1 Reply Latest reply on Sep 1, 2007 5:58 AM by Anonymous

    Average Disk Queue length Monitor question

      Hi all,


       Thank you for all the help so far, I have a few of questions about setting up a monitor for average disk queue length.  I created monitors for the (3) available monitors on my server.  I have a 2 different 'drives' one in the local C: drive and the other is an array, the D:\ drive.


      As stated, I have creared the following monitors:


      1.3.6.1.4.1.9600.1.1.1.1.7.2.67.58
      1.3.6.1.4.1.9600.1.1.1.1.7.2.69.58
      1.3.6.1.4.1.9600.1.1.1.1.7.6.95.84.111.116.97.108


       I think the first corresponds to the C: drive, the second corresponds to the D: and the third is the sum of them together.  Can I get clarification on this point?  Does anyone know if my assumption is correct?


      My second question is this, within the monitor while it is running,  there is a 'value' variable that populates with a number as I increase the load on the disk queues.  However it never seems to be exactly the same as the number reported by Perfmon.  Of course, it could be that I am just not be able to physically switch back and forth while forcing a test to see the same numeric value.  Should I expect this number to be the same as reported by Perfmon?


      My third question also relates to the 'value' variable.  Unless the load is greater than 1 it doesn't seem to report anything.  Ever.  However the value is reported to four decimal places behind the integer.  For example, most of the time it sits at 0.0000 while Perfmon is happily reporting 0.027 or so.  Is this by design or is this an incorrectly defined monitor?


      Finally I have set my Analysis of Test Results to the Boolean range


      Pass if :


      <= 2


      OR


      <=0


      My goal was to set a test that would pass and not alert so long as the average disk queue length didn't sustain a load greater than 2 for 2 minutes or more.  If it does, then I want to alert the administrators and take action.  Is my logic consistent? 


      thanks for all the help,


      Jerry

        • Re: Average Disk Queue length Monitor question
          Hi Jeff, In response to your questions: Yes, you are right this is correct; the third value is for the total.  The same as with CPUs, you will get one OID for the CPU or each CPUs and one OID for CPU(s) total.  You could confirm this by contacting the SNMP Informant team. “My second question is this, within the monitor while it is running there is a 'value' variable that populates whit a number as I increase the load on the disk queues.  However it never seems to be exactly the same as the number reported by Perfmon.  Of course, it could be that I am just not be able to physically switch back and forth while forcing a test to see the same numeric value.  Should I expect this number to be the same as reported by Perfmon?” Are the numbers close?  These numbers are snapshots in time and may vary depending on when they where taken.  However, they should be as close as possible.  To validate this, try increasing the load constantly, are both the Perfmon counter and the SNMP value from the SNMP Monitor increasing?  The SNMP Informant support team might be able to provide more information on the OIDs they use and what is exactly corresponds to in regards to the Perfmon counter. In regards to your third question, I would use the following setup for the Monitor: Pass if Value matches a set of Boolean Rules: < 2 Then have the Monitor's timing set as follow: Timing section: Maximum Test Duration:  20 seconds  Delays Between Tests While: Up: 60  seconds Warn: 60 seconds Down: 60 seconds Lost: 60 seconds   Notification Control section: Accumulated Failures per Alert:  2  Maximum Alerts to Send:  3 

          This will have the Monitor test every 60 seconds and fail if the average disk queue length is higher than 1 for 2 minutes.

           

          I hope this helps.  Please let us know the information you get from SNMP Informant.

           

          Cheers,

           

          Stephane