| openstackgerrit | Sam Morrison proposed openstack/nova master: Allow ability for non admin users to use all filters on server list. https://review.openstack.org/526558 | 00:04 |
|---|---|---|
| *** mriedem has quit IRC | 00:15 | |
| *** mriedem has joined #openstack-placement | 00:19 | |
| mriedem | cdent: pep8 gonna getcha | 00:19 |
| cdent | argh | 00:19 |
| cdent | because I'm an idiot, I have managed to make the functional tests pass in the EdLeafe repo: edleafe, efried | 00:29 |
| cdent | will annotate a pull request or something, but it is not for merging | 00:30 |
| mriedem | so i'd be +2 on the reshaper api | 00:35 |
| mriedem | pep8 import order fix needed obviously | 00:35 |
| cdent | mriedem: if you're gung ho, please feel free | 00:35 |
| mriedem | i'll +1/+2 once that's done and then it can stick through rebases | 00:35 |
| mriedem | ok | 00:35 |
| cdent | I'm up way past may bedtime | 00:36 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: [placement] Add /reshaper handler for POST https://review.openstack.org/576927 | 00:36 |
| mriedem | yup | 00:36 |
| mriedem | +2 | 00:36 |
| mriedem | assuming efried will rebase in the morning | 00:37 |
| *** tetsuro_ has quit IRC | 00:40 | |
| *** Nel1x has joined #openstack-placement | 00:45 | |
| cdent | efried, edleafe: for whenever you wanna see a huge diff: https://github.com/EdLeafe/placement/pull/1 | 00:48 |
| cdent | goodnight all | 00:51 |
| *** cdent has quit IRC | 00:51 | |
| *** tetsuro has joined #openstack-placement | 00:54 | |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Adds view builders for keypairs controller https://review.openstack.org/347289 | 01:12 |
| *** mriedem has quit IRC | 01:48 | |
| *** deepak_mourya__ has joined #openstack-placement | 02:21 | |
| *** alex_xu has joined #openstack-placement | 02:27 | |
| *** dansmith has joined #openstack-placement | 02:28 | |
| *** deepak_mourya_ has quit IRC | 02:29 | |
| *** htimsnad has quit IRC | 02:29 | |
| *** deepak_mourya__ is now known as deepak_mourya_ | 02:29 | |
| openstackgerrit | zhufl proposed openstack/nova master: Blacklist test_create_server_with_tags for nova-cells-v1 https://review.openstack.org/595399 | 02:37 |
| *** Nel1x has quit IRC | 02:53 | |
| openstackgerrit | Chen proposed openstack/nova master: Normalize dashless 'resource provider create' uuid https://review.openstack.org/567191 | 03:24 |
| *** nicolasbock has quit IRC | 03:44 | |
| openstackgerrit | Chen proposed openstack/nova master: Fix create_resource_provider docstring https://review.openstack.org/595453 | 04:12 |
| openstackgerrit | melanie witt proposed openstack/nova master: Correct the release notes related to nova-consoleauth https://review.openstack.org/595455 | 04:57 |
| *** lei-zh has joined #openstack-placement | 05:40 | |
| *** lei-zh1 has joined #openstack-placement | 05:55 | |
| *** lei-zh has quit IRC | 05:58 | |
| openstackgerrit | zhufl proposed openstack/nova master: Blacklist test_create_server_with_tags for nova-cells-v1 https://review.openstack.org/595399 | 06:50 |
| *** e0ne has joined #openstack-placement | 07:13 | |
| *** lei-zh1 has quit IRC | 07:22 | |
| *** lei-zh1 has joined #openstack-placement | 07:32 | |
| *** tssurya has joined #openstack-placement | 07:39 | |
| *** cdent has joined #openstack-placement | 08:20 | |
| openstackgerrit | Zhenyu Zheng proposed openstack/nova master: Only lock when race condition happens in context.set_target_cell https://review.openstack.org/595530 | 08:25 |
| *** sean-k-mooney has joined #openstack-placement | 08:46 | |
| *** tetsuro has quit IRC | 08:47 | |
| openstackgerrit | Zhenyu Zheng proposed openstack/nova master: Only lock when race condition happens in context.set_target_cell https://review.openstack.org/595530 | 09:21 |
| *** lei-zh1 has quit IRC | 09:31 | |
| openstackgerrit | Chris Dent proposed openstack/nova master: [placement] Add functional test to verify presence of policy https://review.openstack.org/595559 | 09:39 |
| cdent | gibi, efried: you prolly like that ^ | 09:40 |
| *** tetsuro has joined #openstack-placement | 10:30 | |
