DX NetOps

 View Only
  • 1.  Spectrum Contact Lost "Server Down" Alarms

    Posted Jul 21, 2016 05:07 PM

    Hello Guys,

     

    Im new with spectrum and im a little confused about some OOTB alrmrs, i would like to know whats the difference between "Application contact lost", and "contact lost" alarms, im guessing onw is for ICMP contact lost and another is for SNMP but im not sure.

     

    Im trying to generate an alarm when a device cant be contacted using ICMP for a certain amount of time, how can i achive this ? Right now if i try that by restarting the device it down trigger an alarm regarding to server down, neither when the device is on "Initial" state.

     

    Thanks in advanced



  • 2.  Re: Spectrum Contact Lost "Server Down" Alarms

    Posted Jul 22, 2016 08:47 AM

    If the device is managed by snmp in Spectrum, Spectrum does not ping the device on a regular basis. The only time Spectrum will ping an snmp managed device is if the snmp request fails according to the DCM Timeout and Retry settings. If the snmp requests do fail, then Spectrum will ping the device. If the ping is successful, then Spectrum will logically conclude the device is up but not responding to snmp. A major "MANAGERMENT AGENT LOST" alarm is then asserted on the model. If the ping fails, then Fault Isolation is triggered to try and determine the root cause. Spectrum will then either assert a suppressed or critical condition on the model.

     

    A device manged by icmp in Spectrum is pinged according to the poll interval configured on the model. These models are commonly referred to a Pingable models. Out of the box, if the ping fails to one of these models, Spectrum asserts a blue initial condition. This is because Pingable models are considered insignificant models. This is controlled by the "Value_When_Red" attribute on the model. A value of "0" indicates an insignificant model. Set the value to a number greater than 0 if you wish to see a critical alarm on Pingable models.