Summary:ASTERISK-07856: asterisk locks, taking most of the box down, when performing attendant transfers
Reporter:Brad Barnett (bbarnett)Labels:
Date Opened:2006-10-02 14:36:15Date Closed:2006-10-03 08:50:37
Versions:Frequency of
Description:This bug does not exist in Asterisk 1.2.10, Debian version 1.2.10.dfsg, but does exist in Debian version 1.2.11.dfsg-1 and  This bug also exhibits itself more frequently, if one uses a longer blind transfer or attendant transfer series of digits, for example **55 for attendant transfer.  None the less, the bug will exhibit itself using ** for attendant transfer, as well.

A number of transfer can be done, I've had as many as twenty occur, as few as two before the lockup occurs, but it always does.  I hooked up a serial console, because sometimes most of the time the machine locks solid (does not respond to any pings, dead keyboard, etc).  There was no output from the serial console, and I suspect that wctdm must be involved, as it would likely take a kernel module to kill the machine in such a way.

I am limited in the testing I can do, although I will be on site to work on echo cancellation at a later date.  I can likely perform additional down time tests if it helps.  Unfortunately, I am unable to get any logging information out of the box when it happens.  :/

All transfers are being performed over SIP phones, via local lan.  There are three zap channels for incoming lines, and a forth disconnected (waiting for an expensive analog speaker phone to be connected).  TDM400P card.  Debian pre-compiled kernel 2.6.16-2-686.

Comments:By: Serge Vecher (serge-v) 2006-10-03 08:50:17

bbarnett: this has been fixed recently in the 1.2 branch. Please checkout out the 1.2 branch from SVN or wait for the 1.2.13 release. Thanks.