| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-02-10 | |||
| 20:36:15 | superdan | also, we only had two actual migrations in ocata and they were small | |
| 20:36:29 | superdan | mriedem: yep, but flavors, aggregates, server_groups, keypairs at least | |
| 20:37:25 | superdan | oh wait | |
| 20:37:29 | superdan | that was in newton | |
| 20:37:37 | openstackgerrit | Pushkar Umaranikar proposed openstack/nova master: Placement api: set custom json_error_formatter in resource_class https://review.openstack.org/430463 | |
| 20:37:50 | openstackgerrit | Maciej Szankin proposed openstack/nova master: conf: resolved final todos in libvirt conf https://review.openstack.org/429042 | |
| 20:37:51 | superdan | so we did that already in the pike migration.. sweet, nevermind, I lost track | |
| 20:39:01 | mriedem | ocata migration...? | |
| 20:39:38 | openstackgerrit | Maciej Szankin proposed openstack/nova master: conf: resolved final todos in libvirt conf https://review.openstack.org/429042 | |
| 20:39:38 | superdan | mriedem: https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/migrate_repo/versions/345_require_online_migration_completion.py | |
| 20:39:46 | mriedem | yeah | |
| 20:41:41 | mriedem | efried: you need to run py27 on your spec thing | |
| 20:41:44 | mriedem | zuul is unhappy | |
| 20:42:13 | efried | mriedem Roger that. (Who knew an RST needed python validation) | |
| 20:42:31 | efried | (I'm sure it's documented, isn't it) | |
| 20:43:51 | mriedem | leakypipes: https://review.openstack.org/#/c/432442/ fits, thanks | |
| 20:43:55 | dims_ | mriedem : "cat woke me up at 3am" :) | |
| 20:43:58 | mriedem | superdan: want to get that? | |
| 20:44:17 | superdan | mriedem: gross. | |
| 20:44:21 | mriedem | dims_: yes, well, i was awake at 2, tossed and turned for 45 minutes, fell asleep, and then the cat was running around making noise and woke me up again | |
| 20:44:47 | mriedem | then i proceeded to think about cinder v3 for some reason | |
| 20:45:03 | superdan | mriedem: I have to -1 that | |
| 20:45:30 | mriedem | superdan: the spec or the mid morning cinder v3? | |
| 20:45:37 | superdan | mriedem: the spec | |
| 20:46:42 | superdan | well, I guess maybe he's copy/pasting a bit.. grr, I will note my complaint and proceed I guess | |
| 20:49:30 | mriedem | because of "behaviour" with a "u"? | |
| 20:49:54 | superdan | eff yeah man | |
| 20:50:01 | mriedem | you know jay eats blood pudding for breakfast right? | |
| 20:50:23 | mriedem | breakfaust | |
| 20:50:36 | superdan | heh | |
| 20:56:39 | openstackgerrit | Eric Fried proposed openstack/nova-specs master: PowerVM Driver Integration (Pike) https://review.openstack.org/432387 | |
| 20:57:50 | openstackgerrit | Pushkar Umaranikar proposed openstack/nova master: Remove unused columns from BuildRequest table in nova_api db https://review.openstack.org/422259 | |
| 21:03:19 | openstackgerrit | Maciej Szankin proposed openstack/nova master: conf/compute: remove invalid TODOs https://review.openstack.org/432453 | |
| 21:06:56 | efried | mriedem Done and done. Thanks for the feedback. | |
| 21:07:12 | openstackgerrit | Maciej Szankin proposed openstack/nova master: conf/libvirt: remove invalid TODOs https://review.openstack.org/432455 | |
| 21:43:06 | openstackgerrit | Matt Riedemann proposed openstack/nova-specs master: Fix a couple of spec names https://review.openstack.org/432469 | |
| 21:52:41 | openstackgerrit | Matt Riedemann proposed openstack/nova-specs master: Fix paths when moving specs https://review.openstack.org/432470 | |
| 21:52:42 | openstackgerrit | Matt Riedemann proposed openstack/nova-specs master: Move implemented specs for Ocata https://review.openstack.org/432471 | |
| 21:59:59 | openstackgerrit | Sujitha proposed openstack/nova master: Change os-server-tags default policy https://review.openstack.org/396420 | |
| 22:00:48 | mriedem | melwitt: https://review.openstack.org/#/c/429075/ looks good - leakypipes want to +W that? | |
| 22:03:03 | leakypipes | mriedem: done | |
| 22:03:44 | openstackgerrit | Pushkar Umaranikar proposed openstack/nova master: Placement api: set custom json_error_formatter in inventory https://review.openstack.org/432478 | |
| 22:17:17 | openstackgerrit | Matt Riedemann proposed openstack/nova-specs master: PowerVM Driver Integration (Pike) https://review.openstack.org/432387 | |
| 22:22:21 | efried | mriedem Thanks! | |
| 22:23:15 | mriedem | np | |
| 22:23:27 | openstackgerrit | Pushkar Umaranikar proposed openstack/nova master: Adopts keystoneauth with glance client. https://review.openstack.org/412634 | |
| 22:25:09 | mriedem | superdan: i'm going to start bugging you about stable/newton things, starting with https://review.openstack.org/#/c/408725/ | |
| 22:28:58 | superdan | mriedem: hmm, so this one is in ocata and we're just backporting it if they want to get a jump on the cleanup right? | |
