I was tasked by our Linux admins to create a performance collection to see how many nodes in a system are busy at a time. The script for that uses vmstat and lscpu and takes some time to run. The script is executed every five minutes and the results are returned ok and can be displayed as graph in SCOM.
Now they want to get alerted when a certain threshold is reached. I could create a monitor for that using the very same script. I was wondering if there is a smarter way of doing that by temporarily storing the number the performance collection script has already gathered so that the monitor could pick that up. That would reduce the load on the system to collect the data. One way would be to store it somewhere in the filesystem. Not sure, if that is a good approach though.
How does the other monitors and performance collections like cpu and disk space, do that ?
Thanks in advance