| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-10-16 | |||
| 07:22:39 | openstackgerrit | Brin Zhang proposed openstack/nova master: Follow-up: Add delete_on_termination to volume-attach API https://review.opendev.org/680655 | |
| 07:27:42 | gibi | good morning | |
| 07:47:11 | openstackgerrit | Brin Zhang proposed openstack/python-novaclient master: Add minor version [21] to the test_versions https://review.opendev.org/688599 | |
| 08:06:50 | LiangFang | hi, where can I find the plan of Shanghai PTG? Thanks | |
| 08:07:02 | LiangFang | e.g. the topics | |
| 08:07:18 | LiangFang | how to add topic please? | |
| 08:15:41 | melwitt | LiangFang: http://lists.openstack.org/pipermail/openstack-discuss/2019-October/010126.html | |
| 08:18:39 | kashyap | melwitt: Wauw, you're still awake? | |
| 08:20:08 | LiangFang | melwitt: thanks | |
| 08:23:37 | melwitt | kashyap: yeah :( going to bed soon | |
| 08:24:13 | melwitt | thanks o/ | |
| 08:43:58 | frickler | KeithMnemonic: I haven't tested that scenario yet, but we're running the same setup, so I'd be interested in hearing more about your findings. If you discover a reproducible issue, it would probably be best to create a bug report on lp | |
| 09:13:12 | openstackgerrit | sunjiyun-inspur proposed openstack/nova master: first modify https://review.opendev.org/688884 | |
| 09:31:30 | openstackgerrit | Huachang Wang proposed openstack/nova master: Set instance CPU policy to 'share' through image property https://review.opendev.org/688603 | |
| 11:07:22 | openstackgerrit | wangfaxin proposed openstack/nova master: Fix misspell word https://review.opendev.org/688901 | |
| 12:53:04 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Mask the token used to allow access to consoles https://review.opendev.org/220622 | |
| 13:29:06 | mriedem | bauzas: another scenario where we could leak allocations https://bugs.launchpad.net/nova/+bug/1848343 | |
| 13:29:07 | openstack | Launchpad bug 1848343 in OpenStack Compute (nova) "MigrationTask rollback can leak allocations for a deleted server" [Undecided,Triaged] | |
| 13:29:59 | mriedem | does your audit command detect the scenario that a resource provider has VCPU allocations but the instance consumer does not exist? | |
| 13:30:43 | dansmith | mriedem: guh, sorry about those last few issues, but thanks for fixing them | |
| 13:32:00 | mriedem | np | |
| 13:32:31 | mriedem | i reckon i'll spend some time today to write a recreate test for same-cell resize for that latent bug above | |
| 13:48:55 | openstackgerrit | Huachang Wang proposed openstack/nova master: [WIP] To create single NUMA node instance in function '_get_numa_topology_auto' https://review.opendev.org/688932 | |
| 13:48:55 | openstackgerrit | Huachang Wang proposed openstack/nova master: [WIP] Assign and track instance pinning cpu through 'cpu_pinning' field https://review.opendev.org/688933 | |
| 13:48:56 | openstackgerrit | Huachang Wang proposed openstack/nova master: [WIP] Add a new instance CPU allocation policy: mixed https://review.opendev.org/688934 | |
| 13:48:57 | openstackgerrit | Huachang Wang proposed openstack/nova master: [WIP] virt/libvirt: Get host pin cpuset according instance cpu_pinning https://review.opendev.org/688935 | |
| 13:48:57 | openstackgerrit | Huachang Wang proposed openstack/nova master: [WIP] metadata: export the vCPU IDs that are pinning on the host CPUs https://review.opendev.org/688936 | |
| 13:57:56 | KeithMnemonic | frickler i will let you know what i find out. so far it seems some sort of access issue with the backend rbd disk during an evacuation | |
| 14:06:30 | openstackgerrit | Dan Smith proposed openstack/python-novaclient master: Add aggregate-cache-images command and client routines https://review.opendev.org/687141 | |
| 14:07:55 | openstackgerrit | Matt Riedemann proposed openstack/python-novaclient master: Microversion 2.80: Add user_id/project_id to migration-list API https://review.opendev.org/675023 | |
| 14:09:30 | openstackgerrit | Matt Riedemann proposed openstack/python-novaclient master: Add aggregate-cache-images command and client routines https://review.opendev.org/687141 | |
| 14:15:03 | bauzas | mriedem: sorry, got some network issues last hour, just saw your ping | |
| 14:16:29 | bauzas | mriedem: FWIW, what I do with the audit command is that I look at all the RPs and then look at all the allocations | |
| 14:16:48 | bauzas | if some allocation is not either related to an instance or a migration, I tell it | |
| 14:20:08 | mriedem | with an option to delete the allocation? | |
| 14:21:22 | bauzas | yup | |
| 14:21:46 | bauzas | mriedem: I still have an issue with some placement call because I need to pass a microversion | |
| 14:21:57 | bauzas | but for the moment, I'm updating some unittests | |
| 14:22:32 | mriedem | do you filter the resource providers at all based on inventory or trait to know they are compute nodes, e.g. GET /resource_providers?resources=VCPU:1 | |
| 14:22:46 | mriedem | to like weed out things we won't care about | |
| 14:23:45 | mriedem | MEMORY_MB would probably be more appropriate given PCPU could complicate things now if a node never reports VCPU inventory | |
| 14:34:17 | bauzas | mriedem: we get all allocations but we skip those not related to nova RCs https://review.opendev.org/#/c/670112/7/nova/cmd/manage.py@2794 | |
| 14:40:44 | openstackgerrit | Dan Smith proposed openstack/nova master: Fix up some feedback on image precache support https://review.opendev.org/688172 | |
