Earlier  
Posted Nick Remark
#openstack-nova - 2018-01-25
16:51:00 mriedem looking
16:51:04 mriedem why was it using sets?
16:51:14 gibi efried: done
16:51:16 mriedem and are there problems if we're not ensuring uniqueness now?
16:51:31 efried gibi Thanks! stephenfin ^
16:51:36 gibi stephenfin: sorry
16:51:47 stephenfin gibi: Nope, suits me :)
16:51:49 efried mriedem The only real reason it was using sets before was for O(1) operations.
16:51:59 stephenfin bauzas has been bugging me about this vGPU thing anyway
16:52:10 mriedem stephenfin: the vgpu stuff is all approved
16:52:14 efried mriedem And as super-double-insurance that we don't have cycles in the "tree". But that's prevented elsewise in a number of ways.
16:52:16 stephenfin Ha!
16:52:22 stephenfin I should work on bugs more often
16:52:25 mriedem god this provider tree stuff,
16:52:31 stephenfin Everyone does my work for me
16:52:43 mriedem all reminds me of some graph traversal stuff i wrote like 10 years ago at ibm
16:52:54 efried mriedem But the trees should be small enough that O(lognN) is no big deal, and is a rounding error anyway against all the placement API calls that feed it.
16:52:57 mriedem back when i was young and smart
16:53:20 efried Not sure how that makes me feel, considering I'm older than you (pretty sure)
16:53:36 mriedem <- 36
16:53:40 efried 41
16:53:44 mriedem jesus you're old
16:53:47 efried ikr
16:53:54 efried Man, when I was 36 I could do anything.
16:54:07 efried Now I sound like a bowl of rice krispies getting out of bed.
16:55:25 efried So anyway, I think I now have 8 patches at the mercy of zuul. Bottom one of that series has been thrashing around for three full days.
16:55:36 mriedem ooo nice osc-placement docs are published https://docs.openstack.org/osc-placement/latest/
16:57:40 mriedem and releasenotes, although the version isn't what i expected https://docs.openstack.org/releasenotes/osc-placement/unreleased.html
16:59:16 edmondsw stephenfin if you're looking for something to review... :) https://review.openstack.org/#/c/526094/
16:59:49 stephenfin edmondsw: Darn, I thought I'd got away with the powerVM stuff
17:00:01 edmondsw and they just reel ya back in...
17:00:07 edmondsw that's the last one
17:01:13 dansmith efried: looks like that set has a pretty healthy set of +2+W already, up to a patch with a -1 and no response
17:03:30 openstackgerrit Stephen Finucane proposed openstack/osc-placement master: Resolve nits from I552688b9ee32b719a576a7a9ed5e4d5aa31d7b3f https://review.openstack.org/537971
17:03:45 stephenfin mriedem, melwitt: Fancy pushing that through? ^
17:04:50 mriedem stephenfin: i know you love to `` everything
17:04:54 mriedem but that's going to look weird in a shell
17:05:56 stephenfin Yeah, that's a good point actually
17:06:04 stephenfin cliff should probably strip those
17:06:05 mriedem it's the best point i've ever made i think
17:07:11 openstackgerrit Stephen Finucane proposed openstack/osc-placement master: Resolve nits from I552688b9ee32b719a576a7a9ed5e4d5aa31d7b3f https://review.openstack.org/537971
17:10:01 mriedem stephenfin: you didn't even have nits in https://review.openstack.org/#/c/536858/ :)
17:10:26 stephenfin Let's call it muscle memory
17:10:45 stephenfin Or me not waiting for the CI to build docs before approving them 🙈
17:11:32 dansmith er, I guess that one is not part of the main stack
17:11:48 jlvillal stephenfin, dansmith mriedem: You may want to move your jobs out of the Zuul 'integrated' queue.
17:12:04 jlvillal We (Ironic) set our queue as 'ironic'
17:12:20 jlvillal I notice the 'integrated' queue is enormous at the moment.
17:12:25 jlvillal http://zuul.openstack.org/
17:12:29 dansmith jlvillal: um, what? doesn't that mean you don't co-gate with other projects?
17:12:44 jlvillal dansmith, Yeah.
