| *** cdent has quit IRC | 00:23 | |
| openstackgerrit | wanghongtao proposed openstack/nova master: Fix the metadata re to match the unicode https://review.openstack.org/536236 | 00:43 |
|---|---|---|
| *** edmondsw has quit IRC | 01:54 | |
| *** edmondsw has joined #openstack-placement | 01:55 | |
| openstackgerrit | tianhui proposed openstack/nova master: Fix bug to api-ref https://review.openstack.org/571375 | 01:59 |
| *** edmondsw has quit IRC | 01:59 | |
| openstackgerrit | Yikun Jiang (Kero) proposed openstack/nova master: Fix invalid raise in test_compute_mgr https://review.openstack.org/571610 | 02:25 |
| openstackgerrit | Merged openstack/nova stable/queens: libvirt: Skip fetching the virtual size of block devices https://review.openstack.org/571425 | 02:27 |
| *** nicolasbock has quit IRC | 03:04 | |
| *** edmondsw has joined #openstack-placement | 03:40 | |
| *** edmondsw has quit IRC | 03:44 | |
| openstackgerrit | Merged openstack/nova master: libvirt: configure trust mode for vfs https://review.openstack.org/458514 | 04:19 |
| *** e0ne has joined #openstack-placement | 04:21 | |
| *** e0ne has quit IRC | 04:30 | |
| *** e0ne has joined #openstack-placement | 05:02 | |
| *** e0ne has quit IRC | 05:16 | |
| *** edmondsw has joined #openstack-placement | 05:28 | |
| *** edmondsw has quit IRC | 05:33 | |
| *** ttsiouts has joined #openstack-placement | 06:37 | |
| *** ttsiouts has quit IRC | 06:55 | |
| *** ttsiouts has joined #openstack-placement | 06:59 | |
| *** tssurya has joined #openstack-placement | 07:07 | |
| *** ttsiouts has quit IRC | 07:19 | |
| *** edmondsw has joined #openstack-placement | 07:20 | |
| *** edmondsw has quit IRC | 07:24 | |
| openstackgerrit | Yikun Jiang (Kero) proposed openstack/nova master: Adapt _validate_instance_group_policy to new policy model https://review.openstack.org/571465 | 07:34 |
| *** ttsiouts has joined #openstack-placement | 08:14 | |
| *** ttsiouts has quit IRC | 08:17 | |
| *** ttsiouts has joined #openstack-placement | 08:22 | |
| openstackgerrit | Elod Illes proposed openstack/nova stable/pike: placement: Fix HTTP error generation https://review.openstack.org/571218 | 08:23 |
| openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: Initial change set of z/VM driver https://review.openstack.org/523387 | 08:47 |
| openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: Spawn and destroy function of z/VM driver https://review.openstack.org/527658 | 08:47 |
| openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: add snapshot function https://review.openstack.org/534240 | 08:47 |
| openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: add power actions https://review.openstack.org/543340 | 08:48 |
| openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: add get console output https://review.openstack.org/543344 | 08:48 |
| *** belmoreira has joined #openstack-placement | 09:01 | |
| *** edmondsw has joined #openstack-placement | 09:08 | |
| *** edmondsw has quit IRC | 09:13 | |
| openstackgerrit | jichenjc proposed openstack/nova master: not reraise DiskNotFound if instance is resized https://review.openstack.org/571410 | 09:19 |
| *** ttsiouts has quit IRC | 09:33 | |
| openstackgerrit | Stephen Finucane proposed openstack/nova-specs master: Add 'numa-aware-vswitches' spec https://review.openstack.org/541290 | 09:42 |
| *** ttsiouts has joined #openstack-placement | 09:44 | |
| *** ttsiouts has quit IRC | 10:01 | |
| *** nicolasbock has joined #openstack-placement | 10:55 | |
| *** edmondsw has joined #openstack-placement | 10:56 | |
| *** edmondsw has quit IRC | 11:01 | |
| *** ttsiouts has joined #openstack-placement | 11:06 | |
| *** ttsiouts has quit IRC | 11:16 | |
| *** ttsiouts has joined #openstack-placement | 11:18 | |
| openstackgerrit | Chen proposed openstack/nova master: WIP https://review.openstack.org/571472 | 11:25 |
| *** ttsiouts has quit IRC | 12:05 | |
