asterisk users - Nov 2005 - 1.2.0 PRI dropping calls occasionally...

After upgrading to 1.2.0 (from a three-week-prior CVS version), I've
suddenly had people starting to complain of lost calls. They'd be there,
and suddenly they'd drop out - they could be in a conversation, or more
often, the caller would be on old, and suddenly the light would go out,
and the caller would be gone.

I haven't noticed anything unusual in the logs, nothing that stands out
anyway. Is anyone else experiencing anything like this?

--Rob

On 12/1/05, Rob Thomas <rob@wpm4l.com> wrote:
> After upgrading to 1.2.0 (from a three-week-prior CVS version), I've
> suddenly had people starting to complain of lost calls. They'd be
there,
> and suddenly they'd drop out - they could be in a conversation, or more
> often, the caller would be on old, and suddenly the light would go out,
> and the caller would be gone.
>
> I haven't noticed anything unusual in the logs, nothing that stands out
> anyway. Is anyone else experiencing anything like this?
This is probably completely unrelated, particularly as it occurs on a
1.0.9 system, but we have one box which occasionally reports a PRI
"D-Channel up" message, resulting in killed calls, even though the
D-Channel is already up and working.

In our case it was fax-tone detection (for echo canceller switch-off)
in zaptel which seems to cause it. We have compiled this feature out
on this one server as a workaround. This can be spotted by looking for
the zaptel message in dmesg showing tone detection.

Strangely, the EXACT same software build and O/S with the same PRI
hardware and driver does not cause this problem on any of our other
servers.

Regards,
Steve

Did you ever solve this? I have noticed this also. I made a few changes
at the same time I moved to 1.2.0. One being that I switched providers
and thought it may be them.

 

Please let me know. Thanks!

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.digium.com/pipermail/asterisk-users/attachments/20051222/8e05d82b/attachment.htm