[asterisk-users] Asterisk database weird behaviour

[Andrew Nowrot]

I am using realtime architecture in my Asterisk 11.3.0 system; for cdr,
dialplan, sip accounts, voicemail and queues. My database (on separate
machine)  is postgres 8.4 in drbd cluster with one active and one passive
node controlled by corosync and pacemaker. So 99% of my sip accounts is
defined in realtime, but I also have some statically in sip.conf

I have a following problem:
Each time I have to interact with the database (backup with pg_dump or
preform synchronisation between two nodes) asterisk stops responding to sip
registrations.
The bizarre thing is that it also can`t reach users defined in
sip.conffile (no database). Asterisk can execute
dialplan defined in extensions table, it can also insert data in to
cdrafter the call is completed, the only problem is with sip messages.
I thought that this is caused by transactions that are put on database
during for example backup, but with the problem in progress I could log in
to postgres console with pgsql and I was able to execute sql (updates or
inserts), so no transaction is active.

Does anyone have similar behaviour or does anyone know what is causing this
and how to fix it?

Cheers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-users/attachments/20130902/5a155035/attachment.htm>