| *** tetsuro has quit IRC | 10:40 | |
| *** tetsuro_ has joined #openstack-placement | 10:40 | |
| *** tetsuro has joined #openstack-placement | 10:50 | |
| *** tetsuro_ has quit IRC | 10:50 | |
| *** tetsuro has quit IRC | 10:51 | |
| *** tetsuro_ has joined #openstack-placement | 10:51 | |
| gibi | cdent: looking | 10:53 |
| *** tetsuro has joined #openstack-placement | 10:54 | |
| *** tetsuro_ has quit IRC | 10:54 | |
| cdent | dansmith: have you seen https://review.openstack.org/#/c/582899/ ? It's got a "mostly I'd like Dan Smith to review this patch" in it from jay | 10:59 |
| cdent | gibi, efried : this looks like an easy win: https://review.openstack.org/#/c/588470/ it already has a +1 from efried, who was looking for a bit of explanation, I guess, before doing a +2 | 11:01 |
| gibi | cdent: the policy functional test looks good to me | 11:03 |
| gibi | cdent: also +2 on https://review.openstack.org/#/c/588470/ , but I let efried to send it through | 11:06 |
| cdent | cool, thanks (on both) | 11:06 |
| cdent | gibi: another if you're in the mood: this fixes a race in aggregate management: https://review.openstack.org/#/c/592654/ | 11:09 |
| *** nicolasbock has joined #openstack-placement | 11:22 | |
| openstackgerrit | Moshe Levi proposed openstack/nova master: libvirt: set vfio driver in interface hostdev https://review.openstack.org/595592 | 11:33 |
| *** mriedem has joined #openstack-placement | 11:33 | |
| *** jaypipes has joined #openstack-placement | 11:42 | |
| cdent | mriedem: https://review.openstack.org/595559 is an auto-confirm policy is there tester | 11:46 |
| cdent | if you'd like to have a look | 11:46 |
| mriedem | will do | 11:47 |
| jaypipes | good morning y'all. | 11:48 |
| * cdent waves | 11:48 | |
| cdent | are you back back jaypipes or only a little back? | 11:49 |
| jaypipes | cdent: fully back. | 11:49 |
| gibi | jaypipes: welcome back | 11:50 |
| jaypipes | gibi: Kösz | 11:51 |
| gibi | jaypipes: :) | 11:56 |
| *** dims_ is now known as dims | 12:37 | |
| openstackgerrit | Merged openstack/nova master: [placement] Regex consts for placement schema https://review.openstack.org/591863 | 12:45 |
| openstackgerrit | melanie witt proposed openstack/nova master: Correct the release notes related to nova-consoleauth https://review.openstack.org/595455 | 12:54 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge keypair extension response into server view builder https://review.openstack.org/584748 | 13:06 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge server usage extension response into server view builder https://review.openstack.org/585262 | 13:06 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge security groups extension response into server view builder https://review.openstack.org/585475 | 13:06 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge extended_status extension response into server view builder https://review.openstack.org/592092 | 13:06 |
| edleafe | cdent: Besides the sheer size, why is that PR a DNM? | 13:20 |
| cdent | edleafe: because it does things like cp -r nova/conf placement/conf without trimming | 13:20 |
| edleafe | cdent: ok. Haven't had time to fully go over it | 13:21 |
| edleafe | External monitor died | 13:21 |
| cdent | it's essentially analysis to find out what matters so we can then do what matters, cleanly | 13:21 |
| gibi | jaypipes, cdent: I think I found a possible race while I read https://review.openstack.org/#/c/592654 It is not super related to the patch itself but I wanted to check with you before I +W that | 13:21 |
| cdent | which seems more friendly to our descendents | 13:21 |
| cdent | edleafe: you have the worst luck with hardware | 13:22 |
| cdent | gibi: you have the best luck with finding bugs | 13:22 |
| edleafe | gibi: wanna trade? :) | 13:22 |
| gibi | edleafe: I don't use any external monitor, so I wouldn't loose to much in the trade ;) | 13:23 |
| efried | cdent, gibi: +A on the trivial refactor. +2 on the policy check, but want to let mriedem +A it. | 13:34 |
| efried | jaypipes: Welcome back. | 13:34 |