| *** edmondsw has joined #openstack-placement | 12:11 | |
| openstackgerrit | Merged openstack/nova master: libvirt: add vf_trusted field for network metadata https://review.openstack.org/566343 | 12:12 |
| *** edleafe is now known as figleaf | 12:14 | |
| openstackgerrit | Merged openstack/nova master: metadata: add vf_trusted field to device metadata https://review.openstack.org/566344 | 12:16 |
| *** tssurya is now known as sususuryashines | 12:16 | |
| *** mriedem has joined #openstack-placement | 12:22 | |
| *** gibi is now known as giblet | 13:06 | |
| *** stephenfin is now known as finucannot | 13:06 | |
| *** jaypipes is now known as leakypipes | 13:18 | |
| *** bauzas is now known as PapaOurs | 13:25 | |
| *** mriedem is now known as hansmoleman | 13:25 | |
| *** efried is now known as fried_rice | 13:26 | |
| openstackgerrit | Stephen Finucane proposed openstack/nova-specs master: Add 'numa-aware-vswitches' spec https://review.openstack.org/541290 | 13:31 |
| *** dansmith is now known as superdan | 13:35 | |
| *** leakypipes is now known as MurderTheLeafBlo | 13:36 | |
| *** MurderTheLeafBlo is now known as leakypipes | 13:36 | |
| *** jroll is now known as jrollinhatin | 13:36 | |
| openstackgerrit | Tsuyoshi Nagata proposed openstack/nova master: nova improvement of maximum attach volumes more than 26 vols https://review.openstack.org/567472 | 13:40 |
| *** ttsiouts has joined #openstack-placement | 13:54 | |
| *** ttsiouts has quit IRC | 13:56 | |
| *** ttsiouts has joined #openstack-placement | 13:56 | |
| openstackgerrit | Jan Gutter proposed openstack/nova master: Use vif.vif_name in _set_config_VIFGeneric https://review.openstack.org/571461 | 13:57 |
| openstackgerrit | Jan Gutter proposed openstack/nova master: Convert vrouter legacy plugging to os-vif https://review.openstack.org/571325 | 13:57 |
| openstackgerrit | Jan Gutter proposed openstack/nova master: Pass virtio multiqueue info to os-vif plugins https://review.openstack.org/571462 | 13:57 |
| *** e0ne has joined #openstack-placement | 13:59 | |
| *** belmorei_ has joined #openstack-placement | 14:01 | |
| *** ttsiouts_ has joined #openstack-placement | 14:01 | |
| *** ttsiouts has quit IRC | 14:01 | |
| *** belmoreira has quit IRC | 14:02 | |
| openstackgerrit | Andrey Volkov proposed openstack/nova master: Test Compute API in multiple cells https://review.openstack.org/547273 | 14:03 |
| *** melwitt is now known as jgwentworth | 14:09 | |
| *** e0ne has quit IRC | 14:21 | |
| openstackgerrit | Merged openstack/nova master: network: update pci request spec to handle trusted tags https://review.openstack.org/458820 | 14:32 |
| openstackgerrit | Tsuyoshi Nagata proposed openstack/nova master: nova improvement of maximum attach volumes more than 26 vols https://review.openstack.org/567472 | 14:56 |
| *** e0ne has joined #openstack-placement | 14:56 | |
| *** e0ne has quit IRC | 15:09 | |
| *** ttsiouts_ has quit IRC | 15:21 | |
| openstackgerrit | Balazs Gibizer proposed openstack/nova master: Add bandwidth related standard resource classes https://review.openstack.org/570847 | 15:22 |
| openstackgerrit | Balazs Gibizer proposed openstack/nova master: Transfer port.resource_request to the scheduler https://review.openstack.org/567268 | 15:22 |
| openstackgerrit | Balazs Gibizer proposed openstack/nova master: Send resource allocations in the port binding https://review.openstack.org/569459 | 15:22 |
| *** sususuryashines has quit IRC | 15:24 | |
| *** ttsiouts has joined #openstack-placement | 15:25 | |
| *** ttsiouts has quit IRC | 15:30 | |
| *** ttsiouts has joined #openstack-placement | 15:30 | |
| *** ttsiouts has quit IRC | 15:31 | |
| openstackgerrit | Artom Lifshitz proposed openstack/nova master: Refactor _build_device_metadata https://review.openstack.org/533804 | 15:32 |
| openstackgerrit | Artom Lifshitz proposed openstack/nova master: Consider hostdev devices when building metadata https://review.openstack.org/533805 | 15:32 |
