| *** harlowja has quit IRC | 01:14 | |
| *** bcoca has quit IRC | 03:01 | |
| *** harlowja has joined #ara | 04:23 | |
| *** harlowja has quit IRC | 05:35 | |
| *** gvincent has joined #ara | 05:49 | |
| *** gvincent has quit IRC | 05:49 | |
| *** gvincent has joined #ara | 05:49 | |
| dmsimard | gvincent: so there's two major problems with the current stable release (even with some patches to reduce the pain) | 13:04 |
|---|---|---|
| dmsimard | *Larger* scale playbooks, i.e, https://i.imgur.com/1TKZE0a.png can either 1) Eat all your RAM 2) Be slow | 13:06 |
| dmsimard | Right now the pagination for the results is done client side (with javascript datatables) and what's eating the memory is the sqlalchemy looping over all the results, i.e paste.openstack.org/show/718864/ | 13:07 |
| dmsimard | http://paste.openstack.org/show/718864/ <-- for some reason that had no http | 13:07 |
| gvincent | with DRF the pagination will fix the ui issue | 13:08 |
| gvincent | you will just return the 1000 max elements and paginate trought the ui if you want to see more | 13:08 |
| dmsimard | right | 13:08 |
| gvincent | the performance issue is only on the ui ? | 13:08 |
| dmsimard | yeah, callback is fine | 13:09 |
| dmsimard | so the thing with the datatables | 13:09 |
| dmsimard | is that since it has all the data, the search box allows you to easily search with regexes etc client side | 13:09 |
| dmsimard | I suppose if we do server-side pagination and we want to keep searching, we'll have to work something out | 13:09 |
| gvincent | see https://github.com/lesspass/backend/blob/master/lesspass/settings.py#L132-L133 | 13:09 |
| dmsimard | nice.. pagination is another thing that was kind of awkward in flask-restful | 13:10 |
| *** bcoca has joined #ara | 13:10 | |
| *** bcoca has joined #ara | 13:10 | |
| gvincent | and for the search | 13:10 |
| gvincent | https://github.com/lesspass/backend/blob/master/lesspass/settings.py#L134-L137 | 13:10 |
| gvincent | everything is already there | 13:10 |
| dmsimard | <3 | 13:10 |
| gvincent | :) | 13:11 |
| gvincent | https://github.com/lesspass/backend/blob/master/api/views.py#L10-L11 | 13:11 |
| dmsimard | gvincent: so the current plan is to land a few patches in the master branch and then cut a 0.15 -- to limit the memory usage and improve performance a bit. | 13:11 |
| gvincent | yes | 13:12 |
| dmsimard | I've been experimenting on the integration jobs starting with the role https://review.openstack.org/#/c/557933/ | 13:12 |
| dmsimard | The general idea would be to really have first party support for the role by using it to deploy ARA in integration test jobs we can always make sure that it works | 13:12 |
| dmsimard | People tend to get confused with WSGI installation and other things despite the docs -- and I suppose we need some way of "packaging" the NPM bits | 13:13 |
| gvincent | my need right now is a small db with fake (but real) data | 13:14 |
| dmsimard | gvincent: from stable release ? or from 1.0 | 13:14 |
| gvincent | 1.0 | 13:14 |
| dmsimard | I can probably easily adapt the callback to use the django standalone client for the time being and then see if that works -- we're going to have to do it anywya | 13:15 |
| gvincent | I see 2 solutions: We work to adapt the actual callback making api request | 13:15 |
| dmsimard | API was already implemented in https://github.com/openstack/ara/blob/feature/1.0/ara/plugins/callbacks/log_ara.py | 13:16 |
| dmsimard | Probably need to adjust some bits but the bulk of the work is done I suppose | 13:16 |
| gvincent | yeah sorry, we should adpat the actual callback to use django standalone | 13:16 |
| dmsimard | i.e, https://github.com/openstack/ara/blob/feature/1.0/ara/plugins/callbacks/log_ara.py#L123 | 13:16 |
| dmsimard | All the API calls the callback needs to do are obvious so I can take those and adapt to django client | 13:17 |
| dmsimard | We can think about optimizing the code later | 13:17 |
| gvincent | <3 | 13:18 |
| gvincent | :D | 13:18 |
| ara-slack | <dmsimard> @harlowja I'd appreciate if you could test https://review.openstack.org/#/c/559840/ -- the parent patch is a rebase of the sqlalchemy pool addition https://review.openstack.org/#/c/524427/ | 13:20 |
| ara-slack | <dmsimard> It doesn't address the memory usage but it prevents it from growing infinitely | 13:21 |
| ara-slack | <dmsimard> If it works for you, please +1 the patches :) | 13:22 |
| *** tbielawa has joined #ara | 13:34 | |
| *** gnupyx has joined #ara | 14:30 | |
| *** rvgate has joined #ara | 15:11 | |
| *** gnupyx has quit IRC | 15:58 | |
| *** harlowja has joined #ara | 16:05 | |
| ara-slack | <harlowja> kk | 16:11 |
| *** tbielawa is now known as tbielawa|lunch | 16:26 | |
| *** harlowja has quit IRC | 17:07 | |
| *** gvincent has quit IRC | 17:33 | |
| *** tbielawa|lunch is now known as tbielawa | 17:46 | |
| *** harlowja has joined #ara | 17:52 | |
| *** tbielawa is now known as tbielawa|brb | 18:25 | |
| *** tbielawa|brb has quit IRC | 18:29 | |
| *** tbielawa has joined #ara | 18:49 | |
| *** BlessJah has quit IRC | 19:32 | |
| *** BlessJah has joined #ara | 19:33 | |
| *** berendt_ has joined #ara | 20:21 | |
| *** berendt has quit IRC | 20:28 | |
| *** berendt_ is now known as berendt | 20:28 | |
| *** tbielawa has quit IRC | 20:29 | |
| openstackgerrit | Merged openstack/ara master: Add support for configuring sqlalchemy pool size, timeout and recycle https://review.openstack.org/524427 | 22:13 |
| openstackgerrit | Merged openstack/ara master: Improve memory usage when dealing with large amount of data https://review.openstack.org/559840 | 22:19 |
| -openstackstatus- NOTICE: zuul was restarted to updated to the latest code; you may need to recheck changes uploaded or approvals added between 21:30 and 21:45 | 22:28 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