| efried | jaypipes: You have quite a bit of interesting reading waiting for you :) | 13:37 |
| cdent | edleafe, efried : made it go: curl http://ds1:8000/ \n {"versions": [{"id": "v1.0", "max_version": "1.29", "min_version": "1.0", "status": "CURRENT", "links": [{"rel": "self", "href": ""}]}]} | 13:40 |
| efried | woot! | 13:41 |
| edleafe | excellent! | 13:41 |
| efried | I'm gonna rebase the series now cdent, mriedem, jaypipes - cool? | 13:44 |
| cdent | efried: reshaper? yes please and thank you | 13:44 |
| cdent | edleafe, efried : all unit tests, all functional tests, all placecat tests are passing | 13:45 |
| * edleafe bows down to cdent | 13:45 | |
| openstackgerrit | Eric Fried proposed openstack/nova master: reshaper: Look up provider if not in inventories https://review.openstack.org/585033 | 13:45 |
| openstackgerrit | Eric Fried proposed openstack/nova master: Make get_allocations_for_resource_provider raise https://review.openstack.org/584598 | 13:45 |
| openstackgerrit | Eric Fried proposed openstack/nova master: Report client: Real get_allocs_for_consumer https://review.openstack.org/584599 | 13:45 |
| openstackgerrit | Eric Fried proposed openstack/nova master: Report client: get_allocations_for_provider_tree https://review.openstack.org/584648 | 13:45 |
| openstackgerrit | Eric Fried proposed openstack/nova master: Report client: _reshape helper, placement min bump https://review.openstack.org/585034 | 13:45 |
| openstackgerrit | Eric Fried proposed openstack/nova master: Report client: update_from_provider_tree w/reshape https://review.openstack.org/585049 | 13:45 |
| openstackgerrit | Eric Fried proposed openstack/nova master: Compute: Handle reshaped provider trees https://review.openstack.org/576236 | 13:45 |
| efried | that's awesome cdent | 13:46 |
| cdent | some major gaps to fill in terms of cleanliness and making things go properly, but the things that need to happen are pretty clear | 13:46 |
| efried | cdent: I'll ask again: how clean/perfect do we need to be before seeding the official repo? | 13:46 |
| cdent | efried: I'm not too concerned about perfection. what I'm more worried about is inconsequential junk being in the way | 13:47 |
| edleafe | If the tests are passing, I think we should do the rest of the work in the official repo | 13:47 |
| efried | (efried, why the hurry? I'm glad you asked. So I don't have to touch github. No other reason.) | 13:47 |
| efried | I would be fine with ^ | 13:47 |
| mriedem | note the nova meeting is in 13 minutes so it would be nice to summarize the activity on the separate git repo for those of us not following along that closely | 13:47 |
| mriedem | i'm mostly interested in preserving git history | 13:47 |
| cdent | I'd rather that we never merge my current patches because they are ... chunky | 13:47 |
| edleafe | mriedem: yeah, that was what involved some work. But it's there | 13:48 |
| mriedem | which it sounds like you've tried to do and then are having to push deltas on top? | 13:48 |
| cdent | if we can pick and choose (as humans) from the bits we like and make something cleaner I think that would be better | 13:48 |
| edleafe | cdent: what do you suggest to deal with your chunkiness? | 13:48 |
| mriedem | nicely worded | 13:49 |
| cdent | reason 1 for not using what we've alread done is that some of the files in it had to be copied in raw from a nova repo, so do it not have history. we need to resolve that to make a better filter | 13:49 |
| cdent | and then redo the work that I've done (in a more clean fashion) on top of that | 13:49 |
| cdent | so with regard to my chunkiness it would be useful for people to gaze upon it and see which parts they like | 13:50 |
| cdent | and understand it fully | 13:50 |
| cdent | so that we can then parcel it out to all who are interested | 13:50 |
| mriedem | is it not possible to cherry-pick stuff if there are deltas? | 13:50 |
| cdent | (i'm not picky) | 13:50 |
| cdent | mriedem: that would be ideal if the base was good, but the base is not good | 13:51 |
| cdent | if we make a good base we can do some interactive cherry picking from my branches | 13:51 |