| *** belmorei_ has quit IRC | 15:40 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Restrict CONF.quota.driver to DB and noop quota drivers https://review.openstack.org/410996 | 15:44 |
| openstackgerrit | Lee Yarwood proposed openstack/nova stable/ocata: libvirt: handle DiskNotFound during update_available_resource https://review.openstack.org/571432 | 16:02 |
| openstackgerrit | Lee Yarwood proposed openstack/nova stable/ocata: libvirt: Skip fetching the virtual size of block devices https://review.openstack.org/571433 | 16:02 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Trim the fat on HostState.instances https://review.openstack.org/569247 | 16:17 |
| openstackgerrit | Sylvain Bauza proposed openstack/nova-specs master: Proposes Multiple GPU types https://review.openstack.org/557065 | 16:26 |
| *** PapaOurs is now known as bauzas | 16:27 | |
| *** fried_rice is now known as fried_rolls | 16:29 | |
| *** hansmoleman is now known as hans_lunch | 16:40 | |
| *** cdent has joined #openstack-placement | 16:50 | |
| *** nicolasbock has quit IRC | 16:56 | |
| openstackgerrit | Dan Smith proposed openstack/nova master: Use oslo.messaging per-call monitoring https://review.openstack.org/566696 | 17:02 |
| *** nicolasbock has joined #openstack-placement | 17:03 | |
| openstackgerrit | Merged openstack/nova master: libvirt: place emulator threads on CONF.compute.cpu_shared_set https://review.openstack.org/510897 | 17:25 |
| *** fried_rolls is now known as fried_rice | 17:33 | |
| *** ttsiouts has joined #openstack-placement | 17:51 | |
| *** e0ne has joined #openstack-placement | 18:00 | |
| fried_rice | leakypipes: I thought we had a big lock preventing claims from happening during rt update | 18:24 |
| *** e0ne has quit IRC | 18:31 | |
| *** e0ne has joined #openstack-placement | 18:31 | |
| *** e0ne has quit IRC | 18:32 | |
| *** e0ne has joined #openstack-placement | 18:35 | |
| *** e0ne has quit IRC | 18:39 | |
| leakypipes | fried_rice: rt update? | 18:53 |
| fried_rice | leakypipes: Yeah, the periodic task wherein update_from_provider_tree is being run. | 18:54 |
| leakypipes | fried_rice: we might lock the provider tree object, but that won't prevent the scheduler from allocating against any of the providers... | 18:54 |
| leakypipes | fried_rice: we need to do the inventory moves and allocation moves (for lots of consumers) in a single transaction. | 18:55 |
| leakypipes | fried_rice: that's the problem domain, AFAIU | 18:55 |
| fried_rice | leakypipes: If this was purely a placement problem, agreed; but it's not. If we were synchronized as I thought we were on the nova side, we would be okay. | 18:55 |
| fried_rice | leakypipes: If that's not the case, then we already have this race condition in spades. | 18:56 |
| fried_rice | leakypipes: Because update_from_provider_tree operates via multiple separate placement calls to flush changes. | 18:56 |
| fried_rice | leakypipes: Help me understand: it's n-cpu that's invoking update_from_provider_tree, but n-sch that's doing GET /a_c and PUT allocations? | 18:58 |
| *** hans_lunch is now known as mriedem | 18:59 | |
| leakypipes | fried_rice: update_from_provider_tree never touches allocations. | 18:59 |
| fried_rice | I understand that's the case today, yes. | 18:59 |
| leakypipes | fried_rice: which is why there are no current race conditions with it. | 18:59 |
| *** ttsiouts has quit IRC | 19:00 | |
| fried_rice | Still a problem with allocations (from the scheduler) straddling changes to providers/inventories during update_from_provider_tree though. | 19:00 |
| fried_rice | Once we have consumer generations, if both sides are trying to muck with allocations on the same instance, they'll bounce "naturally". Having trouble seeing it as a different issue (vs. a tweak on the same issue). | 19:02 |
| leakypipes | fried_rice: is that you that is light blue in the etherpad? | 19:02 |
