| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-10-17 | |||
| 17:42:31 | mriedem | that's one of them | |
| 17:42:37 | mriedem | and not called out specifically | |
| 17:42:44 | mriedem | for --all-tenants, | |
| 17:42:51 | mriedem | i'd think we could just list all instance mappings in down cells, | |
| 17:42:54 | mriedem | not filtered on project_id | |
| 17:43:03 | mriedem | up to whatever limit | |
| 17:49:31 | mriedem | so apparently filtering on project_id only works if you're also using all_tenants | |
| 17:49:37 | mriedem | otherwise we just filter on the context.project_id | |
| 17:50:42 | dansmith | um, what? | |
| 17:51:15 | dansmith | meaning --all_tenants is required in order to say --but-only-this-one ? | |
| 17:51:26 | mriedem | https://github.com/openstack/nova/blob/9c5d4eb2008df02837985779d87fedb17b4f70bf/nova/api/openstack/compute/servers.py#L206L248 | |
| 17:51:29 | mriedem | yes... | |
| 17:52:32 | dansmith | hmm | |
| 18:24:58 | openstackgerrit | Jack Ding proposed openstack/nova master: Add I/O Semaphore to limit concurrent disk ops https://review.openstack.org/609180 | |
| 18:32:31 | awaugama | mriedem: when you have a chance, it looks like the value is set on the compute node. after a reboot I saw this in the nova-compute log | |
| 18:32:34 | awaugama | 2018-10-17 17:44:00.277 1 DEBUG oslo_service.service [req-c9c4f04c-cfc6-4fe3-868d-206f9329419d - - - - -] cpu_allocation_ratio = 1.0 log_opt_values /usr/lib/python2.7/site-packages/oslo_config/cfg.py:3023 | |
| 18:33:15 | mriedem | jroll: where in the ironic API reference would one find anything about conductor_groups? https://developer.openstack.org/api-ref/baremetal/ | |
| 18:33:33 | mriedem | i mean i see https://docs.openstack.org/ironic/latest/contributor/webapi-version-history.html#rocky-11-1-0 | |
| 18:33:37 | mriedem | but nothing in the API reference | |
| 18:34:21 | jroll | mriedem: apparently I forgot to update that when I added it :( | |
| 18:34:26 | mriedem | awaugama: ok so it's 1.0 in config, it's 1.0 in the compute_nodes.cpu_allocation_ratio column in the db, but it's 16.0 in the associated resource provider VCPU inventory in placement | |
| 18:34:36 | awaugama | yeah | |
| 18:34:46 | mriedem | well wtf | |
| 18:34:49 | jroll | mriedem: it would be in node CRUD as its own field, I'll get that done real quick | |
| 18:34:56 | mriedem | jroll: a uuid or what? | |
| 18:35:33 | mriedem | jroll: and you can update a node's conductor_group? | |
| 18:35:34 | jroll | mriedem: a string, up to 255 characters IIRC: https://github.com/openstack/ironic/blob/b8ffcc0f0298fca5b4b36ad016e2c3b2f0e81710/ironic/common/utils.py#L530 | |
| 18:35:45 | mriedem | ok so it's just some tag | |
| 18:35:50 | awaugama | I'll sit down with sylvain tomorrow and we can do some debugging, will let you know if we find anything | |
| 18:35:51 | mriedem | special tag | |
| 18:35:52 | jroll | yeah | |
| 18:36:10 | jroll | alphanumeric, plus - _ . | |
| 18:42:25 | artom | - _ . is what I look like after a few drinks | |
| 18:42:52 | jroll | hah | |
| 18:53:13 | jroll | mriedem: api-ref for you https://review.openstack.org/611415 | |
| 18:53:49 | openstackgerrit | Artom Lifshitz proposed openstack/nova-specs master: Re-propose numa-aware-live-migration spec https://review.openstack.org/599587 | |
| 18:58:05 | mriedem | jroll: i've brought the wrath | |
| 18:58:47 | dansmith | hrm, pretty sure this functional timeout on the down cell series is a real deadlock on our cell cache | |
| 18:59:11 | jroll | mriedem: thanks, valid points | |
| 18:59:46 | jroll | for context I haven't touched our API ref in a long time :P | |
| 19:00:26 | mriedem | ugh | |
| 19:00:39 | mriedem | so GET /v1/nodes/detail is deprecated for GET /v1/nodes?detail=True, | |
| 19:01:00 | mriedem | but the request filter and response parameters for the latter don't mention anything possible in the former | |
| 19:08:37 | jroll | I'm not sure it's even properly deprecated | |
| 19:08:58 | jroll | added a note to the detail=True parameter | |
| 19:29:28 | mriedem | so we're deprecating the force flag from the evacuate and live migration apis, | |
| 19:29:40 | mriedem | wouldn't it behoove us to deprecate that as an option from nova commands as well? | |
| 19:29:49 | mriedem | or at least doc it up real good that you shouldn't use it? | |
| 19:32:50 | artom | mriedem, wait, deprecate or remove? | |
| 19:33:01 | melwitt | dansmith: would appreciate your review on mah backport https://review.openstack.org/610673 | |
| 19:33:05 | artom | Because for removal the usual novaclient microversion stuff applies, no? | |
| 19:34:08 | mriedem | artom: if we don't want people using the force flag to live migrate or evacuate an instance, | |
| 19:34:17 | mriedem | so much so that we're deprecating the api parameter, | |
