| *** sigmavirus24_awa has quit IRC | 03:27 | |
| *** asahlin has quit IRC | 03:28 | |
| *** asahlin has joined #openstack-searchlight | 03:31 | |
| *** sigmavirus24_awa has joined #openstack-searchlight | 03:34 | |
| *** lakshmiS has joined #openstack-searchlight | 05:02 | |
| *** lakshmiS has quit IRC | 05:11 | |
| *** lakshmiS has joined #openstack-searchlight | 05:14 | |
| *** asahlin has quit IRC | 10:27 | |
| *** nikhil_k has quit IRC | 10:27 | |
| *** mugsie has quit IRC | 10:27 | |
| *** sigmavirus24_awa has quit IRC | 10:27 | |
| *** sigmavirus24_awa has joined #openstack-searchlight | 10:29 | |
| *** asahlin has joined #openstack-searchlight | 10:30 | |
| *** nikhil_k has joined #openstack-searchlight | 10:30 | |
| *** mugsie has joined #openstack-searchlight | 10:30 | |
| *** asahlin has quit IRC | 10:30 | |
| *** nikhil_k has quit IRC | 10:30 | |
| *** mugsie has quit IRC | 10:30 | |
| *** asahlin has joined #openstack-searchlight | 10:33 | |
| *** nikhil_k has joined #openstack-searchlight | 10:33 | |
| *** mugsie has joined #openstack-searchlight | 10:33 | |
| *** lakshmiS has quit IRC | 13:21 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 13:30 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 14:38 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:38 | |
| openstackgerrit | Steve McLellan proposed openstack/searchlight: More efficient result filtering https://review.openstack.org/207682 | 15:49 |
|---|---|---|
| openstackgerrit | Steve McLellan proposed openstack/searchlight: More efficient result filtering https://review.openstack.org/207682 | 15:56 |
| rosmaita | sjmc7: got a minute? | 16:14 |
| sjmc7 | i'm on a phone call - ten minutes ok? | 16:14 |
| rosmaita | sure, i'll be here | 16:14 |
| rosmaita | thanks | 16:14 |
| sjmc7 | rosmaita, done | 16:57 |
| sjmc7 | sorry, meeting overran a bit | 16:57 |
| rosmaita | np | 16:58 |
| rosmaita | got a question for you about https://review.openstack.org/#/c/207682/6/searchlight/api/v1/search.py | 16:58 |
| sjmc7 | yep | 16:58 |
| rosmaita | i have not been as immersed in searchlight lately as i should be | 16:59 |
| sjmc7 | there's no time lik eht present to reform | 16:59 |
| rosmaita | so this is more of a project philosophy type question | 16:59 |
| sjmc7 | ah :) | 16:59 |
| rosmaita | i'm looking at the try at line 70 | 16:59 |
| sjmc7 | ok | 17:00 |
| rosmaita | so if there's a document type in the result set that we don't have a plugin for, we bail | 17:00 |
| sjmc7 | ah, yes. i questioned myself on this | 17:00 |
| rosmaita | so that implies that we expect there to be plugins for all doc types? | 17:00 |
| rosmaita | i was wondering whether we should jsut remove the hit | 17:01 |
| sjmc7 | yeah, i could see that. i'm not a huge fan of silent failures | 17:01 |
| rosmaita | i know, that's why it's kind of a philosophy question | 17:01 |
| rosmaita | we could log it | 17:01 |
| rosmaita | but if you're missing the nova plugin, you would get a lot of messages! | 17:02 |
| sjmc7 | yeah. i was trying to think how the situation could come about | 17:02 |
| rosmaita | right | 17:02 |
| rosmaita | yeah, if it would be mis-configuration or something, best thing to do would be to break completely as you do, the operator would definitely notice | 17:03 |
| sjmc7 | right.. in theory we restrict queries to plugins that are registered | 17:03 |
| rosmaita | i guess my question is that in general, we would expect all deployers of searchlight to have all plugins for all doctypes? | 17:04 |
| sjmc7 | well, i'd expect you only to get search results for stuff you had | 17:05 |
| rosmaita | right | 17:05 |
| rosmaita | and this is all server-side | 17:05 |
| sjmc7 | usually i like raising exceptions for stuff the user can correct | 17:05 |
| sjmc7 | in this case i'm not sure that's true | 17:05 |
| rosmaita | well, you log and raise a 500 ... gives a pretty good clue | 17:09 |
| sjmc7 | :) | 17:10 |
| sjmc7 | remove results is tricky | 17:10 |
| sjmc7 | because things like paging get messed up | 17:10 |
| sjmc7 | thinking about it more, i think an error is appropriate. this isn't something that should happen | 17:10 |
| rosmaita | i hadn't thought about that but that's a good point | 17:11 |
| rosmaita | yeah i agree with you | 17:11 |
| sjmc7 | the only reason i explicitly caught it was because when it did happen (i had it misconfigured) i got a log entry that was 'glance' or something | 17:11 |
| sjmc7 | because keyerrors are stupid | 17:11 |
| rosmaita | well, that shows it can happen! | 17:12 |
| sjmc7 | :) only cause i'm flipping between patches | 17:12 |
| sjmc7 | some of which have the nova plugin and some dont, some had different index names | 17:12 |
| *** sigmavirus24 is now known as sigmavirus24_awa | 19:05 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 19:59 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 21:59 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 21:59 | |
| *** david-lyle has quit IRC | 23:24 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!