| fried_rice | I haven't looked at the etherpad yet, so no. | 19:02 |
| leakypipes | cdent: are you the light blue? | 19:03 |
| cdent | yeah | 19:03 |
| leakypipes | got it, thanks. | 19:03 |
| leakypipes | cdent: strange that the line numbers are different between us. | 19:03 |
| cdent | yes not sure what's up with that | 19:03 |
| cdent | i've pasted an alternate allocations dict at the bottom | 19:04 |
| *** openstackgerrit has quit IRC | 19:04 | |
| leakypipes | cdent: ack, thx | 19:05 |
| fried_rice | leakypipes: So your suggestion effectively dovetails with mine, right? The virt driver still needs to provide the information; the difference is that update_from_provider_tree needs to do the allocation & inventory move in this single-transaction API call you've proposed, rather than stepwise as I described. | 19:07 |
| leakypipes | fried_rice: yep, that's the different in a nutshell. | 19:08 |
| fried_rice | ack | 19:08 |
| leakypipes | I'm pretty positive superdan is going to throw up when he sees the proposal for another HTTP endpoint. | 19:10 |
| cdent | can we put a "you must be this tall to ride this URI" on this thing? | 19:10 |
| superdan | I dunno what we're talking about, but barf | 19:12 |
| leakypipes | hehe | 19:13 |
| * cdent passes superdan a tissue | 19:13 | |
| * superdan wipes on his sleeve like a barbarian | 19:13 | |
| * leakypipes passes superdan a superkerchief | 19:14 | |
| superdan | that's a sleeve yes? | 19:15 |
| * superdan has to run out for a bit | 19:15 | |
| leakypipes | superdan: don't you mean you have to fly out for a bit? I mean, when you have the power... | 19:16 |
| fried_rice | cdent, leakypipes: If we're looking for "the smallest thing that could possibly work" - I think we can avoid the race conditions if we implement a flag on the allocations endpoint that allows me to exceed total - reserved. | 19:16 |
| leakypipes | fried_rice: not really a fan of that approach. just means mucking with reserved amounts to get around the fact that we need to perform multiple operations in a single transaction. | 19:17 |
| fried_rice | The update_from_provider_tree algo starts off by effectively "locking" the inventories by setting reselved = total. Then we don't have to worry about the scheduler interfering while we're mucking with stuff. And we use that "overallocate" flag to move the allocations. Then set reserved back to the proper value when done. | 19:18 |
| fried_rice | leakypipes: True story, but it gives us time to implement a more comprehensive solution, which is going to take time. | 19:18 |
| leakypipes | fried_rice: that sounds more complicated and tribal knowledge-y than just adding an HTTP endpoint that does what we're trying to do in the first place? | 19:18 |
| cdent | that flag would have a less than great failure mode too (start the multi-step process, fails halfway through) | 19:19 |
| fried_rice | leakypipes: If we don't solve this upgrade problem in rocky, xen's vgpu nrp bp is dead in the water. | 19:19 |
| leakypipes | cdent: excellent point. | 19:19 |
| cdent | fried_rice's point is good too | 19:19 |
| leakypipes | fried_rice: I'm less concerned with that than I am with doing this right. | 19:19 |
| leakypipes | same for the multi-vgpu type stuf. | 19:20 |
| fried_rice | It will be very disappointing if nrp isn't usable in rocky. | 19:21 |
| leakypipes | fried_rice, cdent: if we can reach some rough agreement on the HTTP API, I can have the implementation of this hacked together this weekend. | 19:21 |
| fried_rice | I might have to be /nick efired | 19:22 |
| cdent | i've got to lunch, but will look back here and on the etherpad on when I return (I'm in PST today) | 19:22 |
| leakypipes | cdent: ack, thx | 19:23 |
| *** e0ne has joined #openstack-placement | 19:24 | |
| *** openstackgerrit has joined #openstack-placement | 19:44 | |