| cdent | becasue quite a lot off stuff is just renaming imports and mocks | 13:51 |
| mriedem | is this because the dir structure needed to move up to /placement instead of /nova? | 13:51 |
| cdent | if people want, I can just do all that, but I was hoping for at least efried and edleafe to be involved so there's more breadth of understanding of the process involved (not just the changes themselves) | 13:52 |
| cdent | mriedem: yes, pretty much | 13:52 |
| cdent | but there's also some decsion that edleafe made in the moving things around that we might want to discussion (edleafe see the etherpad about the 'api' package and whether we want that) | 13:53 |
| edleafe | cdent: I can add the files you copied to the extraction script, and re-run it | 13:53 |
| cdent | edleafe: that woul be an excellent start, but if we can first figure out which conf bits we actually need (instead of just saying all of them) that might be tidier (it doesn't _matter_, but is nice) | 13:53 |
| edleafe | yeah, I saw that. I can redo the directory shuffling | 13:54 |
| edleafe | It doesn' | 13:54 |
| edleafe | It doesn't mater to me if that is done in the private repo or on the openstack repo | 13:54 |
| edleafe | (the conf bits) | 13:55 |
| cdent | another open issue (that I'm hacking around right now with stuff from placedock) is db sync | 13:56 |
| cdent | but we can also fix that later if we like | 13:56 |
| edleafe | I think that stuff like the db sync is fine to do in the openstack repo | 13:56 |
| efried | gibi: Re https://review.openstack.org/#/c/592654/ are you holding +W until a test/bug has been written for the race condition? (Assume it would be a separate patch regardless) | 13:56 |
| edleafe | It's part of the process | 13:56 |
| *** takashin has joined #openstack-placement | 13:57 | |
| gibi | efried: I wanted to check with Jay about that race before I +A but you are right I can file a bug instead | 13:58 |
| gibi | efried: +A-d | 13:59 |
| efried | cool, thx | 14:00 |
| jaypipes | sorry, was at breakfast. reading back. | 14:02 |
| efried | gibi: Also, I think we're ready to land https://review.openstack.org/#/c/590041/ | 14:02 |
| gibi | efried: added to my list | 14:03 |
| efried | thx | 14:03 |
| cdent | edleafe, efried : https://github.com/EdLeafe/placement/pull/2 has the latest work. | 14:06 |
| efried | ack | 14:06 |
| openstackgerrit | Merged openstack/nova master: Merge extended server attributes extension response https://review.openstack.org/584590 | 14:06 |
| edleafe | cdent: ack. Will take a look soon | 14:06 |
| efried | cdent: Attending nova mtg? | 14:06 |
| cdent | yeah, i'm there | 14:06 |
| jaypipes | gibi: addressed your concerns in the single-shot patch. | 14:07 |
| jaypipes | efried: I've read most of it already. | 14:07 |
| jaypipes | efried: just haven't had a chance to respond to gibi's consumer generation ML post. | 14:08 |
| gibi | jaypipes: how _increment_provider_generation can ensure that the generation is correct my case if it doesn't get the expected consumer generation that was in the request? | 14:09 |
| jaypipes | gibi: only one thread would ever be able to update generation. the next thread that attempts to update the generation will get a failure. | 14:13 |
| gibi | jaypipes: ohh I see. the RP is passed as a parameter and that RP is read from the db _before_ the generation check was doen in the handler | 14:13 |
| jaypipes | right | 14:13 |
| jaypipes | gibi: this is how all the generation increment code works (for provider and consumer) | 14:13 |
| gibi | jaypipes: thanks. I think I got it now. The race I envisioned does not exist | 14:14 |
| jaypipes | gibi: np | 14:16 |
| openstackgerrit | Stephen Finucane proposed openstack/nova master: tests: Create functional libvirt test base class https://review.openstack.org/407055 | 14:25 |
| openstackgerrit | Stephen Finucane proposed openstack/nova master: tests: Move mocking to setUp https://review.openstack.org/595802 | 14:25 |
| *** takashin has left #openstack-placement | 15:01 | |
