[collectd] Bugreport: Collectd is rising load dramatically

[Johannes oetzi Ott]

On Sat, Sep 16, 2006 at 10:51:28AM +0200, Florian Forster wrote:
> Hi Johannes,
> 
> On Fri, Sep 15, 2006 at 10:30:52PM +0200, Johannes oetzi Ott wrote:
> > when starting collectd the load rises up to a constant level above 3.0
> > and an average of about 9.0. Yesterday load  wents up to about 60.0 
> 
> that's weird, with collectd being a single-thread (/-process) daemon it
> shouldn't rise the load by more then one, ever. So my guess is that it's
> blocking or slowing another process, however that should work..
> 

That was the thing I meant.

> With a load at around 60 it should be quite cozy on the run-quere. Could
> you check which commands/processes are waiting to be run? An easy method
> of doing that is (with the GNU `ps'):
>  $ ps ax -o state,command | egrep '^R'
> 

There are normally only a few apache-process shown. But they always seem
to be there about a second. So I don't really know what processes are slown 
down.

> Also, please try disabling all plugins and try all plugins successively
> by themselves. I've heard faint complaints about the `ntpd' plugin
> (``Didn't work, so I disabled it and didn't bother''), so you might want
> to try that one first.
> 

I have done that now. The plugin rising the load when being activated is
not ntpd (this works fine), but it is the ping-plugin.

> > I think that there is a problem with amd64 dualcore cpu of the server.
> 
> That I doubt very much. I know that collectd runs on machines with many
> more CPUs, so _that_ isn't a problem. Also, 64bit processors are no
> problem either, so I don't see why the architecture should play any role
> in this..
> 

I know that collect is running on multi CPUs-machines, but the sucking
dualcore amd-cpu is really problematic as I had to regonized under debian.
It was just a guess what could be the reason

> Regards,
> -octo
> -- 
> Florian octo Forster
> Hacker in training
> GnuPG: 0x91523C3D
> http://verplant.org/


But now back to the problem. As I mentioned above the ping-plugin is the 
problem.

Perhaps there is a problem with that the other traffic of the machine is 
very high, but this is just a guess again.

These are the hosts I try to ping.

<Plugin ping>                                                                                                                                                            
   Host t-online.de                                                                                                                                                      
   Host arcor.de
   Host strato.de
   Host irc.gamesurge.net
   Host recall-revolutions.de
</Plugin>

We have a average response-time about 16.2 ms per host.

Hope you perhaps know a solution.

Best regards

oetzi

-- 
Johannes Oetzi Ott  \      oetzi at gletschereis.net     / ascii ribbon campaign _
                     \  http://www.gletschereis.net  / - against html mails  ( )
                      \      GPG-ID: 7737 71E3      / - against microsoft     X
                       \                           /   attachments & vcards  / \
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: Digital signature
Url : http://mailman.verplant.org/pipermail/collectd/attachments/20060919/b49626cb/attachment.pgp