| 14:40:45 | openstackgerrit | Dan Smith proposed openstack/nova master: WIP: Add image precaching docs for aggregates https://review.opendev.org/687348 | |
| 14:40:45 | openstackgerrit | Dan Smith proposed openstack/nova master: WIP: Log some stats for image pre-cache https://review.opendev.org/688173 | |
| 14:43:01 | openstackgerrit | Merged openstack/nova-specs master: Add 'Feature Liaison' spec process https://review.opendev.org/685857 | |
| 14:46:59 | mriedem | bauzas: i'm saying we could optimize here https://review.opendev.org/#/c/670112/7/nova/cmd/manage.py@2921 by filtering out providers that aren't going to be compute nodes | |
| 14:47:08 | mriedem | i.e. providers that don't provide MEMORY_MB inventory | |
| 14:47:22 | mriedem | although, | |
| 14:47:31 | mriedem | i guess that would break ironic node providers... | |
| 14:47:32 | mriedem | so nevermind | |
| 14:48:12 | bauzas | mriedem: well, maybe we could use consumer types once it's done | |
| 14:48:36 | bauzas | ie. asking for RPs having some "Nova" consumer type | |
| 15:13:13 | mriedem | bauzas: not that i was actively reviewing it, but approving a policy change like https://review.opendev.org/#/c/685857/ without at least the majority of the core team acking it seems wrong | |
| 15:13:34 | mriedem | i guess my fault for not being more involved | |
| 15:13:46 | efried | https://www.zdnet.com/article/the-openstack-train-keeps-chugging-on/ Nova features are mentioned (albeit pretty inaccurately) | |
| 15:14:00 | mriedem | i heard there is a cyborg spec! | |
| 15:14:10 | mriedem | the marketing machine does it's job again | |
| 15:14:14 | mriedem | *its | |
| 15:14:16 | efried | ikr | |
| 15:15:36 | mriedem | i guess that was because of https://releases.openstack.org/train/highlights.html#cyborg-accelerator-life-cycle-management | |
| 15:22:44 | efried | I guess now that we've made such a big deal about SEV, AMD had better find someone to support it. I like how they plastered SUSE's departure all over the place without connecting the two at all. | |
| 15:34:26 | mriedem | "we" :) | |
| 15:36:21 | dansmith | ugh. | |
| 15:40:49 | dansmith | mriedem: a couple questions on your reviews from that client patch | |
| 15:40:51 | dansmith | if you please | |
| 15:48:22 | bauzas | mriedem: I feel bad then, my apologies | |
| 15:48:38 | mriedem | dansmith: replied | |
| 15:48:38 | bauzas | mriedem: we can honestly revert the change for asking at least for more | |
| 15:49:24 | bauzas | mriedem: do you want me to provide the revert change (and then the revert of revert) ? | |
| 15:49:27 | bauzas | efried: WFY ? | |
| 15:49:29 | mriedem | no | |
| 15:49:44 | mriedem | i'd say if someone on the core team has a stink about what was merged we can amend later | |
| 15:50:05 | bauzas | ok, I can write a ML email to mention it got merged then | |
| 15:50:20 | mriedem | sure | |
| 15:50:26 | bauzas | and ask for people to provide some change if they disagree with some stuff | |
| 15:50:28 | bauzas | ok, doing | |
| 15:50:33 | bauzas | mriedem: and again, apologies | |
| 15:51:07 | bauzas | FWIW, those were my notes for the +2 "OK, after commenting a lot about this spec process, I think I'm quite +2 with it, even if I could still have some concerns. Given it's important to have a consensus about this, I think it's more important to just accept it and then providing other follow-ups if there would be some points." | |
| 16:00:41 | efried | bauzas: I already wrote an email mentioning it's merged. | |
| 16:00:55 | bauzas | efried: ack | |
| 16:01:40 | efried | http://lists.openstack.org/pipermail/openstack-discuss/2019-October/010158.html | |
| 16:02:18 | efried | Let's be honest, if we waited for a majority of cores, we would be waiting forever. Not because they necessarily object, but because they don't care. | |
| 16:03:03 | mriedem | that's why i said it's fine to just amend if people care or take issue with it at this point | |
| 16:03:08 | efried | ++ | |
| 16:21:04 | mriedem | dansmith: replied on https://review.opendev.org/#/c/688832/ - that's just a wip right now that i pushed late last night, it's definitely not in final form, | |
| 16:21:12 | mriedem | i'm working on a recreate test for the latent same-cell resize bug | |
| 16:22:12 | openstackgerrit | Eric Fried proposed openstack/os-traits master: Add COMPUTE_NODE trait https://review.opendev.org/688969 | |
| 16:26:09 | melwitt | mriedem: question on this in case you didn't see https://review.opendev.org/687427 | |
| 16:28:19 | dansmith | mriedem: aight | |
| 16:30:12 | dansmith | mriedem: so on this object-or-id thing, | |
| 16:31:05 | dansmith | mriedem: if I take either are you expecting me to validate the uuid if it was given, or just pass that in the body? Like, leave the actually-check-with-glance part in the CLI shell, and just make the python API able to do either, but not validate? | |
| 16:37:20 | dansmith | mriedem: I guess based on the other examples, just include and not validate | |
| 16:37:48 | mriedem | the cli would get the image from glance based on name or id, | |
| 16:37:55 | mriedem | the python api binding method would just pass through | |
| 16:38:09 | dansmith | yeah | |
| 16:38:13 | mriedem | that's how other api methods like create, rebuild, rescue work | |
| 16:38:25 | dansmith | I'm not a fan, but consistency is more important for usre | |
| 16:39:48 | mriedem | if going directly to the api we'll barf and return a 400 if the image doesn't exist | |
| 16:39:49 | mriedem | so i think we're ok | |