| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-05-06 | |||
| 13:42:31 | efried | o/ | |
| 14:29:51 | mriedem | hi | |
| 14:59:00 | gibi | tssurya: ack | |
| 14:59:57 | gibi | tssurya: the size of that patch is a bit intimidating | |
| 15:10:08 | mriedem | the locked reason one? | |
| 15:10:14 | mriedem | anything involving API samples is going to be big | |
| 15:14:51 | gibi | mriedem: yes, that one. But I started reading it... | |
| 15:26:42 | openstackgerrit | Merged openstack/nova stable/queens: libvirt: Stop ignoring unknown libvirtError exceptions during volume attach https://review.opendev.org/657050 | |
| 15:32:42 | openstackgerrit | Merged openstack/nova stable/queens: libvirt: Always disconnect volumes after libvirtError exceptions https://review.opendev.org/657111 | |
| 16:06:36 | amansi26 | fungi:In order to remove this particular warning http://paste.openstack.org/show/750709/ . Do I need to add cafile option at both the groups (glance and ssl)? or how is it? | |
| 16:24:01 | mriedem | amansi26: it appears that way yes https://docs.openstack.org/nova/latest/configuration/config.html#glance.cafile | |
| 16:32:29 | mriedem | fyi i saw a grenade job fail because it took 202 seconds to spawn an instance http://logs.openstack.org/61/621061/5/check/grenade-py3/6aabc8e/logs/screen-n-cpu.txt.gz#_May_05_16_41_08_772233 and the tempest timeout is 196 seconds | |
| 16:32:32 | mriedem | maybe just a slow node | |
| 16:33:31 | cdent | oh noes, i can't find it | |
| 16:36:38 | efried | cfriesen: on TPM, did we decide to do implicit version? | |
| 16:40:48 | efried | so | |
| 16:40:48 | efried | => required=$version present, hw:sev_model=$model present => $model at $version | |
| 16:40:48 | efried | => required=$version present, hw:sev_model absent => TIM at $version | |
| 16:40:48 | efried | => required=$version absent, hw:sev_model=$model present => nova adds required=$version, where $version = 2_0 if $model is CRB else 1_2 | |
| 16:42:37 | efried | dah, tpm_model, not sev_model :) | |
| 16:46:06 | cdent | yeah, I'm going to need you to come in on saturday | |
| 16:46:22 | efried | that's tps_report | |
| 16:46:25 | cfriesen | efried: yeah. I started looking at the implementation yesterday and to avoid code duplication I'd have to move a bunch of code around into helper functions | |
| 16:46:42 | efried | really? | |
| 16:46:44 | mriedem | i know this is a bit fugly, but can we at least get another core on the (non-fugly and quite beautiful) recreate test for this stein regression? https://review.opendev.org/#/q/topic:bug/1825034+(status:open+OR+status:merged) | |
| 16:46:44 | efried | okay. | |
| 16:47:01 | efried | mriedem: looking. | |
| 16:48:03 | mriedem | thanks. we at least have 1-2 regression fixes merged back to stable/stein so it'd be good to do a release soonish | |
| 16:55:42 | cfriesen | efried: both the version and the model can be specified in both the image and the flavor, so I've got helper functions down in the libvirt driver to retrieve them. We'd need to move those up to where they're accessible from the sched request filter, which in turn means mucking around with test cases, etc. | |
| 16:56:43 | efried | cfriesen: Oh, okay, well, I would have likely -1'd anyway requesting that the validation be done in the request filter :) | |
| 16:58:10 | efried | No reason to wait until we get all the way to the compute to check for conflicting values, if we can fail at the scheduler | |
| 16:58:35 | cfriesen | efried: yesterday I was looking at whether we could do validation in _validate_flavor_image*. | |
| 16:59:01 | efried | yeah, could. But that doesn't mean you shouldn't *also* do it in the request filter. | |
| 16:59:31 | efried | oh, maybe it does | |
| 16:59:54 | cfriesen | validate all the things everywhere. :) | |