17:12:54 mriedem kind of defeats the purpose of gating
17:12:59 dansmith doesn't that defeat the whole thing we're doing here?
17:13:00 dansmith yeah.
17:13:07 jlvillal Well we gate with our patches.
17:13:30 jlvillal But yeah, not with cinder, glance, nova, etc...
17:13:30 dansmith right, but it means you don't get tested against ours, and we don't get tested with yours
17:14:28 jlvillal True. We did that after we migrated over.
17:14:37 mriedem can someone turn on the sdague signal
17:14:43 jlvillal Not sure if the tradeoffs are worth it or not.
17:15:43 jlvillal Too bad they don't have a 'depend' on queue. Instead of everyone goes in the same bucket queue
17:17:17 dansmith jlvillal: depends-on between queues just makes it one queue no?
17:18:57 jlvillal dansmith, Not a gerrit depends-on for one.
17:19:29 jlvillal dansmith, Now if 'ironic' depends on 'nova'. And 'glance' depends on 'nova' But ironic and glance don't depend on each other.
17:20:01 jlvillal dansmith, So in the gate queue a failed glance wouldn't cause the ironic job to get re-queued.
17:20:32 jlvillal dansmith, They do not have this idea of a 'depend' in zuul queues. There is just a queue.
17:20:42 jlvillal It was a random idea.
17:21:03 dansmith I see what you mean, but I'm not sure the benefit is worth that complexity
17:21:54 jlvillal dansmith, Yeah probably right. I just saw your nova jobs that have been 18 hours + in the gate queue. So that is why I mentioned it.
17:22:11 dansmith yeah, we're definitely feeling that pain, but..
17:34:49 mriedem gibi: once the two approved versioned notification transformation changes are merged https://review.openstack.org/#/q/topic:bp/versioned-notification-transformation-queens+status:open i'm going to close that bp and open another one for rocky for anything left
17:38:06 mriedem so is bauzas out today?
17:38:22 dansmith apparently
17:38:38 dansmith he's not on our calendar as being out, but I've pinged him a few times yesterday and today already and haven't seen any response
17:39:03 dansmith he was doing some conf thing earlier in the week, but I'm surprised he hasn't popped back up yet
17:48:15 mriedem oh right snowbunnies
17:55:23 jackie-truong dansmith: RE: https://review.openstack.org/#/c/537728/
17:55:33 jackie-truong dansmith: I wasn't sure if we needed to reassign instance._context
17:55:44 jackie-truong dansmith: The more I look at it, I'm thinking "no"
17:55:49 jackie-truong dansmith: Should I remove?
17:56:11 dansmith jackie-truong: it's done on L342 why would you have to do it again?
17:56:37 jackie-truong dansmith: i was a little confused as to how things were getting passed around
17:56:57 dansmith jackie-truong: you're calling that method :)
17:57:13 dansmith jackie-truong: it's set on instance above, then you pass it to another method, so... it's still set
17:57:28 jackie-truong dansmith: kk i'll remove
17:59:14 jackie-truong outlook pop-ups ;-)
17:59:49 dansmith I figured you had an army of interns, one patch per intern, tasked with rapid-fire refreshing until midnight
18:00:08 jackie-truong nope, that was just me sitting on the couch, eating nachos
18:00:22 dansmith hah nice
18:05:05 openstackgerrit Jackie Truong proposed openstack/nova master: Reduce complexity of _from_db_object https://review.openstack.org/537728
18:05:26 dansmith jackie-truong: why do you keep pushing these up one by one? do you not have them in a proper stack?
18:06:57 jackie-truong dansmith: i don't think i know what i'm doing when it comes to tying the patches together for gerrit
18:07:16 dansmith jackie-truong: so you're doing what? rebasing the second one on the first, then the third on the second, etc?
18:07:19 dansmith all in separate branches?
18:07:24 mriedem git rebase -i
18:07:26 mriedem is your friend
18:07:28 dansmith yeah
18:07:34 mriedem rebase the entire series at once,
18:07:40 mriedem modify the patches you need
18:08:16 mriedem https://docs.openstack.org/infra/manual/developers.html#rebasing-a-commit

Earlier   Later