*** efried has joined #openstack-placement | 05:13 | |
*** efried has quit IRC | 05:36 | |
*** alex_xu has quit IRC | 06:11 | |
*** alex_xu has joined #openstack-placement | 06:14 | |
*** tetsuro has joined #openstack-placement | 06:30 | |
*** tetsuro has quit IRC | 07:20 | |
*** mriedem has joined #openstack-placement | 07:40 | |
*** cdent has joined #openstack-placement | 08:04 | |
*** e0ne has joined #openstack-placement | 08:13 | |
mriedem | cdent: i've got a debug patch up for that grenade-py3 job failure, not sure what's going on there | 08:13 |
---|---|---|
cdent | I saw your comments on it perhaps being osc-placement | 08:13 |
cdent | which makes sense | 08:13 |
cdent | and good find if it is that | 08:14 |
mriedem | dansmith: this is the placement fixture stuff from gibi so nova can continue to use it https://review.openstack.org/#/c/599551/ | 08:14 |
mriedem | i'm going to setup a py3 devstack to mess with that as well | 08:16 |
dansmith | mriedem: is there a nova-side patch that depends on that to show that it works for nova functional tests? | 08:18 |
dansmith | I'm a bit fuzzy on how we're going to get the in-master placement bits via pip for our functional tests without needing a release of placement in between every feature that crosses the boundary | 08:19 |
cdent | dansmith: there's a couple of wips from gibi : https://review.openstack.org/#/c/601641/ and https://review.openstack.org/#/c/599556/ but I'm not sure of the state on those as it's been quite a while | 08:20 |
dansmith | cdent: so we're going to have placement in our requirements and placement-from-git in our test-requirements? | 08:21 |
dansmith | even still, I don't think we'll be able to test nova patches against un-merged placement patches that way | 08:21 |
openstackgerrit | Merged openstack/nova-specs master: Spec: allocation candidates in tree https://review.openstack.org/603585 | 08:22 |
cdent | I don't recall the details because it's been now months since I was investigating that, but in terms of testing un-merged stuff I think the solution to that is the same as ever: depends-on. | 08:23 |
cdent | I can pick it back up now that the alembic stuff and other testing stuff has made some progress | 08:24 |
dansmith | I'm saying I don't think depends-on will fix that unless I'm missing something | 08:24 |
dansmith | meaning zuul would have to know to re-write our test-requirements | 08:24 |
* cdent is still on his first cup of coffee | 08:24 | |
cdent | so apologies for being slow | 08:25 |
cdent | but tox-siblings may be helpful on this | 08:25 |
cdent | but I need to refresh my brain | 08:25 |
cdent | https://docs.openstack.org/infra/manual/zuulv3.html#installation-of-sibling-requirements | 08:27 |
mriedem | i'm not sure the test-requirements thing is going to be an issue, | 08:28 |
mriedem | i didn't know we had a git link to placement in nova's test-requirements, i'm not sure what that's for, | 08:28 |
cdent | I think that's just an artifact of gibi's testing, not a thing that will stick around | 08:29 |
mriedem | but i'd expect placement gets released to pypi, nova's test-requirements depends on placement, and we use depends-on for zuulv3 to pull in and install whatever dependent change we need | 08:29 |
dansmith | yeah that link from cdent says that zuul will specifically *fix* the venv which would be installed incorrectly from test-requirements | 08:29 |
dansmith | like test-requirements will be wrong, install the venv wrong, and then zuul fixes it before we run tests in it | 08:29 |
dansmith | (wrong in the sense that it won't initially reflect the depends-on) | 08:30 |
dansmith | that's definitely different in zuulv3, so I'm just behind on that I guess | 08:30 |
