Dear members...

I run a java application on my CentOS Server. We run this via a YAJSW wrapper and we start this one up as a service.

All things look ok, but at 2 sites this service is shut down. There are no errors in the log, just a normal shutdown message in the /var/log/messages file. It looks like somebody has typed the shutdown command for this service, only the site administrator said that it was impossible.

Has any of you had any similar experiences?

Hope you can help me.

Wilco

Some applications provide their own start/stop utilities. Often because you want them to start/stop automatically on reboot you create init scripts (RHEL6) or systemd units (RHEL7) that use the app's provided utilities. This means anyone who has access to the app's provided utilities can stop the app even if it was started originally by init.d (or systemd).

What "user" runs the application in question? Who has access to become that user (e.g. in same group as that user or sudo access to su to that user)?

I know, we wrote the application ourself and also the startup and stop script. So no suprises in here.

I know that there is something like a OOM trap in Linux, but I cannot find any automated kill commands in the logfiles, which should be there if the OOM progam shuts it down.

By the way, the program has root access.

OOM would not just "shutdown" processes. It attempts to more aggressively reclaim memory not in active use at the moment. If you're having OOM issues the main symptom would be extreme performance degradation as it steals the memory then the process that was using it claims it again ("memory flapping").

Can you post what you see in the log when this "shutdown" occurs?

When the shutdown occurs on RHEL6 in your init script are you creating a file under /var/lock/subsys at start for this service as you should? If so when you see the "shutdown" occurring is that file gone? If not then it likely isn't the init script being used to do the shutdown.

Of course there are reasons why a process might be "killed" rather than "shutdown".

Yes, I read about OOM killing processes. They also write there should be a KILL message somewhere in the logging. I cannot find a KILL message, so I do not thing an OOM is the issue. It is one server that has this issue, Other servers (customers) don't have this issue. It must be something network related or even site related. We keep monitoring. Thx for your responses by the way. Helps a lot.

Wilco