| 17:00:08 | cfriesen | might be overkill | |
| 17:00:44 | efried | yeah, that does validation, which is fine, but you would have to duplicate some of the logic to do the implicit required trait and/or model business. | |
| 17:00:59 | efried | probably not a huge amount of overlap, actually. | |
| 17:01:18 | efried | Anyway, the right place for all of that logic is *not* the virt driver. | |
| 17:02:33 | cfriesen | yeah, it's high-level enough that it would apply to any implementation in other virt drivers | |
| 17:02:34 | efried | um, the logic to do implicit model lives in the virt driver though. | |
| 17:03:51 | cfriesen | I'll rework the change. Hopefully soonish. | |
| 17:04:10 | efried | no hurry, I just wanted to record the design we decided on before I forgot it :) | |
| 17:23:53 | efried | mriedem: https://review.opendev.org/#/c/653098/ <== If you want to add a comment, like: | |
| 17:23:53 | efried | # The following (archive stuff) is used to prove that the migration created a "fake instance". It is not necessary to trigger the bug. | |
| 17:23:53 | efried | $archive stuff | |
| 17:23:53 | efried | # Since the above (archive stuff) removed the fake instance, do the migration again to recreate it so we can exercise the code path. | |
| 17:23:53 | efried | ...I'll fast approve | |
| 17:25:13 | openstackgerrit | Dan Smith proposed openstack/nova master: Add a workaround config toggle to refuse ceph image upload https://review.opendev.org/657078 | |
| 17:53:01 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add regression test for bug 1825034 https://review.opendev.org/653098 | |
| 17:53:02 | openstack | bug 1825034 in OpenStack Compute (nova) "listing deleted servers from the API fails after running fill_virtual_interface_list online data migration" [High,In progress] https://launchpad.net/bugs/1825034 - Assigned to Matt Riedemann (mriedem) | |
| 17:53:53 | mriedem | efried: done | |
| 17:53:55 | efried | mriedem: +A | |
| 17:53:58 | mriedem | thanks | |
| 17:54:03 | efried | mriedem: Not sure if the fix needs a rebase now? | |
| 17:54:13 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Exclude fake marker instance when listing servers https://review.opendev.org/653158 | |
| 17:54:33 | mriedem | ^ | |
| 17:54:47 | efried | neat | |
| 17:54:49 | efried | +A | |
| 18:01:36 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/stein: Add regression test for bug 1825034 https://review.opendev.org/657420 | |
| 18:01:38 | openstack | bug 1825034 in OpenStack Compute (nova) "listing deleted servers from the API fails after running fill_virtual_interface_list online data migration" [High,In progress] https://launchpad.net/bugs/1825034 - Assigned to Matt Riedemann (mriedem) | |
| 18:01:38 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/stein: Exclude fake marker instance when listing servers https://review.opendev.org/657421 | |
| 18:14:11 | rm_work | mriedem: correction -- apparently we use the log driver and ship notifications to splunk via logs, so no RMQ involved >_> | |
| 18:15:58 | mriedem | rm_work: ah ok, i don't hear of many people doing that | |
| 18:17:15 | rm_work | yeah, apparently some of our team didn't realize either... I had no idea what we were doing on the nova side of things at all, but finding out more about our deployments daily by asking around :D | |
| 18:45:07 | openstackgerrit | Merged openstack/nova master: Add regression test for bug 1825034 https://review.opendev.org/653098 | |
| 18:45:08 | openstack | bug 1825034 in OpenStack Compute (nova) stein "listing deleted servers from the API fails after running fill_virtual_interface_list online data migration" [High,In progress] https://launchpad.net/bugs/1825034 - Assigned to Matt Riedemann (mriedem) | |
| 18:59:58 | openstackgerrit | Jay Pipes proposed openstack/nova master: add InstanceList.get_all_uuids_by_hosts() method https://review.opendev.org/623557 | |
