NSClient++ Help (#1) - CheckCounter vs. Perfmon (#654) - Message List

CheckCounter vs. Perfmon

Hi gang,

I'm monitoring some perfmon counters using nsclient++ 0.3.8.

I see weirdness between perfmon and the graphs I get in nagios (using rrdtool/pnp4nagios).

does anyone know how to explain this?

for the perfmon graph, I set up a 'Performance Logs and Alerts' Counter Log and sampled at the same interval as the nagios check. I figured there would be some minor variation between the two, but this is pretty significant.

I'd appreciate any insight!

 http://www.oddelement.com/images/perfmon.png  http://www.oddelement.com/images/nagios.png thanks,

addodd

oddboy
08/30/10 22:13:25 (18 months ago)
Tree View Flat View (newer first) Flat View (older first)
  • Message #1949

    Humm...

    Well counters (in nsclient++) are samples when you ask for them (not in between) so the values (for such a sporadic counter) will most likely not be that usable I would think.

    But I am by no means an expert. I guess a more sensible concept in this case would be to continually sample the values storing some form of min/max or something like that (which by the way is not really supported in nsclient++)

    Michael Medin

    mickem
    08/30/10 22:29:34 (18 months ago)
    • Message #1950

      thanks for the reply.

      for each graph, the settings are the same. query the perfmon counters (LogicalDisk?(E:)\Disk Reads/sec and Writes/sec) at 5 minute intervals. graph the results.

      I'm just surprised that the graphs are so different.

      I know rrdtool does some consolidation on the data it collects, but the data is just soooo different, it's alarming.

      these items may be very sporadic, but they can be useful over time to highlight trends.

      slightly related, on another drive (where the SQL Server log files are stored), the graphs for perfmon and nagios are almost identical... just not on a drive that has both read/write activity.

      weird.

      addodd

      oddboy
      08/30/10 23:09:20 (18 months ago)
      • Message #1951

        Humm...

        again no expert here but...

        The value is not "read/write per second during the last 5 minutes" it is the absolute value of read/write per second during the last second or so. Which roughly translates to something not that useful since you poll every 5 minutes. You would have to be very lucky to get the same values (unless your IO is very even). But if you are writing when perfmon polls (1 second before) you could very well be reading when Nagios polls a second later... (again, unless your IO is even)

        Michael Medin

        mickem
        08/31/10 06:52:04 (18 months ago)
Tree View Flat View (newer first) Flat View (older first)

Subscriptions