0

I'd like to start keeping track of which binaries users run on our Linux cluster. I'd like log the results on each node, and then aggregate them to find the total. I'd like to eventually to get to the point where I can run a report at the end of the month and see that /usr/bin/foo was run 9000 times.

My question is about the logging aspect. What facility is best to achieve this? process accounting? auditd?

Specifically I'd like the log to keep track of the path to each executable run, and how often it was executed.

2

Years ago I administered an LSF cluster and we did exactly this. We used system processing accounting. The systems I managed were running Solaris but it shouldn't be any different for Linux.

| improve this answer | |
0

I don't know for sure but I would guess that looking at the audit subsystem to put specific audit events on the binaries you want to watch would be the best way. My hazy recollection is that the process accounting system logs a lot more than what you may want.

| improve this answer | |
0

If your looking for a quick and free solution. Use SGE "Scheduling system".

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.