Forum Discussion

Manog1978's avatar
2 years ago

Alert Suppress

Is there any feasibility do not monitor the particular process, example, there is backup job running on the server and due to on this , the CPU and memory process goes to High and getting many alerts in LM.

instead of using SDT Mode, if there any possible do not monitor particular process.

  • Dynamic thresholds would be a good solution as well. 

10 Replies

  • As a more complex answer… I could see a way to use the current CPU load of a given process to have thresholds that push threshold adjustments for another DS on that device via the REST API in the background.  Kind of a process dependent dynamic threshold rather than an averaged one like the main LM product offers.  This would be a DS that would drive scripting in the background rather than being used to throw an alert.  If using powershell (I’m stubbornly not learning Groovy as we use powershell throughout the entire rest of our workflow)… 

    get-wmiobject win32_process | where-object name -eq “nameOfYourProcess”

    The returned object will give you current workload status of that process in the properties that can tell you the CPU load at the moment.  If CPU > 50%, REST API for the device if DS CPU Threshold < 95%  increase the CPU DS threshold.  If CPU < 50% && DS CPU Threshold = 95%, lower the CPU DS threshold.

  • Good Morning Manog1978,
    I would need a little bit more information in order to answer this fully.  From my current understanding, you are referring to individual DataSources for monitoring CPU statistics and Memory statistics on a device.  If this is the case, they are polling metrics for these individual components vs. the specific processes and services running.  You also do have the ability to monitor individual processes and services as outlined in the ‘Monitoring Web Pages, Processes, Services and UNC Paths’ support document.  Monitoring has to be set up manually for these items.

    As for a potential solution to the CPU and Memory monitoring, you could consider tuning the alert threshold on this device to be less sensitive.  Also, if the backup jobs occur at specific times, you have the ability to set thresholds to be active during certain time periods.  Additional information can be found in the ‘Tuning Static Thresholds for Datapoints’ support document.

  • Hi

    Thank your for quick responce. i would like to stop monitoring for specific processes as like backup job and if we enable SDT, not sure how long to be set and it is not best practices.

  • Like @mark.lemm mentioned, if you’re talking about alerts from the CPU datapoint, there’s not really a way to do that in that DS. That DS just asks the device, “what’s your cpu usage” and the device responds with the overall number. The device would have to know that you don’t want to include certain processes in the overall number, which leads down the path of a different DS. 

    If you are talking about one of the several process monitoring options that are available, just disable the instance corresponding to the unwanted service or put SDT on the instance.

  • Hi

    Thank you for update. 

    I tried to implement Tuning Static Thresholds  and SDT Option to stop the CPU alert when Backup Job utilization more cpu, but it is not correct solution i need to provide the some exception if the backup job utilization more cpu some certain time interval instead of SDT, the alert should not trigger, as it is getting more noise in Event management, 

    Please advise if any alternate solution.

  • Hi Stuart,

    We are receiving multiple CPU alert from LogicMointor and multiple incidents are created in ServiceNow, When we checked in LM, the alert triggered in particular time frame example (10 am to 12 am) some time it may be varify based on backup job process running in the Host since it is utilization more cpu

    I would like to stop monitor or provide exception for backup job when it is increasing cpu.

    Is it possible in LM. 

  • Great.. thank you very much. thanks for all providing such a solution.