pmacct-discussion  

Re: [pmacct-discussion] reloading config & accuracy

Paolo Lucente
Tue, 23 Jun 2009 02:05:57 -0700

Hi Chris,

On top of the previous questions, maybe i skipped two things: a) the
most obvious question: which version are you verifying such behaviour? 
Is it 0.11.6? b) another alternative is to let me troubleshoot on the
system itself - granted that the daemon can be restarted and the box
is geared with compiler and debugger. 

Cheers,
Paolo


On Tue, Jun 23, 2009 at 08:15:38AM +0100, Paolo Lucente wrote:
> Hi Chris,
> 
> On Tue, Jun 23, 2009 at 09:48:38AM +0300, Chris Wilson wrote:
> 
> > I also get huge spikes which are unaccountable for, even after recompiling 
> > with 64-bit counters. It's not a single entry but rather all flow records 
> > for a few minutes at a time, then it goes back to normal. It's way more 
> > than my poor router or internet connection could sustain (gigabytes per 
> > second). This is using pcap rather than netflow. I'm suspecting memory 
> > corruption in pmacct.
> > 
> > How would I get the average packet size for you?
> 
> In this case i would rather like to get privately a full tcpdump trace
> of the traffic at the moment the spike occurs (so that i can reproduce
> the issue myself and troubleshoot). So far, I had a limited number of
> reports of occasinal spikes specifically for pmacctd (never for nfacctd
> or sfacctd). 
> 
> Can you recognize any patterns? Ie. always same host(s) involved? Let me
> know.
> 
> Cheers,
> Paolo


_______________________________________________
pmacct-discussion mailing list
http://www.pmacct.net/#mailinglists