[rrd-users] MRTG/RRD Intermittent Graphs

Darren.Terry at cox.com Darren.Terry at cox.com
Mon Feb 4 17:15:33 CET 2008


     This past week I migrated a rather large RRD/MRTG system to new
servers and updated software. The system was running on a redhat 9
server with dual P3 xeons running at 1.4GHz with 1GB of memory and U160
SCSI disks in RAID 1. The new system is a dual P4 xeon system with dual
cores, 4 146GB U320 SCSI disks and 8GB of memory... this system is using
RAID 5. The MRTG system has a total of 20,757 RRDs (i.e. interfaces).
The new server has much, much more bandwidth to the disks and also much,
much more memory and processor cycles.

     Now that I have the initial post out of the way, now comes the
problem. Previously this system was polling all of the devices with no
problems at all, no gaps in the graphs. However, with this new system,
there's one group of devices that don't seem to be graphing like they
should. There are gaps in the graphs for these devices. When polling the
poll doesn't finish before the next one needs to start. This doesn't
make sense... at all. This new system should be able to poll all of
these devices with no issues. If anyone here has any experience with
this, it would be awesome if you have any pointers on how to remedy this


More information about the rrd-users mailing list