| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-11-20 | |||
| 12:22:52 | moshele | stephenfin: regarding the revert https://review.openstack.org/#/c/518833/ | |
| 12:23:03 | stephenfin | Yup? | |
| 12:23:23 | moshele | stephenfin: so I have a fix https://review.openstack.org/#/c/519066/ | |
| 12:23:35 | stephenfin | moshele: Excellent. I'll take a look at that today :) | |
| 12:23:50 | moshele | stephenfin: but I didn't find the guy to say if it not breaking heat | |
| 12:23:55 | efried | cdent Am I missing the part of the series where GET /allocation_candidates returns the new format expected by PUT /allocations ? | |
| 12:24:30 | stephenfin | moshele: Hmm, that probably wouldn't be a bad idea. ramishra was the nick. I wonder if he's on #heat? | |
| 12:25:28 | moshele | stephenfin: I was looking for him in openstack-heat :) | |
| 12:25:49 | stephenfin | Ahh, yeah. Look at the welcome message ;) | |
| 12:27:01 | moshele | stephenfin: feel free to review the patch I should solve the problem | |
| 12:27:14 | moshele | s/I/ti/ | |
| 12:27:21 | stephenfin | Will do (y) | |
| 12:29:56 | cdent | efried: line 286ish https://review.openstack.org/#/c/510626/14/nova/api/openstack/placement/handlers/allocation.py | |
| 12:30:20 | cdent | oh sorry, that’s the put you want to know the get | |
| 12:30:56 | cdent | it’s in the _allocations_dict method | |
| 12:31:10 | cdent | GET was already nearly correct, it just needed the project_id and user_id added | |
| 12:31:21 | cdent | GET being different was why PUT was changed to be more like GET | |
| 12:31:46 | efried | cdent I'm talking about GET /allocation*_candidates* | |
| 12:32:07 | cdent | ah, sorry, names are hard on a monday | |
| 12:33:06 | efried | cdent The latest changes to the spec were that the resource request bit of the GET /allocation_candidates response was going to be changed to match what PUT /allocations expects so that you can still just turn it around and PUT it without looking at it (and converting it) | |
| 12:34:19 | cdent | oh, yeah, I think that’s just not done yet, as that was decided right before I started travelling | |
| 12:34:25 | cdent | so I’ll stick a -W on there somewhere | |
| 12:35:02 | openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient master: [functional] Remove duplication of boot helper https://review.openstack.org/521071 | |
| 12:35:03 | openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient master: [ci] Use pseudo-random names for new resources https://review.openstack.org/479777 | |
| 12:35:20 | efried | cdent Okay, cool. Didn't want to embarrass my self by +1ing that. Dodged a bullet, phew! | |
| 12:36:40 | cdent | My memory is starting to surface: I updated the spec and volunteered you or jay to do the work, but then you and jay fell in the refactoring hole related to resources etc and so now it is back to me. | |
| 12:36:41 | cdent | Or something | |
| 12:37:11 | efried | cdent Yes, that's the long and short of it. Though refactoring isn't the *only* thing that happened - see update email 41.75 :) | |
| 12:37:26 | cdent | that’s the “etc” part | |
| 12:37:47 | cdent | speaking of: thanks for doing those. you can carry on doing it if you’re feeling keen, or I’m happy to take it back if not | |
| 12:40:12 | efried | cdent I'd be fine either way. I've been happily slinging code for virt drivers getting hands on ProviderTree, but that stuff (as well as granular resource request work) is high atop other series work, so not urgent. Whereas symmetric stuff is pretty close to done. | |
| 12:40:58 | efried | cdent So if you're buried under other stuff, I can take it on for the sake of expediency. But if you've got the bandwidth to finish it, you should. | |
| 12:41:00 | cdent | the “those” I meant was the weekly updates | |
| 12:41:12 | efried | Hah. Oh. | |
| 12:41:22 | efried | You're welcome - and yes, please take them back. | |
| 12:41:40 | cdent | roger that. and I’ll take back the symmetric stuff | |
| 12:41:47 | efried | cdent I made sure to leave you #42 :) | |
| 12:41:53 | cdent | so kind | |
| 12:42:18 | efried | cdent And speaking of taking stuff over - you feel like running the scheduler meeting today? | |
| 12:42:36 | efried | edleafe- is out | |
| 12:44:14 | cdent | oh is he now | |
| 12:44:23 | cdent | yeah, I can do it | |
| 12:46:17 | efried | cdent Cool | |
| 12:46:44 | efried | cdent Being an American in the UK, are you going to be out later this week or no? | |
| 12:46:46 | cdent | just to check my calendar it is scheduled for a 1.25 hours from now? | |
| 12:46:55 | cdent | no | |
| 12:47:17 | cdent | Usually I take holiday back to the US to see family, but with sydney just now I couldn’t stand the idea of yet more international travel | |
| 12:47:51 | efried | cdent Okay. Yes, scheduler meeting is 1.25 hours from now if my UTC clock is sane. (DST is stupid) | |
| 12:48:01 | efried | cdent FYI I'll be out W-F | |
| 13:27:14 | niraj_singh | efried: i was going through ( https://review.openstack.org/#/c/512329/ ) this patch. A new exception is cretaed i.e. "ServiceUserTokenNoAuth". But i didn't find where this exception is handled. | |
| 13:36:10 | openstackgerrit | Takashi NATSUME proposed openstack/nova stable/pike: Add 'delete_host' command in 'nova-manage cell_v2' https://review.openstack.org/513717 | |