| 22:30:00 | mriedem | superdan: yeah, it was reportd by someone moving to newton | |
| 22:30:09 | mriedem | so i fixed on ocata with the intent of backporting | |
| 22:30:14 | mriedem | hence the comments in there | |
| 22:33:01 | superdan | yeah, reading, I just want to get my head wrapped around it, since this is fairly strange | |
| 22:33:14 | superdan | people would have run the online migrations when they upgraded to newton and wouldn't again until the next time | |
| 22:36:42 | mriedem | this is to resolve the issue for anyone upgrading from mitaka before the hit the issue, unless i'm not catching your point | |
| 22:37:36 | superdan | yeah, I think I get it | |
| 22:37:51 | mriedem | in mitaka we had the build request mirror the instance, | |
| 22:38:04 | mriedem | failed instance build in the api (i think) which didn't get deleted, | |
| 22:38:06 | mriedem | something like that | |
| 22:38:15 | mriedem | then in newton, we serialized the instance into the build request, | |
| 22:38:29 | mriedem | but when you try to load that old build request from mitaka, it's data format is different and blows up on load | |
| 22:38:52 | superdan | yeah | |
| 22:39:00 | superdan | why didn't we just do the delete.filter(instance_uuid=null()).limit($limit) ? | |
| 22:39:05 | superdan | instead of the pre-query? | |
| 22:39:11 | superdan | I mean too late now, just thinking | |
| 22:39:20 | mriedem | this https://review.openstack.org/#/c/306023/ | |
| 22:39:45 | mriedem | because it's sticking with the limit/count semantis in the online migrations cli | |
| 22:39:54 | mriedem | we talked about just doing the delete straight up on master | |
| 22:40:29 | mriedem | https://review.openstack.org/#/c/408727/2/nova/objects/build_request.py | |
| 22:40:35 | superdan | I think we'd still be able to do the same thing with just delete, but anyway | |
| 22:40:55 | superdan | did we also make it not blow up on load? | |
| 22:41:07 | superdan | this makes instance list blow up until you run this right? | |
| 22:42:16 | mriedem | i'd have to dig into all of that context again | |
| 22:42:23 | mriedem | you were +2 on the ocata change fwiw :) | |
| 22:42:57 | superdan | I know, I just don't have this in my head anymore and.. it's a weird thing to backport | |
| 22:43:23 | mriedem | the whole idea was to backport it from the start | |
| 22:43:27 | mriedem | the blow up on load is here https://review.openstack.org/#/c/298455/16/nova/objects/build_request.py@110 | |
| 22:44:06 | mriedem | with a mitaka db record that doesn't have the instance_uuid, that's going to blow up either way, so i'm not sure how we'd handle it, unless we checked for a uuid recored? | |
| 22:44:09 | superdan | I know, I know | |
| 22:44:10 | mriedem | *record | |
| 22:44:54 | superdan | um | |
| 22:45:04 | mriedem | https://github.com/openstack/nova/blob/stable/mitaka/nova/db/sqlalchemy/api_models.py#L143 | |
| 22:45:05 | superdan | did we not bump the version on that because it wasn't sent over RPC? | |
| 22:45:09 | superdan | but it was stored in the db... | |
| 22:45:31 | mriedem | "Note that the BuildRequest object version is not being bumped because this object is not yet passed over the wire anywhere or looked up anywhere, so there are no backwards compatibility concerns." | |
| 22:46:57 | superdan | yeah, I'm just trying to figure out why we didn't try to handle the explosion | |
| 22:47:01 | superdan | but I'll just stop asking questions I guess | |
| 22:48:35 | mriedem | i'm assuming no one just ever thought of it | |
| 22:50:15 | superdan | in newton, we were listing buildreqs and putting them in front of the instance list, | |
| 22:50:21 | cfriesen | I'm trying to run devstack on master and getting this: | |
| 22:50:24 | cfriesen | pkg_resources.VersionConflict: (packaging 16.5 (/usr/local/lib/python2.7/dist-packages), Requirement.parse('packaging>=16.8')) | |
| 22:50:37 | superdan | and presumably this but means BuildRequestList.get_all() would fail to yield anything until you run this migration | |
| 22:50:52 | superdan | which makes the api broken until you get this online migration | |
| 22:50:52 | cfriesen | anyone know what's going on there? | |
| 22:51:05 | superdan | but we could have just made BuildRequestList.get_all() ignore a known exception there and log | |
| 22:51:13 | superdan | which seems like an obvious fix, so I'm just wondering why we didn't do it | |
| 22:55:43 | mriedem | i also backported https://review.openstack.org/#/c/420051/ and i'm not sure if we care or not now | |
| 22:58:44 | mriedem | i think we want all of these in newton if we want people to use placement before ocata https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:stable/newton+topic:bug/1647316 | |
| 22:58:50 | mriedem | which, maybe we care less about now | |
| 22:58:57 | mriedem | anyway, gotta run now | |
| 22:59:03 | superdan | I care bigly | |