cdent | it's a) cool and interesting, b) problematic for local dev | 08:30 |
dansmith | yeah the local dev thing is different and annoying | 08:31 |
dansmith | but not unlike other situations we have | 08:31 |
mriedem | fwiw openstack/placement is already in required-projects for the base dsvm job in zuul | 08:33 |
mriedem | we needed that for the grenade stuff | 08:33 |
mriedem | http://git.openstack.org/cgit/openstack-infra/openstack-zuul-jobs/tree/zuul.d/jobs.yaml#n957 | 08:33 |
cdent | anybody know: What's a reasonable base zuul job if you just wanna run a shell script on a simple linux box? | 08:43 |
*** e0ne has quit IRC | 08:47 | |
cdent | mriedem: weirdly the output columns are in a different order between py2 and py3 | 08:48 |
cdent | 2: http://logs.openstack.org/96/617896/1/check/neutron-grenade/a3f2510/logs/grenade.sh.txt.gz#_2018-11-14_08_35_19_465 | 08:48 |
cdent | 3: http://logs.openstack.org/96/617896/1/check/grenade-py3/c5c1aba/logs/grenade.sh.txt.gz#_2018-11-14_08_44_04_936 | 08:48 |
*** e0ne has joined #openstack-placement | 08:49 | |
*** e0ne has quit IRC | 08:50 | |
mriedem | hmm, some column hash thing in the dict results from osc i'm guessing | 08:50 |
* cdent nods | 08:50 | |
cdent | can we use -c to limit the output further? | 08:51 |
*** e0ne has joined #openstack-placement | 08:51 | |
mriedem | i think i asked that on the original grenade change to avoid a lot of the piping and grep'ing and there was a reason for those, but can't remember now, looking it up | 08:51 |
* cdent has no cloud handy at the moment | 08:52 | |
mriedem | https://review.openstack.org/#/c/597206/6/projects/60_nova/resources.sh@57 | 08:53 |
mriedem | and above that | 08:53 |
mriedem | but that was on listing providers, not inventory | 08:54 |
mriedem | i think we can use -c on getting the specific inventory | 08:54 |
mriedem | i shall push a change | 08:54 |
mriedem | allocation show won't work with -c b/c the CLI dumps the 'resources' json blob | 08:55 |
mriedem | so...that's fun | 08:55 |
cdent | "fun" | 08:55 |
*** e0ne has quit IRC | 08:57 | |
*** e0ne has joined #openstack-placement | 08:59 | |
*** e0ne has quit IRC | 09:10 | |
*** e0ne has joined #openstack-placement | 09:12 | |
*** e0ne has quit IRC | 09:13 | |
*** e0ne has joined #openstack-placement | 09:14 | |
*** e0ne has quit IRC | 09:15 | |
mriedem | ta da https://review.openstack.org/617914 | 09:21 |
mriedem | we should maybe add an "openstack resource provider allocation show <consumer_uuid> <resource_class>" CLI | 09:22 |
cdent | might make sense, at least for these kinds of things. another option would be to curl and pipe to jq, but I'm drawn that way | 09:24 |
*** cdent has quit IRC | 09:43 | |
*** rubasov has joined #openstack-placement | 10:04 | |
openstackgerrit | Jim Rollenhagen proposed openstack/nova-specs master: Use conductor groups to partition nova-compute services for Ironic https://review.openstack.org/609709 | 10:05 |
*** cdent has joined #openstack-placement | 10:06 | |
gibi | cdent, dansmith: making nova unit test run with splited out placement is on my todo list, but lately spent zero time on it unfortunately | 10:25 |
gibi | s/unit/functional/ | 10:26 |
cdent | gibi: do you think you'll be able to get back to it soonish? If not, I can probably pick it up. | 10:27 |
gibi | cdent: not earlier than next week but next week I can spend a day or two on it I think | 10:29 |
cdent | gibi: okay, cool, if I get a chance before then I might poke at it | 10:30 |
gibi | cdent: are we on the same page about running nova functional test with which version of splitted placement if no depends-on specified? a) master b) last released | 10:31 |