| 13:36:41 | efried | niraj_singh One sec. | |
| 13:37:54 | efried | niraj_singh Okay - what do you mean "handled"? | |
| 13:38:17 | efried | niraj_singh The motivation behind this patch was to make an exception that was easier for the user to understand. | |
| 13:38:32 | efried | niraj_singh So it's not actually supposed to be handled anywhere - it bubbles up to the user. | |
| 13:40:33 | efried | niraj_singh Sorry, by which I mean, it ends up in the logs where the operator can see it. | |
| 13:46:49 | stephenfin | moshele: Done | |
| 13:47:20 | stephenfin | Even though I suggested it, I'm not sure whether we should log an exception message without raising it. I'll let someone else comment on this | |
| 13:48:53 | niraj_singh | efried: I thought it should handle any where. ok thanks for information. | |
| 13:54:03 | openstackgerrit | priyaduggirala proposed openstack/nova master: Rename parameters in call() of nova/image/glance.py https://review.openstack.org/508533 | |
| 13:57:41 | efried | Scheduler meeting in three minutes in #openstack-meeting-alt | |
| 13:57:46 | cdent | whoa | |
| 13:57:59 | efried | Did I just read your mind? | |
| 13:58:00 | cdent | I was, like, literally, typing, literally, like, the same, like thing | |
| 13:58:14 | efried | Great minds, and all that. | |
| 13:58:36 | cdent | we are merely expressions of the platonic nova scheduler dev | |
| 13:59:01 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: [placement] Add 'Location' parameters in API ref https://review.openstack.org/521541 | |
| 14:00:01 | efried | I had an almost-mother-in-law when I was a youngster who would always know the second half of pithy aphorisms, which was slightly annoying. In this case, "Great minds think alike" apparently concludes, "And fools seldom differ" | |
| 14:02:37 | stephenfin | sean-k-mooney: I'm +2 on this if you want to take a second look https://review.openstack.org/#/c/493082/ | |
| 14:03:43 | stephenfin | efried: nested resource providers are on my review radar for this week | |
| 14:03:58 | stephenfin | (RE: the review request you sent at some point last week :)) | |
| 14:04:06 | efried | stephenfin That's great to hear. Can you join the sched meeting in -alt ? | |
| 14:04:13 | stephenfin | Now? | |
| 14:04:16 | efried | now | |
| 14:05:33 | cdent | +1 | |
| 14:13:19 | sean-k-mooney | stephenfin: just taking a look now | |
| 14:13:56 | sean-k-mooney | ah the printable mixin | |
| 14:14:19 | stephenfin | bauzas: This one's another easy +W, IMO https://review.openstack.org/#/c/466051/ | |
| 14:15:32 | bauzas | stephenfin: ack | |
| 14:16:46 | stephenfin | cdent: Seen this? https://review.openstack.org/#/c/511936/ Are we expecting to have a python-placementclient library? | |
| 14:17:36 | cdent | there’s going to be an osc plugin, but no plans for a specific client | |
| 14:17:56 | cdent | in large part because it’s “just json” | |
| 14:18:49 | cdent | stephenfin: https://review.openstack.org/#/q/topic:bp/placement-osc-plugin+(status:open+OR+status:merged) | |
| 14:19:45 | efried | cdent I would be in favor of a client lib that performs useful API calls, like SchedulerReportClient does, and parlays "just json" into useful python objects, like ProviderTree. | |
| 14:20:22 | cdent | I was really hoping we’d keep ProviderTree hidden in the bowels of nova and not make that a thing that people commonly want to use... | |
| 14:20:52 | cdent | But if that’s not the case, then sure, I guess. The deal simply was that until multiple implementations with commonanilities showed up, why bother? | |
| 14:21:07 | efried | It doesn't have to be ProviderTree itself, but in that vein. | |
| 14:21:38 | openstackgerrit | Merged openstack/os-vif master: Add VersionedObjectPrintable mixin https://review.openstack.org/493082 | |
| 14:21:49 | efried | cdent Oh, I can answer that one: so that we don't have multiple implementations with commonalities show up, which we then have to reconcile and consolidate. | |
| 14:22:14 | cdent | sigh | |
| 14:22:37 | cdent | reconciliation and consilidation is a _good_ thing. it allows for compare and contrast amongst a variety of thoughts | |
| 14:22:41 | cdent | rather than imposing a single vision | |
| 14:23:18 | efried | But then you wind up with things you can't get rid of because momentum. | |
| 14:23:38 | cdent | The way nova chooses to do things internal to the report client isn’t necessarily right for the rest of the world and vice versa | |
| 14:23:56 | cdent | and if you can’t just use requests for the API when we may as well give up and go home because we’ve failed | |
| 14:24:30 | efried | cdent This is interesting in light of some of the work I did last week. Let's chat after the meeting. | |
| 14:25:16 | cdent | (or at least by 3) | |
| 14:25:30 | efried | cdent Okay, well, if you get a chance, have a look at this: https://review.openstack.org/#/c/521098/4/nova/scheduler/client/report.py | |
| 14:26:34 | efried | A non-trivial method to collate all the information about a group of providers. Call it ProviderTree or whatever, IMO it's conceivably a useful thing to be able to do for any client. | |
| 14:27:12 | cdent | In idealized world mode, the fact that we need a provider tree (or whatever) is a reason to go home | |