| 19:34:26 | mriedem | you could argue that we should not have it in the CLI either | |
| 19:34:31 | mriedem | like, at all | |
| 19:34:34 | mriedem | even for older microversions | |
| 19:34:43 | artom | mriedem, but it still exists for old microversions | |
| 19:34:45 | artom | In the API | |
| 19:34:47 | mriedem | yes i know | |
| 19:34:52 | artom | So, we have to keep client support | |
| 19:35:02 | mriedem | not really | |
| 19:35:03 | artom | So if they specifically --os-compute-version <old>, they have it | |
| 19:35:06 | artom | Otherwise, it's gone | |
| 19:35:36 | mriedem | once all allocations are nested, you won't be able to force at all | |
| 19:35:43 | mriedem | regardless of microversion | |
| 19:35:52 | mriedem | anyway, it was just a thought | |
| 19:36:05 | mriedem | should probably start by putting the big fat warnings in the API reference into the CLI option descriptions | |
| 19:36:52 | artom | mriedem, ah I see. Well we still have to keep the old API intact, no? Just now we'll return a 400 or something. | |
| 19:37:04 | artom | If they send a force flag | |
| 19:37:59 | mriedem | it'll be some kind of error | |
| 19:38:04 | mriedem | don't know if it's a 400 or 409 | |
| 19:38:07 | mriedem | it's in gibi's spec | |
| 19:40:11 | mriedem | aspiers: just a few hundred comments in your spec https://review.openstack.org/#/c/609779/ | |
| 19:43:30 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Fix typo in libvirt.hw_machine_type help https://review.openstack.org/611422 | |
| 19:48:58 | openstackgerrit | Jack Ding proposed openstack/nova-specs master: High Precision Event Timer (HPET) on x86 guests https://review.openstack.org/607989 | |
| 19:51:43 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Document each libvirt.sysinfo_serial choice https://review.openstack.org/611426 | |
| 19:55:36 | artom | mriedem, you led me astray, I demand a refund: https://review.openstack.org/#/c/587439/4/nova/tests/unit/compute/test_compute_mgr.py@7439 | |
| 19:59:55 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: Ensure attachment cleanup on failure in driver.pre_live_migration https://review.openstack.org/587439 | |
| 20:00:22 | mriedem | umm | |
| 20:00:27 | mriedem | what's the point of even asserting those then | |
| 20:01:08 | artom | I guess Matt was trying to make sure we call all of the things? But... in a for loop, because he didn't feel like writing out each individual method name? | |
| 20:01:49 | mriedem | ^O^ | |
| 20:01:56 | mriedem | that's me shrugging, not a bat | |
| 20:02:24 | artom | Or a yelling Asian person? | |
| 20:02:30 | artom | (Can I say that? Is that racist?) | |
| 20:02:54 | mriedem | it's very racist | |
| 20:03:16 | artom | Dammit. Hilter 2.0 right here, friends. | |
| 20:05:47 | openstackgerrit | Merged openstack/nova master: Merge used_limits extension response into limit view builder https://review.openstack.org/606031 | |
| 20:15:31 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Fix block_device_mapping_v2 mention in server create API reference https://review.openstack.org/611433 | |
| 20:34:34 | openstackgerrit | Matt Riedemann proposed openstack/python-novaclient master: Recommend against using --force for evacuate/live migration https://review.openstack.org/611436 | |
| 21:15:09 | openstackgerrit | melanie witt proposed openstack/nova master: Add rolling upgrade info to enable_consoleauth workaround reno https://review.openstack.org/611218 | |
| 21:15:09 | openstackgerrit | melanie witt proposed openstack/nova master: Add nova-status upgrade check for consoles https://review.openstack.org/611214 | |
| 21:48:07 | openstackgerrit | Matt Riedemann proposed openstack/nova-specs master: Support initial allocation ratios https://review.openstack.org/552105 | |
| 21:51:20 | mriedem | holy fart i'm +2 on the initial allocation ratios spec https://review.openstack.org/#/c/552105/ | |
| 21:52:38 | melwitt | and the peasants rejoiced! | |
| 22:37:51 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: Add API ref guideline for examples https://review.openstack.org/604060 | |
| 23:19:44 | openstackgerrit | Merged openstack/nova stable/rocky: Handle volume API failure in _post_live_migration https://review.openstack.org/611083 | |
| 23:19:50 | openstackgerrit | Merged openstack/nova master: Ignore uuid if already set in ComputeNode.update_from_virt_driver https://review.openstack.org/611162 | |
| #openstack-nova - 2018-10-18 | |||
| 00:42:37 | openstack | bug 1798163 in OpenStack Compute (nova) rocky "Placement incomplete consumers online migration fails" [Critical,In progress] https://launchpad.net/bugs/1798163 - Assigned to Matt Riedemann (mriedem) | |
| 00:42:37 | openstackgerrit | Merged openstack/nova stable/rocky: Add recreate test for bug 1798163 https://review.openstack.org/611314 | |