can find some time, I’ll get the backtrace and submit a bug report. I don’t think I’ll be able to show the locks as the cli ignores input when it’s hung.

I posted the bug report. Joshua suggested to test with the newer version. I did and this problem does not exist in 13.26. At the point the voicemail is forwarded, the folder is created and the message is forwarded properly.

The bug report is closed. I just wanted to update here also in case someone else has the same problem.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.