| openstackgerrit | Kashyap Chamarthy proposed openstack/nova master: libvirt: Remove usage of migrateToURI{2} APIs https://review.openstack.org/567258 | 19:44 |
| fried_rice | leakypipes, cdent: Proposed slight tweak to the format at the bottom of the etherpad. Glues together inv+allocs for one RC. | 20:06 |
| fried_rice | I think it handles splits and new inventory creation correctly. | 20:06 |
| superdan | leakypipes: my cape is in the shop | 20:14 |
| *** e0ne has quit IRC | 20:19 | |
| cdent | fried_rice, leakypipes I don't have strong opinion on the shape of things. fried_rice's has an advantage of being fairly distinct from any other structure we use, which in this case I think is useful | 20:28 |
| openstackgerrit | Merged openstack/nova master: Restrict CONF.quota.driver to DB and noop quota drivers https://review.openstack.org/410996 | 20:56 |
| openstackgerrit | Merged openstack/nova master: Fix invalid raise in test_compute_mgr https://review.openstack.org/571610 | 20:56 |
| openstackgerrit | Merged openstack/nova master: Implement granular policy rules for placement https://review.openstack.org/524425 | 21:06 |
| openstackgerrit | Jay Pipes proposed openstack/nova master: Add a microversion for consumer generation support https://review.openstack.org/565604 | 21:12 |
| *** edmondsw has quit IRC | 21:25 | |
| *** superdan is now known as dansmith | 21:25 | |
| openstackgerrit | Jay Pipes proposed openstack/nova master: placement: always create consumer records https://review.openstack.org/567678 | 21:31 |
| openstackgerrit | Jay Pipes proposed openstack/nova master: add consumers generation field https://review.openstack.org/557958 | 21:31 |
| openstackgerrit | Jay Pipes proposed openstack/nova master: placement: Allocation.consumer field https://review.openstack.org/565405 | 21:31 |
| openstackgerrit | Jay Pipes proposed openstack/nova master: rework allocation handler _allocations_dict() https://review.openstack.org/565407 | 21:31 |
| openstackgerrit | Jay Pipes proposed openstack/nova master: Add a microversion for consumer generation support https://review.openstack.org/565604 | 21:31 |
| *** mriedem has quit IRC | 21:42 | |
| openstackgerrit | Merged openstack/nova master: [placement] default to accept of application/json when */* https://review.openstack.org/568630 | 21:46 |
| openstackgerrit | Chris Dent proposed openstack/nova master: Extract part of PlacementFixture to placement https://review.openstack.org/568359 | 21:53 |
| fried_rice | leakypipes: Responded in etherpad (this is a weird way to have a conversation) | 22:11 |
| openstackgerrit | karim proposed openstack/nova master: Handle rebuild of instances with image traits https://review.openstack.org/569498 | 22:16 |
| *** figleaf is now known as edleafe | 22:20 | |
| fried_rice | cdent: Found out that gabbi query_parameters doesn't preserve param order. Is that a bug? | 22:25 |
| fried_rice | cdent: I assume the http spec supports semantic significance to the ordering of qparams | 22:26 |
| cdent | fried_rice: I'm not certain, let me do some looking | 22:26 |
| *** dabo has joined #openstack-placement | 22:26 | |
| *** edleafe has quit IRC | 22:28 | |
| *** dabo is now known as edleafe | 22:28 | |
| openstackgerrit | Eric Fried proposed openstack/nova master: Test for multiple limit/group_policy qparams https://review.openstack.org/568713 | 22:29 |
| fried_rice | cdent: FYI ^ | 22:30 |
| cdent | fried_rice: everything I'm reading is that there's no spec and that it is framework dependent and they don't offer guarantees | 22:35 |
| fried_rice | vay | 22:35 |
| cdent | since gabbi's rep of query_parameters is a dict it is not ordered | 22:35 |
| fried_rice | well, I figured that part was the case. | 22:35 |
| fried_rice | framework dependent, meaning the server? | 22:35 |
| cdent | yeah | 22:36 |
| fried_rice | cdent: Then I would say it is a bug, because it means gabbi isn't "supporting" servers where ordering is significant. | 22:36 |