cdent | based on the earlier discussion, because of tox-siblings the gate ought/should use master | 10:31 |
gibi | cdent: OK, so we assume we always release nova and placement in synch, that works for me | 10:32 |
gibi | cdent: I will ping you before I start picking up the work next week | 10:35 |
cdent | ✔ | 10:35 |
*** rubasov has quit IRC | 10:36 | |
*** mriedem has quit IRC | 10:40 | |
cdent | gibi: which one of these is the right one? https://review.openstack.org/#/c/599556/ or https://review.openstack.org/#/c/601641/ | 10:41 |
cdent | or is it worth starting over? | 10:41 |
cdent | or something else? :) | 10:42 |
* gibi tries to remember :) | 10:45 | |
gibi | cdent: I think my list trials was with https://review.openstack.org/#/c/596291 + https://review.openstack.org/#/c/601641 | 10:46 |
gibi | s/list/last/ | 10:46 |
cdent | ah, right, I forgot about the one I started, thanks | 10:51 |
gibi | cdent: yeah I started out on a different track first but then ended up going back to your original trial as that was workable | 10:53 |
* cdent plays with it | 10:54 | |
*** e0ne has joined #openstack-placement | 10:57 | |
*** e0ne has quit IRC | 11:30 | |
*** cdent has quit IRC | 11:36 | |
*** cdent has joined #openstack-placement | 12:12 | |
*** mriedem has joined #openstack-placement | 12:19 | |
*** tetsuro has joined #openstack-placement | 12:37 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Adapt placement database fixture for external use https://review.openstack.org/617935 | 12:46 |
openstackgerrit | Merged openstack/nova-specs master: Use conductor groups to partition nova-compute services for Ironic https://review.openstack.org/609709 | 12:52 |
*** tetsuro has quit IRC | 12:57 | |
*** tetsuro has joined #openstack-placement | 13:01 | |
mriedem | cdent: edleafe: i'm hacking on some sqlite model sync testing stuff | 13:17 |
mriedem | just fyi | 13:17 |
cdent | cool, thanks | 13:17 |
cdent | I'm hacking on external placement nova func tests | 13:17 |
cdent | and am breaking everything | 13:18 |
*** e0ne has joined #openstack-placement | 13:23 | |
*** tetsuro has quit IRC | 13:32 | |
mriedem | so, we just don't have any models sync testing in the placement change as far as i can see, | 13:33 |
mriedem | nova gets that by extending an abc from oslo.db | 13:33 |
mriedem | i mean there is ModelsMigrationSyncMixin | 13:34 |
mriedem | but it's not being used | 13:34 |
mriedem | oh nvm i missed it for sqlite | 13:34 |
mriedem | i see, in nova there is a single mixin which extends both the walk and sync classes from oslo.db, so nova just has a single set of sqlite/mysql/pg tests which run both the walk versions and models sync tests | 13:37 |
*** tetsuro has joined #openstack-placement | 13:37 | |
cdent | i (very briefly) tried to make a subclass that used 'sqlite' as the driver but hit some bumps and then had to leave the room I was in | 13:37 |
*** e0ne has quit IRC | 13:38 | |
mriedem | got it | 13:40 |
mriedem | working for sqlite i mean | 13:40 |
cdent | huzzah | 13:42 |
*** openstackgerrit has quit IRC | 13:51 | |
mriedem | i'll update and rebase the series | 13:52 |
*** openstackgerrit has joined #openstack-placement | 13:54 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Adapt placement database fixture for external use https://review.openstack.org/617935 | 13:54 |
*** tetsuro has quit IRC | 13:55 | |
*** tetsuro has joined #openstack-placement | 13:55 | |
openstackgerrit | Matt Riedemann proposed openstack/placement master: Added alembic environment https://review.openstack.org/614350 | 13:59 |
openstackgerrit | Matt Riedemann proposed openstack/placement master: Delete the old migrations https://review.openstack.org/611440 | 13:59 |
