[smokeping-users] After a lot of alerts, smokeping do not update the rrd anymore

G.W. Haywood ged at jubileegroup.co.uk
Mon Aug 6 16:59:03 CEST 2012


B11;rgb:ffff/ffff/ffffHi there,

On Mon, 6 Aug 2012, Nicolas KARP wrote:

> I ran several smokeping for several months now and all was fine until we
> had an issue with our ISP... Following these issues, smokeping sent a lot
> of alarms which is what we wanted :-) But after that, smokeping stopped to
> update the RRD file and the graphs were not updated.
>
> Do you know what could cause this problem with smokeping ?  Too much
> unresponsive devices ?

Are you by any chance using logarithmic graphs?

I've seem Smokeping fail to update graphs (and the smokeping processes
hanging using 100% CPU) when using logarithmic graphs.  I have a patch
which has been in testing for a week but seems stable enough to release.

The problem I had was that sometimes when you use logarithmic graphs the
minimum for the vertical scale is greater than the maximum, and this
causes something in RRD to get confused.

My fix is to make sure that the minimum is less than the maximum, but it
seems to me that a fix is also needed in RRD for the problem behaviour.

In Smokeping.pm:
----------------------------------------------------------------------
   '--title',$desc.$from,
   '--rigid',
! '--upper-limit', ($max->{$s}{$start} > 0.01 ? $max->{$s}{$start} : '0.01'),
   @log,
   '--lower-limit',(@log ? ($max->{$s}{$start} > 0.01) ? '0.001' : '0.0001' : '0'),
--- 1371,1377 ----
   '--title',$desc.$from,
   '--rigid',
! '--upper-limit', $max->{$s}{$start},
   @log,
   '--lower-limit',(@log ? ($max->{$s}{$start} > 0.01) ? '0.001' : '0.0001' : '0'),
----------------------------------------------------------------------

Apologies for any line wrapping.

--

73,
Ged.



More information about the smokeping-users mailing list