[OpenSIPS-Users] opensips kernel blocked for more than 120 seconds

[Jeff Pyle]

Hello,

I've got a weird one here.

I have 8 Opensips 1.6 instances in Xen domUs spread across two dom0s.
 Linux on dom0s and domUs is Debian Lenny.  Apparently all at one time they
became kernel blocked, and two minutes later I got messages like the
following on the domUs:

INFO: task opensips:17856 blocked for more than 120 seconds
> "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.


It's still unclear to me whether all Opensips processes hung on all
machines.  During this time, Opensips through syslog was complaining about
DB connectivity issues across the board, including messages like this:

ERROR:db_mysql:db_mysql_connect: driver error(2003): Can't connect to MySQL
> server on 'dbcluster' (111)


'dbcluster' is referenced in /etc/hosts to point to a mysqld instance of
MySQL cluster, different for each domU depending on which dom0 it resides
on.  I was able to connect to 'dbcluster' IPs from the mysql cli client,
however.  There were other indicators the cluster itself was fine,
including other connections from non-Opensips clients on other machines.
 Only Opensips was affected, and it was affected everywhere in my network.

The first time we tried to reset the opensips processes with the rc.d
script it failed to return.  Possibly because it was still having trouble
connecting to the db.  But, then it did restart with no explanation.

Does any of this sound familiar to anyone?


- Jeff
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20111102/13706e22/attachment.htm>