openstackgerrit | Matt Riedemann proposed openstack/placement master: Add a document for creating DB revisions https://review.openstack.org/614024 | 13:59 |
openstackgerrit | Matt Riedemann proposed openstack/placement master: Remove sqlalchemy-migrate from requirements.txt https://review.openstack.org/614539 | 13:59 |
openstackgerrit | Matt Riedemann proposed openstack/placement master: Add a placement-manage CLI https://review.openstack.org/600161 | 13:59 |
cdent | mriedem: that's almost exactly the same as what I did, but I hadn't got the db url quite right | 14:02 |
cdent | soo...yay | 14:02 |
*** tetsuro has quit IRC | 14:07 | |
*** tetsuro has joined #openstack-placement | 14:11 | |
*** efried has joined #openstack-placement | 14:16 | |
cdent | edleafe: you around today? | 14:25 |
edleafe | yep | 14:26 |
cdent | you want to handle matt's final (?) comments on the alembic stuff? | 14:27 |
edleafe | I looked at that earlier, before those comments. I should have time to address them before I descend into meeting hell. | 14:28 |
mriedem | the only hangup i seem to have is the initial empty placeholder migration, | 14:33 |
mriedem | i'm not sure if that's just something with the sqlite backend or what | 14:34 |
mriedem | lemme run it in isolation | 14:34 |
*** e0ne has joined #openstack-placement | 14:35 | |
mriedem | i suspect there might be a sqlite db from a previous run that's not cleaned up | 14:36 |
edleafe | mriedem: the test calls "self.migration_api.upgrade('158782c7f38c')", and you removed migration 158782c7f38c, so that exception should be expected. | 14:36 |
mriedem | oh nvm | 14:36 |
mriedem | i'm just dumb | 14:36 |
mriedem | yup | 14:36 |
mriedem | heh | 14:36 |
edleafe | mriedem: I just changed that to "self.migration_api.upgrade('base')", and that seems to work | 14:37 |
mriedem | ok sec | 14:37 |
mriedem | i've got this change locally so i can just update and rebase if you're ok with that | 14:37 |
mriedem | or you can if you're going to handle any other comments | 14:37 |
mriedem | yup base works | 14:38 |
edleafe | You also have to change the initial migration to have "down_revision = None" | 14:39 |
mriedem | i did, | 14:39 |
mriedem | i pasted my diff in the review comments | 14:39 |
mriedem | i'll let you do whatever you're going to update based on my comments | 14:39 |
edleafe | k | 14:40 |
cdent | efried, mriedem : If we get a chance to talk to belmoria before he leaves before lunch tomorrow about the how's and why's of what's going on with all the info that the reource tracker is moving back and forth to/fro placement that would probably be useful for all of us | 14:41 |
mriedem | was b4ed3a175331_initial.py generated from the models? | 14:41 |
efried | ++ | 14:41 |
mriedem | if so, i didn't realize it was auto-generated | 14:42 |
efried | do you have a line on where he might be hanging out? | 14:42 |
edleafe | mriedem: yeah, alembic generated that | 14:42 |
edleafe | mriedem: that's why cdent didn't want to edit it | 14:42 |
mriedem | edleafe: ok, i didn't realize, i thought those scripts were hand generated | 14:42 |
mriedem | sorry about the noise on that one then | 14:42 |
mriedem | yeah | 14:42 |
mriedem | didn't realize | 14:42 |
edleafe | But as I commented, it is designed to be modified, so I prettied it up | 14:43 |
edleafe | just for you! | 14:43 |
mriedem | <3 | 14:43 |
cdent | efried: he was "in the hall" just now, but he indicated that email would be a useful way to contact him | 14:44 |
*** tetsuro has quit IRC | 14:44 | |
mriedem | would be nice if using email, it could be on the ML | 14:45 |
mriedem | rather than private | 14:45 |
