| *** edmondsw has joined #openstack-placement | 01:07 | |
| *** edmondsw has quit IRC | 01:11 | |
| *** edmondsw has joined #openstack-placement | 01:22 | |
| *** edmondsw has quit IRC | 01:34 | |
| *** edmondsw has joined #openstack-placement | 01:35 | |
| *** edmondsw has quit IRC | 01:37 | |
| *** edmondsw has joined #openstack-placement | 01:37 | |
| openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Return nested providers in get_by_request https://review.openstack.org/567113 | 01:38 |
|---|---|---|
| openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Add traits check in nested provider cadidates https://review.openstack.org/567150 | 01:38 |
| openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Support nested alloc cands with sharing providers https://review.openstack.org/567508 | 01:38 |
| *** edmondsw has quit IRC | 01:38 | |
| *** tetsuro has joined #openstack-placement | 02:30 | |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Remove mox in test_compute_api.py (3) https://review.openstack.org/568184 | 04:07 |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Remove mox in test_compute_api.py (3) https://review.openstack.org/568184 | 04:09 |
| openstackgerrit | caoyuan proposed openstack/nova master: Update auth_url in install docs https://review.openstack.org/568002 | 05:49 |
| *** belmoreira has joined #openstack-placement | 06:08 | |
| openstackgerrit | Shilpa Devharakar proposed openstack/nova master: Update docs, API samples for noVNC 1.0.0 https://review.openstack.org/550173 | 06:29 |
| *** tssurya has joined #openstack-placement | 06:54 | |
| *** rubasov has joined #openstack-placement | 07:03 | |
| openstackgerrit | Surya Seetharaman proposed openstack/nova master: Metadata-API fails to retrieve avz for instances created before Pike https://review.openstack.org/567878 | 07:14 |
| openstackgerrit | Radoslav Gerganov proposed openstack/nova stable/queens: Add random sleep between retry calls to placement https://review.openstack.org/568218 | 07:47 |
| openstackgerrit | Lenny Verkhovsky proposed openstack/nova master: Adding NVMEoF for libvirt driver https://review.openstack.org/482640 | 07:47 |
| *** belmoreira has quit IRC | 08:01 | |
| *** belmoreira has joined #openstack-placement | 08:04 | |
| *** gibi has quit IRC | 08:05 | |
| *** gibi has joined #openstack-placement | 08:07 | |
| *** gibi is now known as Guest44118 | 08:07 | |
| *** alex_xu has quit IRC | 08:32 | |
| *** alex_xu has joined #openstack-placement | 08:33 | |
| *** cdent has joined #openstack-placement | 08:37 | |
| *** Guest44118 is now known as gibi | 08:44 | |
| *** purplerbot has quit IRC | 09:02 | |
| *** purplerbot_ has joined #openstack-placement | 09:02 | |
| *** purplerbot_ is now known as purplerbot__ | 09:02 | |
| *** purplerbot__ is now known as purplerbot | 09:02 | |
| openstackgerrit | sahid proposed openstack/nova master: libvirt: place emulator threads on CONF.cpu_shared_set https://review.openstack.org/510897 | 09:24 |
| *** avolkov has joined #openstack-placement | 09:31 | |
| *** diga has joined #openstack-placement | 09:53 | |
| openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Return all resources in provider_summaries https://review.openstack.org/558045 | 10:04 |
| *** edmondsw has joined #openstack-placement | 11:03 | |
| openstackgerrit | Balazs Gibizer proposed openstack/nova master: Escalate UUID validation warning to error in test https://review.openstack.org/540386 | 11:18 |
| openstackgerrit | Balazs Gibizer proposed openstack/nova master: Suppress UUID warning in map_instance unit tests https://review.openstack.org/568263 | 11:18 |
| *** diga has quit IRC | 11:40 | |
| *** sean-k-mooney has joined #openstack-placement | 11:51 | |
| sean-k-mooney | o/ | 11:51 |
| sean-k-mooney | anyone familar with how ironic placement custom resouces classes work around? | 11:54 |
| openstackgerrit | Julia Kreger proposed openstack/nova master: ironic: add instance_uuid before any other spawn activity https://review.openstack.org/563722 | 11:58 |
| sean-k-mooney | dose placement have a way to assocate a description what a custom resouce class models with the resouce class. e.g. Custom_barmetal_small -> 8 pcpu + 16GB RAM + 2 disk(120GB ssd + 1TB hdd) | 11:59 |
| openstackgerrit | Surya Seetharaman proposed openstack/nova master: Make nova service-list use scatter-gather routine https://review.openstack.org/568271 | 12:02 |
| openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Return nested providers in get_by_request https://review.openstack.org/567113 | 12:10 |
| openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Add traits check in nested provider cadidates https://review.openstack.org/567150 | 12:10 |
| openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Support nested alloc cands with sharing providers https://review.openstack.org/567508 | 12:10 |
| *** cdent has quit IRC | 12:11 | |
