| amoralej_ | sean-k-mooney, wrt https://review.opendev.org/c/openstack/watcher/+/1014459/ , what if we create a new DecisionEngineContext different to the RequestContext that only contains the connections which are reused by the collectors, notification handlers and other clients usage inside the decision-engine? | 11:16 |
|---|---|---|
| amoralej_ | and similar for the applier | 11:16 |
| amoralej_ | in https://review.opendev.org/c/openstack/watcher/+/1014459/comment/61b57970_9ea9a71d/ you mentioned "we shoull not do it per instance but we should likely do it per audit/actionplan". My proposal is to share it per-service, given that we are not using request context but config file credentials | 11:18 |
| amoralej_ | btw, in the tests morenod is doing, it seems to be more bugs in openstacksdk related to this https://bugs.launchpad.net/openstacksdk/+bug/2174410 https://bugs.launchpad.net/openstacksdk/+bug/2174412 | 11:21 |
| amoralej_ | and also, some weird behavior is being observed even with fixes for those, that seems related to https://discuss.python.org/t/memory-leak-when-using-concurrent-futures-threadpoolexecutor-with-glibc-but-possibly-not-with-jemalloc/108311/15 | 11:23 |
| dviroel | amoralej_: o/ i start an etherpad to better list the options for pin/unpin policy: https://etherpad.opendev.org/p/watcher-migrate-pin-policy | 14:21 |
| dviroel | sean-k-mooney: jgilaber ^ if you folks have some time.. | 14:21 |
| jgilaber | dviroel, just in time, I was about to start looking at the spec :) | 14:38 |
| dviroel | jgilaber: oh nice. yeah, so the etherpad is for the pinnin policy part, which we still need to agree around the options and the scope :) | 14:38 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!