Hi all,

My 'top' command below catches my Debian 5 VPS jumping from low average load to high average load with about 6 processes using 100% of the CPU (causing the websites to fail for about 15 minutes or until I restart the VPS). Can I identify what these processes are from their PID? Or would the apache logs be more helpful?

Thank you

Attached Thumbnails

 

You can close apache2 processes, using:-
ps -ef | grep apache2
And then,
Then if needed, you can restart apache service.

Ok thanks but I meant to say that this happens regularly every day - so I am trying to work out the cause.

"cat /proc/<pid>/cmdline".

... or just run top with the "-c" switch.

Am I doing something wrong...


Thanks

Depending on the process, you may need to be root, but more likely the process has already terminated - the /proc/<pid>/... "files" only exist whilst that process is active.

Ok thanks - probably terminated because I restarted the VPS. So does a process get a consistent PID or might the PID vary from one day to another?

Thanks

You mean pgrep apache2 or for sending a signal and allowing processes to close file descriptors:

Killing processes may alleviate the stress the system is under but without (pointing to performing) proper analysis it's like treating symptoms instead of addressing the root cause: not that effective.


Depending on how your machines specs (RAM and other bottlenecks), how access is regulated (like firewall request limiting), how your web server is configured (like stock with all unnecessary modules loaded), what it runs (any interpreter-based Off The Shelf application, homebrewn code, etc, etc) and their dependencies / configuration, web server configuration and access + error log files should be the first things to check. You can also look at a current process with 'strace'. First select the Process Id:
and then
when you think you've got sufficient nfo kill the process with CTRL+C and then read "/path/to/strace.log" for clues.
*Also note that collecting system statistics over time, running any SAR like Atop, Dstat or Collectl, may give a better picture of what happens and when making it easier to search logs for clues.


Restarting a service might help but restarting your VPS is quite drastic as you loose information that may help diagnosing the problem and it isn't how Linux should be handled. And PIDs are assigned sequentially BTW. With this: you'll see all processes sorted by PID and this: confines your view to only Apache2 processes.