| *** tetsuro has left #openstack-placement | 12:15 | |
| openstackgerrit | Naichuan Sun proposed openstack/nova master: XenAPI: define a new image handler to use vdi streaming https://review.openstack.org/486475 | 12:16 |
| openstackgerrit | Surya Seetharaman proposed openstack/nova master: Make nova service-list use scatter-gather routine https://review.openstack.org/568271 | 12:54 |
| efried | sean-k-mooney: Other than docs, nope. | 12:58 |
| sean-k-mooney | efried: that is what i was affraid of, i could be captured in flavors but i was hoping for a way to model it in placement nativly | 12:59 |
| efried | sean-k-mooney: I don't know of any "description" field in placement. The closest we come is the 'name' field of a resource provider afaik. | 13:00 |
| sean-k-mooney | efried: one tought i had was to model the resouces as inventories and set the allocation step_size equal to the total but that has other issues | 13:01 |
| efried | sean-k-mooney: Right, because you have no way to enforce that the flavor consumes all the resources. | 13:02 |
| sean-k-mooney | basically since its a baremetal node everything from the RP is allocated in one go. | 13:02 |
| sean-k-mooney | yep | 13:02 |
| efried | sean-k-mooney: Having said that, if your flavor allocates any one of the resources and the min/step_size are equal to the total, then you would take that node out of the running for other scheduling ops anyway. | 13:02 |
| efried | um, unless we're talking about allocating a network device or something for post-deploy attach. | 13:03 |
| sean-k-mooney | ya that is why i was thinking of doing it that way but i want to support mixed hypervisor envs e.g. barmetal and vm and i do not want vms to be able to land on the barmetal RP | 13:04 |
| efried | sean-k-mooney: Presumably you would control that with a trait | 13:04 |
| sean-k-mooney | efried: i could be we do not have reuired trates for a RP currently | 13:04 |
| sean-k-mooney | e.g. only allocate from this RP if trait X is in allocation request | 13:05 |
| efried | huh? You would put the required trait in the flavor. | 13:05 |
| sean-k-mooney | i dont think it would be that hard to add but its not there today. | 13:05 |
| efried | but the custom resource classes are | 13:05 |
| sean-k-mooney | you would have a new table in placement to map required traits to RP and then request the trait in the flavor | 13:06 |
| edleafe | sean-k-mooney: a baremetal resource provider has an inventory of 1 CUSTOM_* resource. VMs couldn't land on it. | 13:06 |
| efried | I mean, if you're even *talking* about changing to standard resource classes, you must be okay with changing flavors around. | 13:06 |
| efried | sean-k-mooney: That's already in place. | 13:06 |
| efried | edleafe: That's what we're talking about "fixing". | 13:06 |
| sean-k-mooney | i can set required traits on a placement RP | 13:06 |
| efried | sean-k-mooney: yes. | 13:06 |
| *** mriedem has joined #openstack-placement | 13:06 | |
| edleafe | efried: fixing? What's broken | 13:07 |
| edleafe | ? | 13:07 |
| sean-k-mooney | and if you dont request that trait you cant allocate form that RP? | 13:07 |
| efried | In the flavor it looks like: "trait:CUSTOM_FOO": "required" | 13:07 |
| sean-k-mooney | efried: that is not what i want | 13:07 |
| efried | sean-k-mooney: Ah, no. I see, that's what you meant by "required". | 13:07 |
| sean-k-mooney | ya i want to put the required traits on the RP not in the flavor so if you request no tratis in the flavour you cant be allocated from the RP | 13:08 |
| edleafe | that sounds like a filter | 13:08 |
| efried | edleafe: We started with sean-k-mooney asking if he could associate some kind of metadata/description with the baremetal custom resource class, so you would have a way to know e.g. how many CPUs, how much memory, etc. | 13:08 |
| edleafe | efried: yeah, we decided a long time ago not to support metadata in placement | 13:09 |
| sean-k-mooney | edleafe: basicaly i would like to be able to say this custom resouce class give you x ammount of these other resouce classes | 13:10 |
| efried | edleafe: From there we went to talking about representing the ironic node via its *actual* resources. But then we have the problem of a) "partially allocated" baremetal nodes, and all that entails; and b) VMs being able to land there. | 13:10 |
| edleafe | sean-k-mooney: that info belongs in the flavor | 13:10 |
| edleafe | sean-k-mooney: you can set the amounts there, and then override in extra_specs to 0 | 13:11 |
| sean-k-mooney | edleafe: it could be ther yes but that just felt messy | 13:11 |
| sean-k-mooney | edleafe: that was my fallback | 13:11 |
| edleafe | sean-k-mooney: that's the design we put in place specifically for this | 13:11 |
| sean-k-mooney | edleafe: yes but it only works for custom resouces class that model compute hosts | 13:12 |
| edleafe | I don't follow. Why? | 13:12 |
