| opendevreview | Thomas Goirand proposed openstack/oslo.cache master: WIP dont-merge: Clean-up memcache connection sockets https://review.opendev.org/c/openstack/oslo.cache/+/900158 | 09:59 |
|---|---|---|
| opendevreview | Thomas Goirand proposed openstack/oslo.cache master: WIP dont-merge: Clean-up memcache connection sockets https://review.opendev.org/c/openstack/oslo.cache/+/900158 | 10:36 |
| opendevreview | Thomas Goirand proposed openstack/oslo.cache master: WIP dont-merge: Clean-up memcache connection sockets https://review.opendev.org/c/openstack/oslo.cache/+/900158 | 10:37 |
| zigo | Hi there! | 10:40 |
| zigo | On the above patch, we're trying to fix the fact that oslo_cache.memcache_pool never closes connection correctly, leaving a storm of source port in CLOSE_WAIT state. | 10:40 |
| zigo | The thing is, we're not sure how to call self._destroy_connection(conn). Is it ok to just do a while conn = self.queue.pop().connection and close them all, or should we just close *ONE* connection? | 10:40 |
| zigo | Any hint on how to fix this very annoying bug (for us in production) would be highly appreciated. | 10:41 |
| zigo | hberaud: ^ | 10:41 |
| *** elodilles_pto is now known as elodilles | 11:39 | |
| zigo | Looks like my 3rd patch is passing CI. | 11:46 |
| hberaud | zigo: at first glance your implementation looks correct | 12:43 |
| zigo | :) | 13:01 |
| zigo | Thanks. | 13:01 |
| zigo | Removing the WIP. | 13:02 |
| zigo | We'll first test it locally in our billing systems (ie: gnocchi doing thousands of connections ...). | 13:02 |
| hberaud | thanks | 13:12 |
| *** gouthamr_ is now known as gouthamr | 16:09 | |
| *** JayF is now known as Guest6061 | 19:55 | |
| *** JasonF is now known as JayF | 19:55 | |
| *** zigo_ is now known as zigo | 19:57 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!