DX NetOps

 View Only
  • 1.  Node down alarm criteria

    Posted Mar 08, 2017 03:47 PM

    How many SNMP polls or ICMP requests are used by Spectrum before it raises an node down or unreachable alarm?

     

    Thanks.

    Jon



  • 2.  Re: Node down alarm criteria
    Best Answer

    Posted Mar 08, 2017 04:03 PM

    SNMP is controlled by the TryCount attribute 0x110c5 of the model. For most SNMP managed devices that value is 3. Should an SNMP poll fail, by default, Spectrum waits 3 seconds and then retries the same poll. The wait time is controlled by the value of the TimeOut attribute id 0x110c4. By default it is set to 3000 ms (3 seconds). 

     

    So, Spectrum will retry 3 times with a 3 second wait between retries.

     

    If all 3 SNMP retries fail, Spectrum then pings the ip address of the model. By default, Spectrum only pings once. This is controlled by ICMP_trycount attribute id 0x11dac on the FaultIsolation model. Spectrum will wait 3 seconds for a response. This time is controlled by the ICMP_timeout attribute id 0x11dab default value 3000 ms (3 seconds) also on the FaultIsolation model.

     

    If the model support ip redundancy and it is enabled, Spectrum will try to regain communication using a different ip address configured on the device. If it is successful regaining communication with a different ip address, Spectrum will alarm that it changed the ip address. If it is not, Fault Isolation tries to determine where the fault lies and asserts alarms or suppresses accordingly.

     

    Joe