| sean-k-mooney | edleafe: it wont work for things like FPGAs ectra | 13:12 |
| edleafe | I still don't see why | 13:13 |
| sean-k-mooney | well i dont want to be forced to spcify them in the nova flavor for one thing in all cases. | 13:14 |
| openstackgerrit | Eric Fried proposed openstack/nova master: add lower-constraints job https://review.openstack.org/555961 | 13:14 |
| edleafe | Then where would they be specified? They have to be set *somewhere*, no? | 13:15 |
| sean-k-mooney | edleafe: i am currently working on intergrating RSD with openstack and trying to think about how to model composable hardware within placement | 13:15 |
| sean-k-mooney | edleafe: in a composable world acclerators and nic could be modeled as sharing resouce providers and then hotpluged into compute systems dynamically | 13:16 |
| edleafe | sean-k-mooney: that sounds more like Cyborg than Nova | 13:17 |
| *** cdent has joined #openstack-placement | 13:18 | |
| sean-k-mooney | well the point stands in either case that i have to decole the metadata about my custom resouce class form nova flavors. cyborg will not handel all cases so we will need cinder support for volumes, and or neutron support also. | 13:19 |
| efried | sean-k-mooney: yaml to the rescue? | 13:19 |
| edleafe | efried: it does solve all problems | 13:20 |
| sean-k-mooney | efried: i could encode this info in the resouce_class name or some other hack i was hoping to have a more generic solution long term. | 13:20 |
| sean-k-mooney | anyway i can live with doing this in the flavor short term but i think it would be nice to standarise it in placement at somepoint | 13:21 |
| efried | sean-k-mooney: For the sake of discussion, what would that look like? | 13:21 |
| sean-k-mooney | i was thinking either a) a nested resouce_class type structure or b) yaml using resouce_class->quantity mappings + traits | 13:23 |
| sean-k-mooney | e.g. CUSTOM_COMPUTE_BLADE->{ VCPUS:16, RAM:16GB, DISK[{LOCAL_GB:1024,traits=HDD},{LOCAL_GB:256,traits:ssd}]} | 13:27 |
| sean-k-mooney | efried: i had not fully taught about how to model it i was just wondering if it had been solve in the ironic case already | 13:28 |
| sean-k-mooney | efried: edleafe will ye be at the summit out of interest? | 13:29 |
| efried | sean-k-mooney: We can solve the "partial allocation" problem the same way we did for xen: once the node is deployed, the virt driver changes any unallocated inventories to make reserved=total. | 13:29 |
| efried | sean-k-mooney: Yes, we both will be. We're giving a presentation on placement, if you'd like to attend :) | 13:30 |
| edleafe | s/attend/heckle | 13:30 |
| sean-k-mooney | efried: assuming im not on booth duty then i proably will. | 13:30 |
| efried | sean-k-mooney: ---^ but that still doesn't help us with the issue of VMs accidentally getting deployed to the baremetal node. | 13:30 |
| sean-k-mooney | edleafe: no heckling is for the foum not the presentation track :P | 13:31 |
| edleafe | sean-k-mooney: then why am I always heckled in both? | 13:31 |
| sean-k-mooney | efried: ya i think require traits for resocue providers is a simpler solution for that | 13:31 |
| efried | sean-k-mooney: I.e. "here's a trait on a resource provider that the requestor *must* specify, otherwise we won't include that resource provider in the response" ? | 13:32 |
| efried | That's a weird one. | 13:32 |
| sean-k-mooney | efried: yes | 13:32 |
| *** superdan is now known as dansmith | 13:33 | |
| efried | jaypipes_: When you're around... | 13:33 |
| sean-k-mooney | it solve the proablem and is not intrusive to the api its additive | 13:33 |
| *** belmorei_ has joined #openstack-placement | 13:33 | |
| sean-k-mooney | it also wont suck to impement in the sql | 13:33 |
| efried | Yeah, it's simple enough to implement; I'm mainly just having trouble figuring out how to describe it. | 13:34 |
| edleafe | or where to store that tidbit of information | 13:34 |
| efried | For starters, we need a new word, because "required" is taken. | 13:34 |
| *** belmoreira has quit IRC | 13:35 | |
| efried | edleafe: Right, it would have to be a separate table, or column on the resource_provider_traits table. | 13:35 |
| sean-k-mooney | edleafe: as a new table in the placement db with a new enpoint to set/get them. | 13:35 |
| edleafe | efried: IOW, RP metadata | 13:35 |
| efried | :) | 13:35 |
| cdent | let's not have rp metadata, please | 13:36 |
| cdent | or at least, let's finish everything else first before we even talk about it | 13:36 |
| sean-k-mooney | kindof but it is tracking qualitive requirement to the schduler. i was wondering if it could intersect with dansmith's prefilter work | 13:36 |
| sean-k-mooney | cdent: i can implement what i need with a custom filter for now encode the requirement in nova flavors. | 13:39 |
| cdent | sean-k-mooney: glad to hear you won't be left out in the cold | 13:40 |
