Look into memory leaks

[tepickering]

The wfssrv process is quite resource-intensive even when not being actively used. It appears that garbage collection is not being done properly which appears to be an issue with tornado in some circumstances. Need to go through and make sure all of the callbacks are exiting properly, unused objects get deleted manually, and maybe add some direct calls to gc to force garbage collection to happen.

[tepickering]

it looks like the leaks are under control. after several nights of use the wfssrv process looks like:

mmtop    10579  5.3  8.9 1628488 735260 ?      Ssl  Nov01 404:32 /mmt/conda/bin/python /mmt/WFSsrv/wfssrv/wfssrv.py

it starts off much smaller and increases with use, but seems to eventually stabilize around this level of usage. so it looks like it's not actively leaking, but it seems like there are things that could probably be garbage collected.

[tepickering]

here is a baseline process after the server restarted itself, but had not been used yet:

mmtop    10579  5.7  4.5 1219728 370296 ?      Ssl  Nov01 509:47 /mmt/conda/bin/python /mmt/WFSsrv/wfssrv/wfssrv.py