Linux - alert action how to restart daemon

MathieuJM

Hello,

SolarWinds is not really linux friendly and unfortunately we have to monitor Linux box

We monitor a specific Linux server and want to restart some daemon based on alert criteria.

Has onyone be able already to restart Linux daemon on Linux server monitored with agent ?

Typically a simple systemctl restart command would solve most of our issues.

cheers

Find more posts tagged with

alert_trigger_action

linux services

Alerts

Accepted answers

All comments

solatyc

Also looking for a solution like this. Its so easy to set automatic restarts for services in windows servers but no feature like this for Linux. You would think this would be a built in feature by now.

chad.every

This is a very old post that I wrote that talks about how it could be done. Basically kicking off a custom script via the Alert Engine to do it. This was only for a proof of concept, and it isn't very polished, but it did work.

Alert Action - Run Linux Script With Putty

However, if I had to do it again I'd probably skip the Alert Engine and simply use a SAM Application Template with a Linux Script Monitor (since you already have to have an Application Template for your Process monitors). Have an extra monitor in that template that runs a script that monitors the deamons I want up, and if they are stopped to restart them. Build in some type of logic to handle too many restart attempts, etc. If the deamons are all up, then have the script simply output something to signify everything is ok. The pros to this approach, is that it keeps the ssh creds more secure AND will use the Linux Agent to execute the script if it is deployed.

mesverrum

I got to where I do this for all of my custom script monitors. Why just monitor and create alerts when I'm already in here writing code that will execute on the target machine. If the condition fails i write an output that will set the component into Warning and so SAM records that i've initiated a remediation, and if that remediation fails then we actually mark the component as down and we know that whatever is wrong is more complex than my existing automatic fix can handle.