| cdent | I think once a todo list gets to a year or more long, it's time to pause | 13:40 |
| sean-k-mooney | cdent: infact we created a filter to solve this usecase 2 years ago. but i was tring to think is there any way to allow this to work withing the existing placemnet api constructs but it look like the answer is no | 13:40 |
| sean-k-mooney | cdent: ya i dont want to add yet another placement todo to that list | 13:41 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: DNM: test raw image live migration with vfat config drive https://review.openstack.org/567860 | 13:41 |
| jaypipes_ | efried: yes? | 13:51 |
| *** jaypipes_ is now known as jaypipes | 13:51 | |
| efried | jaypipes: We were talking with sean-k-mooney about a suggested feature for placement, wanted to get your reaction. | 13:51 |
| efried | It's a little tough to describe because the word "required" is already taken. | 13:51 |
| efried | But basically it's this: Put a trait on the resource provider that the requestor *must* specify in order for that RP to be returned in an allocation request. | 13:52 |
| edleafe | efried: since it's a reverse of the current required, call it 'deriuqer' | 13:52 |
| efried | The use case is so we can model baremetal nodes with their actual resources instead of a single custom resource class. | 13:52 |
| * edleafe shudders | 13:53 | |
| efried | But we don't want regular VMs to be able to land there. | 13:53 |
| efried | So we want ironic flavors to have to say CUSTOM_BAREMETAL or whatever, and if they don't, they can't land there. | 13:53 |
| cdent | efried: seriously, can we just pause this entire discussion please? | 13:53 |
| cdent | we have enough going on already that by the time this feature comes to pass we will have learned a lot | 13:53 |
| openstackgerrit | Bence Romsics proposed openstack/osc-placement master: Add nested resource providers (v1.14) https://review.openstack.org/546675 | 13:54 |
| sean-k-mooney | i am fine with pausing it for now | 13:54 |
| openstackgerrit | Bence Romsics proposed openstack/osc-placement master: Limit allocation candidates (v1.15, v1.16) https://review.openstack.org/548043 | 13:54 |
| openstackgerrit | Bence Romsics proposed openstack/osc-placement master: Allocation candidates parameter: required (v1.17) https://review.openstack.org/548326 | 13:54 |
| efried | I don't think we're suggesting throwing this into Rocky or anything, cdent | 13:54 |
| efried | just tossing around ideas. | 13:54 |
| cdent | I know, but _all_ I do now is context switch | 13:54 |
| cdent | and never actually make any progress | 13:54 |
| cdent | and it is getting so freaking noisy | 13:54 |
| sean-k-mooney | in my spcific case all the baremetal nodes would be own by the admin tenant so i can just use https://specs.openstack.org/openstack/nova-specs/specs/rocky/approved/placement-req-filter.html or https://specs.openstack.org/openstack/nova-specs/specs/rocky/approved/alloc-candidates-member-of.html to support that | 13:55 |
| efried | oh, that's true too. | 13:55 |
| efried | The former, not the latter, at least not yet, because the only way to specify aggs in flavors will be via the az filter that dansmith is doing. | 13:56 |
| efried | (I'm sure I just used both "az" and "filter" wrong. Please don't yell at me.) | 13:56 |
| * alex_xu will have a lot of chat log to read | 13:57 | |
| *** jgwentworth is now known as melwitt | 13:57 | |
| sean-k-mooney | well effectivly with either i can create a nova aggregate around my specal nodes and restic the placement queries to that subset. albeit via a slightly different mechaniums in both cases | 13:57 |
| efried | Which reminds me, dansmith & jaypipes, we should chat about that at some point. Specifically, see this comment: https://review.openstack.org/#/c/515811/16/nova/scheduler/utils.py@133 | 13:58 |
| efried | jaypipes: You running the sched mtg? | 14:00 |
| jaypipes | efried: ya | 14:01 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add contributor docs on deprecating and removing compute REST APIs https://review.openstack.org/567687 | 14:12 |
| *** cdent_ has joined #openstack-placement | 14:16 | |
| *** cdent has quit IRC | 14:17 | |
| *** cdent_ is now known as cdent | 14:17 | |
| openstackgerrit | Bob Ball proposed openstack/nova master: XenAPI: define a new image handler to use vdi streaming https://review.openstack.org/486475 | 14:20 |
| openstackgerrit | Balazs Gibizer proposed openstack/nova master: Transform missing delete notifications https://review.openstack.org/410297 | 14:22 |
| openstackgerrit | Shilpa Devharakar proposed openstack/nova master: Update docs, API samples for noVNC 1.0.0 https://review.openstack.org/550173 | 14:29 |
| efried | jaypipes, mriedem: Would you please have a look at gibi's comment here: https://review.openstack.org/#/c/515811/16/nova/scheduler/utils.py@167 | 14:32 |