| openstackgerrit | Balazs Gibizer proposed openstack/nova master: Remove noisy DEBUG log https://review.openstack.org/595810 | 15:02 |
| openstackgerrit | Stephen Finucane proposed openstack/nova master: tests: Move mocking to setUp https://review.openstack.org/595802 | 15:09 |
| openstackgerrit | Stephen Finucane proposed openstack/nova master: tests: Create functional libvirt test base class https://review.openstack.org/407055 | 15:09 |
| openstackgerrit | Merged openstack/nova master: Fix nits in resource_provider.py https://review.openstack.org/588470 | 15:12 |
| openstackgerrit | Chen proposed openstack/nova master: Fix create_resource_provider docstring https://review.openstack.org/595453 | 15:31 |
| *** e0ne has quit IRC | 15:51 | |
| *** mriedem is now known as mriedem_sudsy | 16:04 | |
| *** ttsiouts has joined #openstack-placement | 16:08 | |
| dims | looks like the repo is fully pruned and has just commits for placement - nice start - https://github.com/EdLeafe/placement/graphs/contributors | 16:08 |
| openstackgerrit | Merged openstack/nova master: placement: use single-shot INSERT/DELETE agg https://review.openstack.org/592654 | 16:08 |
| *** mriedem_sudsy is now known as mriedem | 16:26 | |
| *** ttsiouts has quit IRC | 16:30 | |
| *** ttsiouts has joined #openstack-placement | 16:30 | |
| *** ttsiouts has quit IRC | 16:35 | |
| *** sean-k-mooney has quit IRC | 16:36 | |
| openstackgerrit | Dan Smith proposed openstack/nova master: Make instance_list perform per-cell batching https://review.openstack.org/593131 | 17:04 |
| openstackgerrit | Dan Smith proposed openstack/nova master: Record cell success/failure/timeout in CrossCellLister https://review.openstack.org/594265 | 17:04 |
| openstackgerrit | Dan Smith proposed openstack/nova master: Optimize global marker re-lookup in multi_cell_list https://review.openstack.org/594577 | 17:04 |
| *** mriedem is now known as mriedem_away | 17:36 | |
| openstackgerrit | Ken'ichi Ohmichi proposed openstack/nova master: tests: Create functional libvirt test base class https://review.openstack.org/407055 | 17:37 |
| *** tssurya has quit IRC | 17:52 | |
| efried | jaypipes: Can we talk about https://review.openstack.org/#/c/585049/12/nova/scheduler/client/report.py@1517 ? | 18:07 |
| jaypipes | efried: sure | 18:08 |
| efried | jaypipes: Cause I'm still not understanding what you'd like to see changed. | 18:08 |
| jaypipes | efried: I haven't seen the latest revision on that patch so if you've changed stuff, lemme look at that first. | 18:09 |
| efried | jaypipes: no, haven't changed anything, just rebases. | 18:09 |
| efried | oh, yeah, I've changed things since PS12, but not that thing. | 18:10 |
| jaypipes | efried: ack, ok. so my point is that instead of invalidating cache entries proactively like you're doing there (and re-fetching the new provider generations) that we only re-fetch reactively when needed. | 18:10 |
| efried | So wrap each of the calls at https://review.openstack.org/#/c/585049/12/nova/scheduler/client/report.py@1651 through 1655 in a retry-if-409? | 18:11 |
| jaypipes | efried: in other words, have a single code path that, when a 409 is seen that mentions a provider generation mismatch, then we re-fetch the current state of the provider. | 18:12 |
| efried | Okay, couple problems I have with that. | 18:12 |
| efried | Foremost being that it is not *always* appropriate to refetch the current state of the provider. | 18:13 |
| efried | I would even contend that it is barely appropriate in the reshape flow, and we could only get away with it there because we're supposed to be doing this with the world shut down | 18:13 |
| efried | Because I'm replacing aggs/traits with what I have in my hand. And in a normal periodic flow, if the generation changed from under me because some other entity changed aggs/traits since the last periodic, I don't want to just refetch and overwrite. | 18:15 |
| efried | (I've said this before, but in case I've managed not to say it to you, this is why I was pulling for reshaper to return a payload with the new generations) | 18:16 |