| cdent | well except that presumably ordering only matters when there are duplicates and a dict doesn't support duplicates | 22:37 |
| cdent | so if you want duplicates you need to put them in the url, and there you can do whatever you want | 22:37 |
| fried_rice | right, I get that there's a workaround, and I used it ^ | 22:37 |
| fried_rice | I'm not a yaml expert; but the syntax for query_parameters can't be a yaml dict, because I'm allowed to duplicate keys in there. | 22:38 |
| cdent | also if a service is relying on the ordering of parameters provided by a framework, I'd argue that's a bad server and I wouldn't want gabbi to support that because it would be supporting framework dependence | 22:38 |
| fried_rice | So this could be "fixed" without requiring a change in the yaml syntax. | 22:38 |
| cdent | so I think the ordering tests you are making are probably a bad idea | 22:38 |
| cdent | and we should not be making and predictions based on such things in our code | 22:38 |
| cdent | s/and/any/ | 22:39 |
| fried_rice | Well, per the commit message, I'm doing it so we don't "accidentally" change the behavior that we've coded into place. | 22:39 |
| fried_rice | I agree the right thing would be to make the server behave better here - by disallowing multiples in this case. | 22:39 |
| cdent | that we've coded that into place was an accident and we shouldn't legislate it | 22:40 |
| cdent | and if the argument is that we have to now because of microversions I'm pretty meh to that | 22:40 |
| cdent | because "we" are still the main clients | 22:40 |
| fried_rice | Well, would you support any level of fixup to this without a microversion? | 22:40 |
| cdent | I would say that fixing it to error on duplicates without a microversions is okay | 22:41 |
| fried_rice | Like the original patch, which only breaks if a consumer happens to be sending in multiple (and *different*) values for these qparams? | 22:41 |
| cdent | but I tend to be in the minority on such things | 22:41 |
| fried_rice | or that, which will break if the consumer sends multiple, even if they're the same. | 22:41 |
| fried_rice | edleafe: thoughts on this, with your API hat? | 22:41 |
| fried_rice | (Guessing he doesn't give a "fig", since it's Friday and he's still edleafe :) | 22:42 |
| cdent | have you tried that and turned on verbose to make sure you are sending what you think you are sending? | 22:43 |
| cdent | yaml allows you to duplicate keys in a dict and then silently doesn't complain | 22:44 |
| fried_rice | no, I didn't try that. | 22:44 |
| fried_rice | but I'm convinced it's not worth "fixing" in gabbi | 22:44 |
| cdent | http://paste.openstack.org/show/722542/ | 22:45 |
| fried_rice | ack | 22:46 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add nova-manage placement heal_allocations CLI https://review.openstack.org/565886 | 22:47 |
| cdent | i'm in meetings at the mo' but will look at that review soon | 22:47 |
| fried_rice | cdent: Apparently according to The Spec, we're doing it all kinds of wrong: https://esdiscuss.org/topic/json-duplicate-keys | 22:48 |
| cdent | fried_rice: what does json have to do with this topic? | 22:49 |
| fried_rice | hm, maybe nothing, except that we're using json schema to validate the querystring?? | 22:49 |
| *** leakypipes has quit IRC | 22:50 | |
| cdent | yeah that's orthogonal | 22:50 |
| fried_rice | I guess the querystring isn't actually json?? | 22:50 |
| fried_rice | sigh | 22:50 |
| cdent | corect | 22:50 |
| cdent | that we use jsonschema is effectively a hack (which happens to work) | 22:50 |
| fried_rice | well, it's because I was trying to figure out how to update the schema to reject duplicates. | 22:50 |
| cdent | max<something> | 22:50 |
| cdent | =1 | 22:51 |
| fried_rice | doesn't look like it. | 22:51 |
| * cdent looks more closely | 22:52 | |
| fried_rice | it's not maxItems | 22:55 |