| efried | Given that perspective, and the fact there's nothing else around here that raises exceptions, I'm inclined to agree with gibi and remove the enforcement. Thoughts? | 14:32 |
| efried | dansmith: Would you please have a look at https://review.openstack.org/#/c/515811/16/nova/scheduler/client/report.py@a368 and let me know whether you would like to do this cutover in a separate patch? | 14:33 |
| dansmith | efried: you're asking if we should support multiple member_of with 1.24, then make this granular change separately yeah? | 14:34 |
| dansmith | if so, I agree | 14:34 |
| dansmith | er, I think we should yes | 14:35 |
| efried | dansmith: Okay. Do you want to propose the interim change or should I? | 14:35 |
| dansmith | I don | 14:36 |
| dansmith | I don't _want_ to, but I probably should at least try | 14:36 |
| dansmith | gimme a few | 14:36 |
| efried | dansmith: Okay. Here's a hint: you're going to need to switch to the tuple (vs. dict) form of url encoding. | 14:36 |
| dansmith | for qs_params you mean? | 14:37 |
| efried | yes | 14:37 |
| dansmith | yup | 14:38 |
| mriedem | efried: re https://review.openstack.org/#/c/515811/16/nova/scheduler/utils.py@167 - if we don't check in nova, placement will bomb out with a 400 right? | 14:54 |
| mriedem | during GET /allocation_candidates | 14:54 |
| *** belmorei_ has quit IRC | 14:54 | |
| mriedem | efried: replied inline | 14:58 |
| openstackgerrit | Dan Smith proposed openstack/nova master: Make scheduler client allow multiple member_of query parameters https://review.openstack.org/568313 | 15:05 |
| dansmith | efried: ^ | 15:05 |
| dansmith | still running functional tests | 15:05 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: DNM: test raw image live migration with vfat config drive https://review.openstack.org/567860 | 15:33 |
| openstackgerrit | Bob Ball proposed openstack/nova master: XenAPI: define a new image handler to use vdi streaming https://review.openstack.org/486475 | 15:35 |
| openstackgerrit | Bob Ball proposed openstack/nova master: XenAPI: Pass expected return codes to resize2fs https://review.openstack.org/568318 | 15:35 |
| *** cdent has quit IRC | 15:43 | |
| efried | mriedem: Yes, ack, and thanks. | 15:54 |
| efried | dansmith: Were you going to post func tests (new or changed) or did you just mean you were running to make sure nothing broke? | 16:05 |
| dansmith | efried: I just meant I hadn't run all the existing functional tests before submitting | 16:06 |
| efried | rgr | 16:06 |
| dansmith | efried: we have coverage for the member_of thing already so the fact that it didn't break any is sufficient I think, | 16:06 |
| dansmith | and I'll have functional tests in my filter patch that uses this with multiple ones | 16:06 |
| efried | rgr | 16:06 |
| openstackgerrit | Eric Fried proposed openstack/nova master: Granular requests to get_allocation_candidates https://review.openstack.org/515811 | 16:14 |
| efried | gibi, mriedem (and jaypipes): Done ^ | 16:15 |
| efried | ...but still need to rebase on dansmith's | 16:15 |
| efried | dansmith: (I just checked the change set where you introduced the quasi-multiple-member_of hack, and there were no func tests there, so I think you're probably clean) | 16:16 |
| gibi | efried: looks good to me | 16:22 |
| efried | gibi: thanks | 16:22 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Implement granular policy rules for placement https://review.openstack.org/524425 | 16:27 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add granular policy rules for /resource_classes* https://review.openstack.org/565578 | 16:27 |
| mriedem | efried: don't you think we should probably have a functional test for https://review.openstack.org/#/c/515811/ ? | 16:28 |
| mriedem | or maybe unit is enough | 16:29 |
| efried | mriedem: We have functional tests (granular.yaml gabbit) for the placement side. The unit tests handle the parsing. You're asking for kind of and end-to-end func test where we go all the way from flavor through placement to scheduling? | 16:30 |
| mriedem | also, do we have any documentation for granular flavors besides https://review.openstack.org/#/c/515811/17/releasenotes/notes/granular-extra-specs-50b26b8f63717942.yaml and specs? | 16:30 |
| mriedem | efried: yes that's what i was thinking of | 16:30 |
| efried | mriedem: No, no docs other than the reno & flavors. I can work up a doc like we did for u-p-t (mostly capy/paste of the spec). | 16:30 |
| mriedem | reno & specs you mean? | 16:31 |
| efried | sorry, yes | 16:31 |
| mriedem | yeah i was thinking something should live in https://docs.openstack.org/nova/latest/user/flavors.html#extra-specs | 16:31 |
| mriedem | could be follow up changes of course | 16:31 |
| efried | mriedem: As for the func test, I thought that would be better waiting for nrp-in-alloc-cands to be done, because that's going to be where we get the most bang out of granular. | 16:31 |
| efried | since we still don't truly claim support for sharing providers. | 16:32 |
