[asterisk-users] Asterisk 1.4 reliability problems

[Ben Willcox]

Hello All,

We have been experiencing some ongoing reliability problems with
Asterisk for quite some time, and I am trying to find out if anyone else
has experienced the same problems.

We are running asterisk 1.4.17~dfsg-2+b1 on Debian Lenny, with a Digium
PRI card, and have approximately 120 sip peers, mostly Snom 360s, with a
few Grandstream GXP2000 and a handful of Handytone 486 units. 

The symptoms, when they occur, are as follows:

-The inability to receive incoming calls to our ISDN PRI (callers get a
busy tone), this starts off becoming intermittent but becomes permanent.

-Asterisk cli commands work once, but then no longer return any data
until disconnecting and reconnecting to the cli, i.e. sip show peers,
show channels etc.

-Internal SIP calls stop working

-Calls remain stuck in queues, the queue members do not ring, and show
as Busy when issuing a 'queue show' command.


We've actually had these sort of problems for many months now, which
originally started when we were running Asterisk 1.2 on Gentoo. We have
done a large amount of fault finding and testing, which has involved a
replacement ISDN card, reinstall on complete different server hardware,
and changing to Asterisk 1.4 on Debian Lenny.

I believe there may be two separate issues here - we did track down one
problem to our cacti and nagios monitoring scripts, which were
connecting and disconnecting to the manager interface several times per
minute, which eventually caused asterisk to give the above symptoms,
although in addition to the above, asterisk would consume 100% cpu on
the box, and eventually need a hard-reboot of the server. I posted about
this to the list a few weeks ago, and it was confirmed that this could
cause such a problem. After stopping these services the problems were
much reduced.

However, we have now completely disabled the manager interface
(enabled=no in manager.conf), and yesterday the problem occurred again -
a restart of asterisk got everything going again.
So really I'm at a loss as to where to go from here. A colleague of mine
also has the same problem at his site running Asterisk 1.4 on Debian
Lenny, he has never used the manager interface, and has completely
different server hardware and ISDN card, so I wonder if it's a Debian
specific problem?

One option is to try reverting back to Asterisk 1.2, but that isn't
really a long-term solution. We also had major problems with 1.2 with
our Snom 360 phones, as with any Snom firmware > 6.2.2 there was a
serious problem whereby on hangup the channels were not cleared down,
meaning we had many outgoing ISDN calls held open for many hours until
we realised the problem. This problem does not occur in Asterisk 1.4,
although we have many log messages such as:

chan_sip.c: Remote host can't match request BYE to call <callid>

so I don't know if this is anything to worry about?

Any help would be gratefully received!

Thanks,
Ben