Custom Alert for Application Process Unavailable (unexpected)
Hi there
I want to customise the "Application Process Unavailable (unexpected)" Incident to wait for 5 minutes before it Alerts and sends an email Servers are often rebooted which raises an alert, but also creates a lot of noise when looking after 2000 servers. I can customise other Incidents like Host CPU etc. by creating another Incident and Measure and changing the Evaluation time-frame to 5 minutes.
Can I do the same with Application Process Unavailable (unexpected)? How would I do that?
Anton
Hi Chris.
Thanks for the feedback. We use this mostly for Infrastructure monitoring, so we cannot shutdown the application gracefully here. This is the only reliable way of detecting if an OS is down. This works well, but we do get quite a number of alerts when machines are rebooted for patching and other routine restarts. It isn't really practical to put each of of the servers in downtime mode every time someone wants to restart a server.
I will put this as a RFE then.
Thanks
Anton
Unfortunately no, you cannot change the interval in which this is triggered due to the way this incident is generated.
The agent for whatever reason was not shutdown cleanly, either a force close, process termination, or fault. I would recommend shutting down the application cleanly before shutting down the server.
When an agent is started, the name of the agent appends the process ID of the application it is monitoring. In 99.9% of cases this process ID is not the same as the old process ID, which means you would not be able to correlate the agent to an older agent in order to evaluate if the same agent connected or not.
Thanks,
Chris