| efried | for that matter, perhaps the doc would be better off waiting for nrp-in-alloc-cands too. | 16:32 |
| jaypipes | efried: k. I'm pretty much done on reviewing tetsuro's member_of+alloc candidates series. needs to be rebased, though, according to gerrit | 16:32 |
| mriedem | we could test the negative scenario we already spent some time talking about, the novalidhost one because scope isn't specified | 16:34 |
| efried | jaypipes: The series starting here is ready, right? https://review.openstack.org/#/c/567113/ | 16:34 |
| jaypipes | efried: that's a different series, but yes, I think it is. | 16:35 |
| mriedem | anyway, functional e2e could be a follow up | 16:35 |
| jaypipes | efried: I'll review that one now. | 16:35 |
| efried | jaypipes: I think that was the "bottom" of the nrp-in-alloc-cands series, and he was going to rebase the others on top. | 16:35 |
| efried | mriedem: Roger that. | 16:36 |
| jaypipes | efried: k | 16:36 |
| mriedem | one thing i just thought of - how many e2e functional tests are going to no longer work once we don't have a running placement service in-tree... | 16:36 |
| mriedem | like, we have to use fixtures for cinder and neutron with stubs for the actual data interactions, we haven't had to do that for placement b/c we have the service running in our functional tests | 16:36 |
| efried | mriedem: I think cdent's plan for that includes... running the placement service in-tree. | 16:36 |
| efried | but I don't know for sure how that's going to work. | 16:37 |
| efried | mriedem: Have you seen his placecat thingy? | 16:37 |
| efried | https://github.com/cdent/placecat | 16:37 |
| mriedem | no | 16:37 |
| efried | Not sure if it's related to how he plans to "install" placement in tree for tests. | 16:38 |
| efried | this would be a good topic for his forum session. | 16:38 |
| efried | mriedem: ^ https://etherpad.openstack.org/p/YVR-placement-extraction ^ | 16:39 |
| efried | not sure if this concern is included in L42-3 or 45. | 16:39 |
| efried | added L46 | 16:40 |
| mriedem | virtualThumbsUp | 16:41 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: add lower-constraints job https://review.openstack.org/555961 | 16:47 |
| openstackgerrit | Sylvain Bauza proposed openstack/nova-specs master: Proposes Multiple GPU types https://review.openstack.org/557065 | 17:07 |
| *** efried has quit IRC | 17:19 | |
| *** efried has joined #openstack-placement | 17:19 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Wait for network-vif-plugged before starting live migration https://review.openstack.org/558001 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add check if neutron "binding-extended" extension is available https://review.openstack.org/523548 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add "bind_ports_to_host" neutron API method https://review.openstack.org/523604 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add "delete_port_binding" network API method https://review.openstack.org/552170 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add "activate_port_binding" neutron API method https://review.openstack.org/555947 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Delete port bindings in setup_networks_on_host if teardown=True https://review.openstack.org/556333 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Implement migrate_instance_start method for neutron https://review.openstack.org/556334 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add VIFMigrateData object for live migration https://review.openstack.org/515423 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Add VIFMigrateData.get_dest_vif https://review.openstack.org/566931 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: libvirt: factor out pre_live_migration plug_vifs call https://review.openstack.org/566932 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: libvirt: use dest host port bindings during pre_live_migration https://review.openstack.org/566933 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: libvirt: use dest host vif migrate details for live migration https://review.openstack.org/551370 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Annotate flows and handle PortBindingDeletionFailed in ComputeManager https://review.openstack.org/551371 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: Port binding based on events during live migration https://review.openstack.org/434870 | 17:25 |
| openstackgerrit | Matt Riedemann proposed openstack/nova master: conductor: use port binding extended API in during live migrate https://review.openstack.org/522537 | 17:25 |
| openstackgerrit | Dan Smith proposed openstack/nova master: Honor availability_zone hint via placement https://review.openstack.org/546282 | 17:39 |
| dansmith | efried: ^ that will make use of multiple member_of, and covers it in more tests | 17:40 |
