Memory leak in 19.1.1?

Users/Site Administrators
1

We seem to be experiencing a memory leak in 19.1.1 pbspro-server

We are meaning to update to 19.1.3 asap - for other reasons - but we aren’t seeing anything in the release notes about fixing a memory leak.

Every two weeks or so we have to restart the pbs.service to reclaim the memory. Eventually the OOM killer takes down DNS and everything cascades from there.

We now have monitoring in place to prevent the cascading catastophe, but we’ve not seen anything on here about memory leaks. Are we imagining this? What should I do to track down the leak if one exists?

2

I’ve just found the information in the INSTALL file but - and I wont lie - that’s a fairly complex change to make on a relatively heavily system with 50+ users running hundreds of jobs. It reads a more like a testing example rather than something designed for production systems. Are there any other methods for checking?