[collectd] two concerns: mysql and network plugins
Florian Forster
octo at verplant.org
Fri Jun 20 10:47:19 CEST 2008
Hi Ruslan,
On Fri, Jun 20, 2008 at 04:06:20AM +0300, Ruslan Bondarev wrote:
> I've noticed, that mysql plugin, if mysql goes away or dies, shuts
> down collectd completely.
that'd be a major bug, but so far I don't see any hints towards this:
The function `getconnection' uses `mysql_ping' to test an existing
connection. If that fails, a warning is printed to the log file and
`mysql_real_connect' is called at increasing intervals until a
connection is established.
Are you positive the mysql plugin shut down your instance? If so, why
are you sure it was the mysql plugin? What was the exit status?
> Also, after message like this:
> [2008-06-13 11:35:24] network plugin: sendto failed: Network is unreachable
Sounds like your network had some trouble at that time.. What was
happening? When writing a bug report please provide as much information
as you can think of, because if you don't we'll improvise to fill in the
gaps and the picture in my mind right now is that you played with the
network, which made the database unreachable, which usually means that
*a lot* of stuff goes amok. When cleaning up you found that collectd had
been restarted so it must have been the mysql plugin's fault.
> [2008-06-13 11:36:19] perl: Initializing Perl interpreter...
> collectd stopped to send data to "master" collectd at all even after
> routing has been restored, and only kill -HUP helped.
There is no SIGHUP handler in collectd, so sending that signal will kill
the daemon. If you only checked the mysql plugin: The mysql plugin tries
to reconnect in increasing intervals so it doesn't spam the network/
database server with connection requests all the time. Maybe everything
worked just fine and you were too impatient and shut down collectd
yourself?
> (notice that collectd restarted itself - perl plugin startup message)
Are you using `collectdmon'? collectd doesn't restart itself - that's
done by `collectdmon', used by the Debian package, for example. If so,
you can find the exit status (see above) or the terminating signal in
the syslog log file.
Regards,
-octo
--
Florian octo Forster
Hacker in training
GnuPG: 0x91523C3D
http://verplant.org/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://mailman.verplant.org/pipermail/collectd/attachments/20080620/f80ebedb/attachment.pgp
More information about the collectd
mailing list