| dansmith | jaypipes: ^ | 17:40 |
| efried | ack | 17:40 |
| *** cdent has joined #openstack-placement | 17:48 | |
| *** tssurya has quit IRC | 17:48 | |
| jaypipes | dansmith: rock on, thanks brother. | 17:52 |
| cdent | efried, mriedem : placecat has nothing to do with fixtures for nova | 17:56 |
| cdent | my thinking/expectation was that we produce a re-usable fixture in placement itself that nova could choose to import | 17:56 |
| cdent | as I understand it zuulv3 makes it pretty easy to use master of another project from one project, so we could effective continue using a fixture that provides the same functionality as the current one | 17:58 |
| cdent | the point of placecat is to play with placement however one might like. if for some reason we really did want a throwaway placement container to run in integration tests (which might use handy) then that would be https://github.com/cdent/placedock (not cat, cat uses dock) | 17:59 |
| cdent | Starting to use throwaway containers in integration tests strikes me a pretty useful thing but not really germane to mriedem's concern | 18:00 |
| cdent | for that I think we should consider doing whatever we can to make the existing tests continue to work | 18:00 |
| mriedem | i expect the easiest thing would just be nova has a test-requirement on placement | 18:03 |
| mriedem | the repo | 18:03 |
| cdent | that's pretty much what I mean | 18:03 |
| cdent | but I had heard that zuul has some new jiggery pokery for ways to make that special | 18:04 |
| cdent | but I guess those special bits are when there are non-test requirements | 18:04 |
| openstackgerrit | Merged openstack/nova master: Suppress UUID warning in map_instance unit tests https://review.openstack.org/568263 | 18:07 |
| openstackgerrit | Mohammed Naser proposed openstack/nova master: Added ability to configure default architecture for ImagePropertiesFilter https://review.openstack.org/566425 | 18:07 |
| openstackgerrit | Mohammed Naser proposed openstack/nova master: Added ability to configure default architecture for ImagePropertiesFilter https://review.openstack.org/566425 | 18:42 |
| openstackgerrit | Eric Fried proposed openstack/nova master: __str__ methods for RequestGroup, ResourceRequest https://review.openstack.org/568353 | 19:07 |
| efried | dansmith, jaypipes: As promised ^ | 19:07 |
| cdent | efried: I like, mostly, but think it has a fairly significant problem | 19:17 |
| efried | tell me | 19:17 |
| cdent | I commented on it, but \n\t in log messages -1 | 19:19 |
| cdent | or in stringifications in general | 19:19 |
| efried | cdent: I was afraid that would be a problem. I just really want the friggin thing to be readable. Kind of the whole point of this change. | 19:20 |
| cdent | yeah, I hear that, but it's problematic for processing | 19:21 |
| efried | cdent: processing of what? | 19:22 |
| * efried goes to read the comment... | 19:22 | |
| cdent | logs | 19:22 |
| cdent | I didn't say much more than what's here | 19:22 |
| efried | butbutbut | 19:22 |
| efried | exception traces have line breaks all over them. | 19:23 |
| efried | ...and that's the only thing that makes them friggin readable, btw. | 19:23 |
| cdent | ikr | 19:23 |
| cdent | I'm stating a personal preference here | 19:23 |
| cdent | sorry NOT stating | 19:23 |
| cdent | I think the difference here between a stacktrace and stringification is about what is consider the atom. And also people complain about stacktraces not being atoms all the time | 19:24 |
| cdent | dan, jay, etc may disagree | 19:24 |
| *** e0ne has joined #openstack-placement | 19:24 | |
| efried | cdent: So what do you suggest to make this readable? | 19:26 |
| cdent | efried: one possibility: do something a bit like what my very old Java recalls: RequestGroup<$WHAT_YOU_HAVE>, RequestGroup< .... | 19:27 |
| *** belmoreira has joined #openstack-placement | 19:38 | |
| openstackgerrit | Zack Cornelius proposed openstack/nova-specs master: Libvirt file backed memory https://review.openstack.org/563704 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: CLI for traits (v1.6) https://review.openstack.org/514643 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: Resource class set (v1.7) https://review.openstack.org/514644 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: Usages per project and user (v1.8, v1.9) https://review.openstack.org/514646 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: CLI allocation candidates (v1.10) https://review.openstack.org/514647 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: New dict format of allocations (v1.11, v1.12) https://review.openstack.org/542819 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: Transactionally update allocations (v1.13) https://review.openstack.org/546674 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: Add nested resource providers (v1.14) https://review.openstack.org/546675 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: Limit allocation candidates (v1.15, v1.16) https://review.openstack.org/548043 | 19:53 |