| 18:59:59 | openstackgerrit | Jay Pipes proposed openstack/nova master: single pass instance info fetch in host manager https://review.opendev.org/623558 | |
| 20:38:42 | redkrieg | does anyone know how I determine which image property the nova scheduler ImagePropertiesFilter is failing a compute host for? | |
| 20:47:35 | efried | redkrieg: Debug logging switched on for n-sch? | |
| 20:48:17 | efried | you'd be looking for one of these: | |
| 20:48:17 | efried | LOG.debug("Instance contains properties %(image_props)s, " | |
| 20:48:17 | efried | "but no corresponding supported_instances are " | |
| 20:48:17 | efried | "advertised by the compute node", | |
| 20:48:17 | efried | or | |
| 20:48:18 | efried | LOG.debug("Instance contains properties %(image_props)s " | |
| 20:48:18 | efried | "that are not provided by the compute node " | |
| 20:48:19 | efried | "supported_instances %(supp_instances)s or " | |
| 20:48:19 | efried | "hypervisor version %(hypervisor_version)s do not match", | |
| 20:52:53 | redkrieg | hmm, I don't see either of those, but I did get this: 2019-05-06 16:50:31.908 1736 DEBUG nova.scheduler.filters.image_props_filter [req-91826f4d-09e1-45ba-92ce-da96721c3edf c7ff06eff7ea4850aeae27eaa7e285af cd8cde005d0841e5958408ef22e8ac72 - default default] (la-skvm-1, la-skvm-1.example.com) ram: 384892MB disk: 7231488MB io_ops: 0 instances: 0 does not support requested instance_properties | |
| 20:52:59 | redkrieg | host_passes /openstack/venvs/nova-18.1.5/lib/python2.7/site-packages/nova/scheduler/filters/image_props_filter.py:117 | |
| 20:54:01 | redkrieg | nevermind, missed that earlier | |
| 20:56:35 | redkrieg | efried: I see in this line that img_hv_type=qemu but the hypervisor says [u'x86_64', u'kvm', u'hvm']. could that be the problem? http://paste.openstack.org/show/750737/ | |
| 20:59:16 | efried | redkrieg: We're kind of surpassing my knowledge here, but that seems like a reasonable guess. | |
| 21:01:26 | redkrieg | thanks for taking a look, changing that property to kvm worked! | |
| 21:04:52 | artom | redkrieg, yeah, I assume you have libvirt+kvm hypervisors | |
| 21:05:10 | artom | So having your virt type set to qemu (aka emulation) doesn't work | |
| 21:05:41 | redkrieg | indeed. I created the image with a new control panel I'm testing and it didn't offer kvm as an option, only qemu. | |
| 21:05:43 | artom | Oh wait, it's an image prop | |
| 21:05:47 | redkrieg | yeah | |
| 21:05:54 | artom | OK, so fuzzier on the details, but you have the right idea :) | |
| 22:32:25 | openstackgerrit | Eric Fried proposed openstack/os-resource-classes master: Propose standard ACCELERATOR_FPGA resource class https://review.opendev.org/657464 | |
| 22:52:55 | openstackgerrit | Merged openstack/nova master: Exclude fake marker instance when listing servers https://review.opendev.org/653158 | |
| #openstack-nova - 2019-05-07 | |||
| 03:57:45 | openstackgerrit | Chason Chan proposed openstack/python-novaclient master: Tiny fix of documentation https://review.opendev.org/657522 | |
| 04:48:21 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: Fix assert methods in unit tests https://review.opendev.org/657525 | |
| 09:07:24 | openstackgerrit | Merged openstack/nova master: Use migration_status during volume migrating and retyping https://review.opendev.org/637224 | |
| 11:27:43 | openstackgerrit | Merged openstack/nova master: Log when port resource is leaked during port delete https://review.opendev.org/657079 | |
| 11:27:58 | openstackgerrit | Merged openstack/nova master: Fix assert methods in unit tests https://review.opendev.org/657525 | |
| 11:30:10 | openstackgerrit | John Garbutt proposed openstack/nova-specs master: Policy Default Refresh spec https://review.opendev.org/547850 | |