mriedem | unless there are sensitive details he doesn't want to expose | 14:45 |
mriedem | unless it's "wtf is going on, this is so complicated we don't know how to maintain it" and that might hurt feelings | 14:46 |
cdent | no, it was a case of "let's have a chat in person somewhere here because we have the opportunity" | 14:46 |
*** tetsuro has joined #openstack-placement | 14:48 | |
*** e0ne has quit IRC | 14:51 | |
*** tetsuro has quit IRC | 14:56 | |
*** mriedem has quit IRC | 15:06 | |
*** cdent has quit IRC | 15:09 | |
*** N3l1x has joined #openstack-placement | 15:10 | |
*** cdent has joined #openstack-placement | 15:12 | |
*** mriedem has joined #openstack-placement | 15:20 | |
mriedem | cdent: efried: i talked to belmiro | 15:20 |
mriedem | i think he was ok with my answers | 15:20 |
mriedem | and then he gave me 2 other issues to worry about | 15:20 |
cdent | yeah, that's why I thought it would be good to have a sit down | 15:20 |
cdent | because he gave me a few too | 15:21 |
*** e0ne has joined #openstack-placement | 15:22 | |
*** efried has quit IRC | 15:26 | |
openstackgerrit | melanie witt proposed openstack/nova-specs master: Propose counting quota usage from placement and API database https://review.openstack.org/509042 | 15:29 |
openstackgerrit | Ed Leafe proposed openstack/placement master: Added alembic environment https://review.openstack.org/614350 | 15:31 |
openstackgerrit | Ed Leafe proposed openstack/placement master: Delete the old migrations https://review.openstack.org/611440 | 15:31 |
openstackgerrit | Ed Leafe proposed openstack/placement master: Add a document for creating DB revisions https://review.openstack.org/614024 | 15:31 |
openstackgerrit | Ed Leafe proposed openstack/placement master: WIP - Show an alembic migration https://review.openstack.org/614025 | 15:31 |
edleafe | mriedem: cdent: ^^ Placeholder migration is no more | 15:32 |
mriedem | edleafe: so close... | 15:35 |
mriedem | If you want to tack on a follow up i'll +2 | 15:35 |
cdent | as far as I can tell the nova-functional jobs aren't doing tox siblings | 15:36 |
edleafe | mriedem: I'll FUP it | 15:36 |
*** tetsuro has joined #openstack-placement | 15:37 | |
mriedem | edleafe: ok | 15:39 |
mriedem | +2 | 15:39 |
cdent | yay | 15:40 |
openstackgerrit | Ed Leafe proposed openstack/placement master: Fix comment in earlier patch https://review.openstack.org/617988 | 15:40 |
edleafe | mriedem: FUP^^ | 15:41 |
openstackgerrit | Takashi NATSUME proposed openstack/placement master: Remove build-openstack-api-ref jobs https://review.openstack.org/614497 | 15:46 |
cdent | wowser, stephenfin, shared tox envs and zuul's tox_install_siblings behavior conflicts: | 15:50 |
cdent | envdir does not exist, skipping siblings | 15:51 |
cdent | oops: http://git.openstack.org/cgit/openstack-infra/zuul-jobs/tree/roles/tox/library/tox_install_sibling_packages.py#n172 | 15:51 |
openstackgerrit | Chris Dent proposed openstack/placement master: Remove sqlalchemy-migrate from requirements.txt https://review.openstack.org/614539 | 15:53 |
openstackgerrit | Chris Dent proposed openstack/placement master: Add a placement-manage CLI https://review.openstack.org/600161 | 15:53 |
stephenfin | cdent: I assume the whoops means different issue? | 16:00 |
cdent | stephenfin: no, the oops was that I pasted the error message, not to the code that causes it. The problem is for real | 16:00 |
cdent | the diff between revisions 5 and 6 on https://review.openstack.org/#/c/617941/ fixes it | 16:01 |
stephenfin | cdent: Sounds like something you should ping mordred about. I know very little about this whole siblings thing | 16:02 |