| openstackgerrit | Matt Riedemann proposed openstack/osc-placement master: Allocation candidates parameter: required (v1.17) https://review.openstack.org/548326 | 19:53 |
| mriedem | 1.6 at the bottom ^ should be happy to go | 19:54 |
| jaypipes | mriedem: ty sir. | 19:56 |
| openstackgerrit | Eric Fried proposed openstack/nova master: __str__ methods for RequestGroup, ResourceRequest https://review.openstack.org/568353 | 19:59 |
| openstackgerrit | Eric Fried proposed openstack/nova master: Granular requests to get_allocation_candidates https://review.openstack.org/515811 | 19:59 |
| efried | cdent: Did that ^ And dansmith jaypipes this is now rebased on dansmith's 1.24-usin patch (entirely supplants it, in fact). | 20:00 |
| openstackgerrit | Chris Dent proposed openstack/nova master: Extract part of PlacementFixture to placement https://review.openstack.org/568359 | 20:00 |
| openstackgerrit | Artom Lifshitz proposed openstack/nova master: Refactor _build_device_metadata https://review.openstack.org/533804 | 20:00 |
| efried | (But I understand why we had to do it that way, sigh.) | 20:00 |
| openstackgerrit | Artom Lifshitz proposed openstack/nova master: Consider hostdev devices when building metadata https://review.openstack.org/533805 | 20:00 |
| openstackgerrit | Zack Cornelius proposed openstack/nova-specs master: Libvirt file backed memory https://review.openstack.org/563704 | 20:00 |
| jaypipes | efried: been trying to get through tetsuro's patch series reviews. | 20:01 |
| cdent | efried: yeah that seems better and more proper (even though it is harder to read) | 20:01 |
| cdent | efried, mriedem : the "Extract part of PlacementFixture" above is related to the discussions earlier about fixtures between placement and nova | 20:02 |
| efried | jaypipes: Roger that, I need to do the same. | 20:03 |
| efried | ack | 20:03 |
| openstackgerrit | Matt Riedemann proposed openstack/nova stable/queens: libvirt: Report the virtual size of RAW disks https://review.openstack.org/568363 | 20:14 |
| *** efried has quit IRC | 20:17 | |
| openstackgerrit | Chris Dent proposed openstack/nova master: Extract part of PlacementFixture to placement https://review.openstack.org/568359 | 20:20 |
| *** avolkov has quit IRC | 20:22 | |
| *** cdent has quit IRC | 20:26 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova stable/pike: libvirt: Report the virtual size of RAW disks https://review.openstack.org/568369 | 20:30 |
| *** belmoreira has quit IRC | 20:33 | |
| *** belmoreira has joined #openstack-placement | 20:42 | |
| *** edmondsw has quit IRC | 20:47 | |
| *** edmondsw has joined #openstack-placement | 20:47 | |
| *** belmoreira has quit IRC | 20:47 | |
| *** edmondsw has quit IRC | 20:52 | |
| openstackgerrit | Mohammed Naser proposed openstack/nova master: Added ability to configure default architecture for ImagePropertiesFilter https://review.openstack.org/566425 | 20:56 |
| *** efried has joined #openstack-placement | 21:04 | |
| *** e0ne has quit IRC | 21:16 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova stable/ocata: libvirt: Report the virtual size of RAW disks https://review.openstack.org/568382 | 21:17 |
| *** edmondsw has joined #openstack-placement | 22:03 | |
| *** efried1 has joined #openstack-placement | 22:05 | |
| *** efried has quit IRC | 22:07 | |
| *** efried1 is now known as efried | 22:07 | |
| *** edmondsw has quit IRC | 22:08 | |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Remove mox in test_conductor.py (1) https://review.openstack.org/564658 | 22:20 |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Remove mox in test_conductor.py (2) https://review.openstack.org/564659 | 22:21 |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Remove mox in libvirt/test_driver.py (1) https://review.openstack.org/566555 | 22:21 |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: api-ref: Example verification for servers.inc https://review.openstack.org/529520 | 22:21 |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Remove mox in test_live_migrate.py https://review.openstack.org/564656 | 22:22 |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Remove mox in tests/unit/api/openstack/compute https://review.openstack.org/566528 | 22:22 |
| *** mriedem has quit IRC | 22:22 | |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Remove mox in tests/unit/api/*/test_volumes.py https://review.openstack.org/564655 | 22:22 |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Fix 500 error while passing 4-byte unicode data https://review.openstack.org/407514 | 22:33 |
| openstackgerrit | Takashi NATSUME proposed openstack/nova master: Adds view builders for keypairs controller https://review.openstack.org/347289 | 22:34 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!