question about thresholds on things are aren't interfaces

 
1
0
-1

Using NMIS to monitor network gear, switch/routers/firewalls, is great.  And when it comes to monitoring other things like servers it does a pretty good job out of the box.

However, I am running into a few things that are a bit frustrating for me. 

  1. the  NMIS VM is reporting that it is in a degraded state.  When I check the status, the only thing that reports "not green" is the hrSystemProcesses = 182.  Not sure why, I have even doubled the RAM, and the runtime graphs are under 10secs as I currently have only 10 items on my local network configured.  Is this normal?
  2. I have configured one of my exchange servers, and NMIS seems to be doing a good job of monitoring all the parts of this 08R2 VM.  However, it is notifying me that the drives are in a "minor" (78.97) and "major" (81.02) levels.  As I certainly want to monitor these values, I would like to be able to adjust what the value=level is on a server to server basis.

 

I may have missed where to make these kinds of adjustments, but thus far the only "threshold" values I have been able to find out how to adjust in the documentation were related to interfaces, not drives utilizations or processes.

 

Thanks for any help.

    CommentAdd your comment...

    2 answers

    1.  
      3
      2
      1

      dale, the process event  is caused by the alerting setup in the Net-SNMP model, and the default threshold for this is 150 processes. My personal gut feeling is that with busy servers this is on the low side, but then it's  configurable (if not in the most comfortable fashion):

      you need to EITHER use the Model editor (menu system -> system configuration -> nmis models; then select model net-snmp, and section system) OR the threshold/alert tuning dialog (menu setup -> thresholding alert tuning; again select model net-snmp and section system).

      after scrolling down a lot you'll see  section alerts and the block that controls hrSystemProcesses. Please note that changing this value has global effects, i.e. it applies to all nodes with Model net-snmp.

       

        CommentAdd your comment...
      1.  
        2
        1
        0

        Hello Dale,

        Thresholds can be tuned from the following menu:
        Setup => Threshold Alert Tuning

        (this is looking at an 8.5.6G  install - let us know the version if you can't find that menu item)

        The table that appears shows the thresholds for all devices and you can change them globally their.  You will also see two drop downs in the top right corner where you can either select a particular device or select a group to which to apply those new thresholds.  So for the one server in question select that server first then change the threshold values.

        hrSystemProcesses = 182

        This is simply a threshold alert about the number of Processes running on the server.  Your server is probably by the sounds of it healthy with that many processes so you may want to change that threshold.  So all services are fine with the CPU, Memory etc and are responding fast.

        1. Dale Hunkele

          Thanks for the prompt response.  I am running 8.5.6G via a VM appliance.

           

           I did find where to find this menu in the documentation, but what I am wanting/needing to tweak isn't listed here.

           

          What is available there is CPU, RAM, and utilization thresholds, which is great for network gear just not servers.

           

          I did find the hard disk thresholds under the alerts, so this mystery is solved.

          I did find references to hrSystemProcesses under the system dropdown.  However there was nothing about theresholds there.

           

          Is the 182 system processes "normal" for monitoring 10 devices?

           

        2. Dale Hunkele

          well, I got the threshold tweaked for the hard drive utilization, and the hrSystemProcesses dropped from 180+ to some number that it doesn't get reported in the status window.  So I don't know what fixed that one.

          Thanks for getting me headed in the right direction.

        CommentAdd your comment...