NSO Automation

[rengonca]

Hello guys,

I am performing a few tests regarding NSO maintenance automation and the behaviour of NSO is not what I was expecting. During the pipeline, I stop the NSO daemon (ncs --stop) and restart it with (ncs --with-package-reload-force). I created 2 pipelines: the first is a dry-run and the second is a commit. In both cases, at the end of the pipeline, Jenkins is killing the NSO daemon. The python script to perform the injection of new configuration (push_config.py) doesn't use a daemon, it is a simple python script that sends commands using python's request library. I think I can trap the SIGTERM SIGKILL SIGQUIT in the "push_config.py" script so it would not let Jenkins kill the daemon. It would likely preclude the graceful stop with the OS too though. This is the current output --> https://pastebin.com/LEMCuNMP Do you guys have any idea regarding this behaviour, did you have this problem in the past?

[yfherzog]

Can you elaborate which aspect is not as expected?

What I could get is that the script push_config.py is failing at some point. Is that the issue?

[rengonca]

Hello yfherzog,

Thank you for your reply.

I was expecting jenkins doesn't kill the ncs service.

I believe is some variable that I need to check in jenkins to dont disturb the ncs and leave it alone in case there is no changes to apply on the devices.

Regards
RG

[yfherzog]

It Doesn't seem to me as if it is related to what you're running against NSO once it's running.

Your checks also confirm that NSO is up before the pipeline ends.

I never witnessed that kind of behavior from Jenkins, but I think it is indeed related to Jenkins configs.

Have a look here:

https://stackoverflow.com/questions/39169457/how-to-tell-jenkins-not-to-kill-processes-after-successful-execution-of-job-in-m?rq=1