| cdent | based on some reading I just did it probably doesn't exist because it's not valid javascript to have duplicated keys. however since the query params become a list we can check that the value is a list of one. but only if the code in utils:validate_query_params is returning a dict of lists. | 22:58 |
| cdent | is reg.GET a dict of strings or lists? | 22:58 |
| cdent | fried_rice: looks like the behavior is a bit weird: https://docs.pylonsproject.org/projects/webob/en/stable/api/multidict.html | 23:00 |
| cdent | and will depend on how jsonscheme.validate is accessing values | 23:00 |
| fried_rice | I'm feeling it's not possible to validate this via schema. | 23:01 |
| fried_rice | if at all, would have to be in the handler. Which is not the end of the world. | 23:01 |
| cdent | this mixed() method (passed to validate) might do the right thing: https://docs.pylonsproject.org/projects/webob/en/stable/api/multidict.html#webob.multidict.MultiDict.mixed | 23:01 |
| cdent | then if something is not a string it ought to kaboom | 23:02 |
| fried_rice | that's not how I read it. | 23:02 |
| cdent | maybe? (sorry for not being able to do more than throw links at you) | 23:02 |
| cdent | "the values are either single values, or a list of values when a key/value appears more than once in this dictionary" | 23:02 |
| fried_rice | I read that as, "if they specified one, you'll get back just the value; if they specified multiple, you'll get back a list." | 23:02 |
| cdent | right which is what you want for the validate (checking for string) | 23:03 |
| cdent | if 'resources' is 'type': 'string' then a list should 400 | 23:03 |
| fried_rice | But that method is on the GET object. | 23:03 |
| fried_rice | Oh, but we're already past the schema validation by the time we would have the opportunity to call mixed(), nah? | 23:04 |
| fried_rice | oh, I see. | 23:04 |
| cdent | no I'm saying change the util method to pass req.GET.mixed() not req.GET | 23:04 |
| cdent | (into jsonschema.validate) | 23:04 |
| fried_rice | yeah, trying. | 23:05 |
| fried_rice | I suspect we're going to blow up in multiple places, though. | 23:05 |
| cdent | will be interesting to know | 23:05 |
| cdent | breaking stuff is good | 23:05 |
| cdent | because it leads to fixing stuff | 23:05 |
| cdent | I suspect that making this change will make the validation much more robust and is probably something we ought to do in nova's api too (if it isn't already there) | 23:06 |
| fried_rice | Okay, that actually does the trick. | 23:06 |
| fried_rice | I haven't gone all the way through, but I suspect our code to handle multi member_of will be broken if we do this. | 23:07 |
| fried_rice | Agree it will make things more robust. Can we do it without a microversion? | 23:07 |
| openstackgerrit | Chris Dent proposed openstack/nova master: Optional separate database for placement API https://review.openstack.org/362766 | 23:07 |
| fried_rice | Anyway, I have to run. | 23:08 |
| cdent | yeah the schema for member_of will need to have it's schema adjusted to accept list | 23:08 |
| fried_rice | list or string | 23:08 |
| openstackgerrit | Chris Dent proposed openstack/nova master: Isolate placement database config https://review.openstack.org/541435 | 23:08 |
| cdent | yeah, right | 23:08 |
| openstackgerrit | Chris Dent proposed openstack/nova master: Ensure that os-traits sync is attempted only at start of process https://review.openstack.org/553857 | 23:08 |
| openstackgerrit | Chris Dent proposed openstack/nova master: Add PLACEMENT_DB_ENABLED=True to the nova-next job https://review.openstack.org/564067 | 23:09 |
| openstackgerrit | Chris Dent proposed openstack/nova master: Use nova.db.api directly https://review.openstack.org/543262 | 23:10 |
| fried_rice | added note to that review. Now I'm really gone. | 23:12 |
| *** fried_rice is now known as efried | 23:12 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!