| efried | jaypipes: The other problem I have, less serious but still bugs me, is doubling the number of API calls. Especially when I *know* I need to do the refetch. | 18:17 |
| *** mriedem_away is now known as mriedem | 18:17 | |
| jaypipes | efried: even if the reshaper returned provider generations that doesn't mean another thread could not have changed things in between that time and the next time you went to update something on that provider, though, right? | 18:17 |
| efried | Correct, which would fail in the path below, which is what we would want. | 18:18 |
| efried | And which would get resolved the next time around | 18:18 |
| efried | Perhaps you're suggesting we do that - just remove the refetch entirely, let the thing below fail, and count on catching it the next time around (which would be on real compute startup, presumably). | 18:19 |
| jaypipes | efried: yes, that's exactly what I'm suggesting. sorry if I have not been clear. | 18:19 |
| efried | Which seems not good UX to me, because we're *guaranteed* that the operation will not complete properly for a large subset of uses. | 18:19 |
| jaypipes | efried: sorry, not on restart of the copmute service or startup. | 18:20 |
| jaypipes | efried: I'm just saying that instead of trying to proactively do cache invalidation that instead we just react whenever a 409 comes up. | 18:20 |
| efried | "just" makes it sound like that's easy to think about and do. It's not. | 18:21 |
| efried | However | 18:21 |
| jaypipes | efried: I've just seen numerous bugs pop up when developers try to handle caches like this instead of just reacting to an expiry event (which is basically what the 409 on generation conflict is) | 18:22 |
| efried | it's possible that the @retrying decorator on _update would do it for us | 18:22 |
| melwitt | I was about to ask that, if the usual retry mechanism would kick in in this case as well | 18:22 |
| melwitt | I don't know much about it but I had thought there was some auto-retry in the case of 409 elsewhere in general | 18:23 |
| jaypipes | melwitt: we don't have a decorator that specifically catches the provider generation conflict, unless I'm mistaken. | 18:23 |
| melwitt | *elsewhere in the placement api | 18:23 |
| jaypipes | ack | 18:23 |
| efried | added pretty recently with I3c5fbb18297db71e682fcddb5bf4536595d92383 | 18:23 |
| efried | jaypipes: ^ | 18:23 |
| melwitt | okay, yeah I thought I had seen it somewhere | 18:24 |
| efried | https://review.openstack.org/#/c/556669/ | 18:24 |
| efried | so thinking this through, what *might* happen if we take that refresh out of there, is: | 18:24 |
| efried | The first time through, the reshaper succeeds, and then the agg/trait update will fail. | 18:25 |
| jaypipes | efried: yeah, sorry, I was saying we didn't have a decorator specifically for that (ala safe_connect is a specific decorator that traps specific exceptions) | 18:25 |
| efried | this will cause ufpt to invalidate the cache and raise the conflict exception that the aforementioned retry is set up to look for. | 18:25 |
| efried | The retry will run _update again, which will attempt to run update_provider_tree again. | 18:26 |
| efried | This time, the reshape already being done, update_provider_tree ought to *not* raise ReshapeNeeded, and instead go through the normal code path, posting back just the agg/trait changes | 18:26 |
| efried | the only thing I'm not completely sure about is how this plays with multiple providers. | 18:26 |
| efried | ufpt is set up to keep trying for other providers, but I'm not sure how completely | 18:27 |
| jaypipes | I have the feeling we're talking past each other. | 18:27 |
| efried | E.g. if there are 4 tries total but 5 providers changing traits, could we run out of tries | 18:27 |
| jaypipes | efried: I'm certainly not suggesting wrapping the entire update_from_provider_tree() in a retrying decorator. | 18:28 |
| efried | it already is. | 18:28 |
| efried | around _update | 18:28 |
| efried | and that's where this is gonna go if I take out that refresh. | 18:28 |
| *** e0ne has joined #openstack-placement | 18:28 | |
