| *** tetsuro has joined #openstack-placement | 00:28 | |
| *** tetsuro_ has joined #openstack-placement | 02:53 | |
| *** tetsuro has quit IRC | 02:56 | |
| *** tetsuro has joined #openstack-placement | 04:29 | |
| *** tetsuro_ has quit IRC | 04:32 | |
| *** dklyle has quit IRC | 05:04 | |
| *** tetsuro_ has joined #openstack-placement | 05:32 | |
| *** tetsuro has quit IRC | 05:34 | |
| *** evrardjp has quit IRC | 05:35 | |
| *** evrardjp has joined #openstack-placement | 05:36 | |
| *** tetsuro has joined #openstack-placement | 06:40 | |
| *** tetsuro_ has quit IRC | 06:44 | |
| *** tetsuro_ has joined #openstack-placement | 07:04 | |
| *** tetsuro has quit IRC | 07:07 | |
| *** tetsuro has joined #openstack-placement | 08:35 | |
| *** tetsuro_ has quit IRC | 08:38 | |
| *** tetsuro has quit IRC | 09:00 | |
| *** martinkennelly has joined #openstack-placement | 09:32 | |
| *** openstackgerrit has joined #openstack-placement | 10:46 | |
| openstackgerrit | Merged openstack/os-resource-classes master: setup.cfg: Configure 'python-requires' https://review.opendev.org/704533 | 10:46 |
|---|---|---|
| *** martinkennelly has quit IRC | 11:43 | |
| *** martinkennelly has joined #openstack-placement | 11:48 | |
| *** dklyle has joined #openstack-placement | 14:11 | |
| openstackgerrit | Merged openstack/os-resource-classes master: Switch to hacking 2.x https://review.opendev.org/710276 | 16:03 |
| *** evrardjp has quit IRC | 17:36 | |
| *** evrardjp has joined #openstack-placement | 17:36 | |
| *** dklyle has quit IRC | 20:39 | |
| *** dklyle has joined #openstack-placement | 20:40 | |
| *** openstackstatus has joined #openstack-placement | 21:16 | |
| *** ChanServ sets mode: +v openstackstatus | 21:16 | |
| *** david-lyle has joined #openstack-placement | 22:01 | |
| *** dklyle has quit IRC | 22:03 | |
| melwitt | dansmith: hey, I could use some help reasoning about whether to have a server_default here for a consumer type. the last two comments are the relevant part. I am really confused about what to do. https://review.opendev.org/#/c/669170/15 would it really be so bad to have a server_default? it would make things a lot easier AFAICT | 22:06 |
| melwitt | and TBC, the patch series began with a server_default and then I got spooked about the bug we had in nova over the instances.hidden field | 22:07 |
| *** david-lyle has quit IRC | 22:08 | |
| dansmith | melwitt: otp right now, bbl | 22:27 |
| melwitt | dansmith: k thanks. it's not urgent, just wanted to leave you a note | 22:30 |
| melwitt | been struggling with it for awhile and now putting out the help beacon | 22:31 |
| *** sean-k-mooney1 has joined #openstack-placement | 23:02 | |
| *** sean-k-mooney has quit IRC | 23:04 | |
| dansmith | melwitt: okay but you know that default= is what nailed us on that, not server_default | 23:19 |
| melwitt | dansmith: I do. so maybe I'm overthinking it for server_default | 23:19 |
| melwitt | server_default will also backfill, but the database server will do it | 23:19 |
| dansmith | it will? or it will set the default field on the column? | 23:20 |
| melwitt | it will | 23:20 |
| melwitt | Mike's last comment on the review explains the tradeoff | 23:21 |
| dansmith | well, my preference for all of this is to make these columns nullable and handle the "default" on read ourselves | 23:22 |
| melwitt | maybe I'm premature optimzation-ing but I'm feeling careful after the hidden field thing | 23:22 |
| dansmith | I don't think that mike's last comment says server_default will fill for sure, but let me roll back a few | 23:23 |
| melwitt | dansmith: it doesn't but I tried it and it does backfill. and also I googled a lot and learned it backfills | 23:24 |
| dansmith | okay | 23:24 |
| dansmith | well, then IMHO, the thing to do is make it nullable and handle it on read | 23:25 |
| dansmith | he did say that regardless, backfilling on large tables will suck with galera | 23:25 |
| melwitt | so I started going in the handle null direction but Mike pointed out that will make the queries inefficient | 23:25 |
| dansmith | welp, I dunno what to say, but data manip during migrations suck for lots of reasons, | 23:27 |
| dansmith | now including that failure we saw earlier | 23:27 |
| *** spatel has joined #openstack-placement | 23:27 | |
| melwitt | yeah.. | 23:28 |
| melwitt | dansmith: this was what started it https://review.opendev.org/#/c/669170/15/placement/objects/usage.py@135 and then so then I went back to thinking whether I should write online_data_migrations for this or just leave it as null and handle it (like hidden) | 23:30 |
| *** spatel has quit IRC | 23:51 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!