cdent | stephenfin I will, but for the time being I'll leave the tox.ini change in place. It's going to be hard for siblings to do what it does without what amounts to consistent naming | 16:03 |
stephenfin | Yeah, that sounds fair. Maybe add a TODO too, just for future reference? | 16:04 |
stephenfin | Lest we forget to investigate it | 16:04 |
cdent | yeah, I'll tidy it up after I see it actually work... | 16:04 |
*** cdent has quit IRC | 16:06 | |
*** irclogbot_2 has quit IRC | 16:08 | |
*** mriedem has quit IRC | 16:10 | |
*** mriedem has joined #openstack-placement | 16:12 | |
*** irclogbot_2 has joined #openstack-placement | 16:17 | |
*** tetsuro has quit IRC | 16:26 | |
mriedem | should be easy to cleanup the alembic docs change https://review.openstack.org/#/c/614024/ | 16:29 |
*** e0ne has quit IRC | 16:33 | |
edleafe | mriedem: Not sure I understand the reason for a separate doc. All the contributor docs are in one place: http://logs.openstack.org/24/614024/19/check/openstack-tox-docs/76a4e0c/html/contributor/index.html. What would be the advantage of replacing the last section with a link? | 16:34 |
*** dims has quit IRC | 16:37 | |
*** dims has joined #openstack-placement | 16:41 | |
mriedem | edleafe: it's just a subdoc | 16:41 |
mriedem | https://github.com/openstack/nova/tree/master/doc/source/contributor | 16:41 |
mriedem | so the index page doesn't hold everything ever, it's for organizing | 16:42 |
edleafe | mriedem: yeah, I can see why that's better, but that sort of refactoring isn't what this patch is doing. Placement isn't set up like that (not yet, anyway) https://github.com/openstack/placement/tree/master/doc/source/contributor | 16:43 |
*** cdent has joined #openstack-placement | 16:46 | |
mriedem | it's not a refactor | 16:47 |
mriedem | create doc/source/contributor/db_schema_migrations.rst | 16:47 |
mriedem | you just add a section to index.rst that ref's the db_schema_migrations doc | 16:47 |
mriedem | don't get hung up on that comment, i can just refactor it later in a follow up, just fix the content issues | 16:48 |
edleafe | It would be nice to refactor those docs like nova's are, but that can be done later. | 16:49 |
*** mriedem has quit IRC | 16:53 | |
openstackgerrit | Ed Leafe proposed openstack/placement master: Add a document for creating DB revisions https://review.openstack.org/614024 | 17:19 |
edleafe | mrieden: ^^ cleaned up | 17:20 |
openstackgerrit | Chris Dent proposed openstack/placement master: Adapt placement database fixture for external use https://review.openstack.org/617935 | 17:32 |
cdent | edleafe: that ^ is probably going to conflict with the alembic changes, but not in any horrible way, and it will come out in the wash cleanly enough | 17:35 |
cdent | and now it is dinner time | 17:35 |
edleafe | cdent: Remember you have to drink my share of beer, too | 17:36 |
cdent | i know, I had a very large one yesterday, does that help? | 17:37 |
edleafe | a little <sob!> | 17:37 |
cdent | I'LL TRY HARDER | 17:38 |
* cdent waves | 17:38 | |
*** cdent has quit IRC | 17:38 | |
edleafe | o/ | 17:38 |
openstackgerrit | Ed Leafe proposed openstack/placement master: Consider root id is None in the database case https://review.openstack.org/613320 | 18:16 |
*** tetsuro has joined #openstack-placement | 20:23 | |
*** tetsuro has quit IRC | 20:25 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Adapt placement database fixture for external use https://review.openstack.org/617935 | 22:19 |
*** e0ne has joined #openstack-placement | 22:35 | |
*** e0ne has quit IRC | 22:36 | |
*** e0ne has joined #openstack-placement | 23:02 | |
*** e0ne has quit IRC | 23:35 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!