| efried | Because I can't take out the refresh *and* add a retry further down in ufpt around just the agg/trait update calls. | 18:29 |
| efried | because refreshing and redriving is not (necessarily) the right thing to do at that point. | 18:29 |
| jaypipes | efried: I don't understand why a retrying decorator cannot be put around smaller code chunks in the update_from_provider_tree() code path. | 18:30 |
| efried | because we don't know why the generation changed. | 18:31 |
| efried | If, say, a cli user adds a trait to a provider in between upt and ufpt, that same bounce will happen at L1655 | 18:32 |
| efried | if we were to retry there, we would overwrite and that trait would go away and the user wouldn't know why. | 18:32 |
| jaypipes | all we care about is that it *did* change, though. again, it's a signal that something changed, so the client code needs to re-get the state it's examining and re-determine if it needs to re-send the updates it wanted to originally or abandon its efforts. | 18:32 |
| efried | yes, re-determine logic happens in upt, not in ufpt. | 18:32 |
| jaypipes | hmm | 18:32 |
| efried | which is why the retry is around _update, not narrower. | 18:32 |
| efried | definitely can't have things outside of upt attempting to guess whether to merge/replace/abort | 18:33 |
| efried | jaypipes: Sorry, I thought this was going to go quicker; I have to run. Resume tomorrow? And/or via the review? | 18:38 |
| jaypipes | efried: sure | 18:39 |
| *** efried is now known as efried_afk | 18:43 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova stable/rocky: libvirt: Don't react to VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED events https://review.openstack.org/595864 | 18:44 |
| *** e0ne has quit IRC | 18:46 | |
| cdent | 'night all | 19:02 |
| *** cdent has quit IRC | 19:02 | |
| openstackgerrit | melanie witt proposed openstack/nova master: Correct the release notes related to nova-consoleauth https://review.openstack.org/595455 | 19:21 |
| openstackgerrit | melanie witt proposed openstack/nova master: Correct the release notes related to nova-consoleauth https://review.openstack.org/595455 | 19:24 |
| openstackgerrit | melanie witt proposed openstack/nova master: Correct the release notes related to nova-consoleauth https://review.openstack.org/595455 | 19:37 |
| *** jaypipes has quit IRC | 19:46 | |
| *** jaypipes has joined #openstack-placement | 19:47 | |
| openstackgerrit | Merged openstack/nova stable/rocky: block_device: Rollback volumes to in-use on DeviceDetachFailed https://review.openstack.org/590801 | 20:17 |
| openstackgerrit | Merged openstack/nova stable/queens: block_device: Rollback volumes to in-use on DeviceDetachFailed https://review.openstack.org/590803 | 20:17 |
| openstackgerrit | melanie witt proposed openstack/nova stable/rocky: Correct the release notes related to nova-consoleauth https://review.openstack.org/595890 | 20:55 |
| openstackgerrit | melanie witt proposed openstack/nova stable/rocky: Correct the release notes related to nova-consoleauth https://review.openstack.org/595890 | 20:57 |
| openstackgerrit | Surya Seetharaman proposed openstack/nova-specs master: Handling a down cell https://review.openstack.org/595892 | 21:03 |
| openstackgerrit | Merged openstack/nova master: Normalize dashless 'resource provider create' uuid https://review.openstack.org/567191 | 21:16 |
| openstackgerrit | Merged openstack/nova master: libvirt: Don't react to VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED events https://review.openstack.org/594508 | 21:30 |
| openstackgerrit | Merged openstack/nova master: Update contributor guide for Stein https://review.openstack.org/591258 | 21:30 |
| openstackgerrit | Merged openstack/nova master: [placement] Add functional test to verify presence of policy https://review.openstack.org/595559 | 21:30 |
| *** mriedem is now known as mriedem_afk | 21:43 | |
| openstackgerrit | Merged openstack/nova master: Add functional test for affinity with multiple cells https://review.openstack.org/585073 | 22:50 |
| *** jaypipes has quit IRC | 23:33 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!