New manually installed agents take several hours before showing health state RRS feed

  • Question

  • In our environment (SCOM 2012 R2) we have around 4000+ client systems. Agent installation is always "manual" i.e. not pushed from the SCOM management servers. The problem we have is the time that it takes for a newly installed SCOM agent/client system to change from the "not monitored" status to where its health status is reported. Typically, this takes a few hours (in excess of 3 hours). 

    I have tried running a trigger on-demand discovery for "Discover Windows Client Computers" against the specific client system that I'm trying to get on-boarded into SCOM, but it does not produce consistent results. Sometimes, the health state of a client system for which I run the script is reported within a few minutes of agent installation, but at other times it takes a few hours (same result as the case where no trigger on-demand discovery is run).

    Is there a way to speed up the process of transitioning the agent status from "not monitored" to a monitored state where health state is shown? 

    Is the trigger on-demand discovery process useful in speeding up this process? 

    • Edited by Sumi Rachel Friday, February 8, 2019 4:08 AM
    Friday, February 8, 2019 4:08 AM

All replies

  • Hi

    This is a normal behavior and as far as I know you have no control over it. It comes up when it comes up.


    Sam (Please take a moment to "Vote as Helpful" and/or "Mark as Answer" wherever applicable. Thanks!) Blog:AnalyticOps Insights Twitter:Sameer Mhaisekar

    Friday, February 8, 2019 5:26 AM
  • Try the following method to check whether it can speed up the process
    1. trigger on-demand discovery for "Discover Windows Client Computers"
    2. restarting the Microsoft Monitoring Agent service on the agent-managed computer or
    3. clearing the agent cache
    Friday, February 8, 2019 6:31 AM
  • Thanks for the suggestions, Roger.

    I've already got #1 and #2 in a script, and #1 runs in a loop until the agent is discovered. It seems to have brought down the timeframe by an hour. As regards #3, given that these are new agents, would there be a significant cache to begin with? I'll give it a go nevertheless, so I'm planning to change my script to: 

    1. install agent

    2. Clear cache and restart service after 15 seconds of installation of agent

    3. Run the on-demand discovery script in a loop till the health state changes from "not monitored"

    Sunday, February 10, 2019 10:26 PM