| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-01-09 | |||
| 16:52:29 | mriedem | ok | |
| 16:52:52 | mriedem | typically applying patches to the CI results for code that's not yet merged is a bit a of a no-no | |
| 16:53:34 | openstackgerrit | Claudiu Belu proposed openstack/nova master: tests: fixes mock autospec usage https://review.openstack.org/447505 | |
| 16:54:19 | efried | mriedem We have a chicken/egg here | |
| 16:55:25 | mriedem | well, that egg wasn't necessary in pike was it? | |
| 16:55:34 | esberglu | mriedem: No we did this in pike as well | |
| 16:55:40 | mriedem | or is the chicken the os-vif stuff added with the ovs change, which the SEA change relies on? | |
| 16:55:43 | mriedem | oh fun | |
| 16:57:55 | efried | mriedem Yeah, the SEA change is on top of the OVS change. Technically we would probably be able to wait to switch on the patching-and-SEA-networking part of the CI until after the OVS patch merges, but for the sake of getting both patches in the review pipeline as quickly as possible... | |
| 17:03:22 | gibi | mriedem: have you planned to join to the notification subteam meeting? | |
| 17:04:17 | stephenfin | claudiub: Want to look at https://review.openstack.org/#/c/447505/ so I can sign off on it? | |
| 17:05:15 | openstackgerrit | Ildiko Vancsa proposed openstack/nova master: WIP: [api] Allow multi-attach in compute api https://review.openstack.org/271047 | |
| 17:07:02 | gibi | mriedem: as nobody joined, I've just closed the meeting, we can discuss thing separatly if neede | |
| 17:08:06 | mriedem | gibi: sorry missed your ping | |
| 17:08:36 | stephenfin | lyarwood: Looked, but my head is gone. I'll hit it tomorrow morning | |
| 17:08:52 | lyarwood | stephenfin: no issues | |
| 17:18:59 | claudiub | stephenfin: done. just fyi. the patch merged in oslotest, and it is included in oslotest 3.2.0. right now, I'm still waiting for the global-requirements bump for oslotest, and then for the OpenStack Bot requirements update to be proposed to nova and be merged, Then finally, that patch is mergeable. :) | |
| 17:19:45 | efried | bauzas You still working on series https://review.openstack.org/#/c/528832/ locally? I see "Done" comments from ~8h ago but no new patch set. | |
| 17:20:09 | openstackgerrit | Claudiu Belu proposed openstack/nova master: tests: fixes mock autospec usage https://review.openstack.org/447505 | |
| 17:27:36 | openstackgerrit | Matt Riedemann proposed openstack/nova master: libvirt: test to make sure volume_use_multipath is properly used https://review.openstack.org/532270 | |
| 17:27:53 | bauzas | efried: yup, I'm in there | |
| 17:28:15 | efried | k. | |
| 17:29:19 | bauzas | dansmith: around ? I have a thought | |
| 17:29:45 | bauzas | dansmith: I'm just building a dictionary of assigned mdevs but looking up the guests by init_host | |
| 17:30:05 | bauzas | dansmith: when I'm assigning a mdev to a guest, I'm also amending that dict | |
| 17:30:29 | efried | bauzas Re+W https://review.openstack.org/#/c/466051/ after rebase? | |
| 17:30:46 | bauzas | dansmith: but for example, say we migrate the instance (or just destroy it), I should call the method recreating the dict | |
| 17:31:22 | bauzas | dansmith: or we could have a mdev said to be assigned while the guest is destroyed | |
| 17:32:04 | bauzas | dansmith: do you think it would be better to call that method by each method (like migrate, rebuild and destroy) or just use a periodic method ? | |
| 17:32:26 | dansmith | bauzas: um, you're not persisting that dict are you? | |
| 17:32:33 | bauzas | if you don't understand my concern, no worries, I'll upload the series soon | |
| 17:33:02 | bauzas | dansmith: by persisting, you mean what ? using an attribute ? | |
| 17:33:19 | dansmith | I mean... persisting it.. I'm not sure what you'd need to update periodically, | |
| 17:33:31 | dansmith | but maybe just upload and we can talk about it concretely | |
| 17:33:34 | bauzas | nevermind, I'll upload | |
| 17:33:40 | bauzas | sure, just working on the tests now | |
| 17:34:41 | bauzas | efried: +Wd | |
| 17:34:45 | efried | thx | |
| 17:35:02 | openstackgerrit | Merged openstack/nova master: Add an additional description for 'token_ttl' https://review.openstack.org/532007 | |
| 17:35:30 | dansmith | mriedem: did you check that this is actually using multiple cells? https://review.openstack.org/#/c/517273/39/nova/tests/functional/compute/test_migration_list.py | |
| 17:38:48 | mriedem | dansmith: i think he copied that from your instance list tests, but it doesn't look like it explicitly creates multiple cells | |
| 17:39:22 | mriedem | https://github.com/openstack/nova/blob/master/nova/tests/functional/compute/test_instance_list.py#L25 | |
| 17:39:24 | mriedem | he needs that | |
| 17:39:27 | mriedem | good point | |
| 17:39:29 | dansmith | but I set | |
| 17:39:32 | dansmith | yeah :) | |
| 17:39:39 | mriedem | removed my +2 | |
| 17:39:54 | mriedem | lay down the -1 | |
| 17:40:05 | dansmith | I done did | |
| 17:44:09 | openstackgerrit | Matt Riedemann proposed openstack/nova master: libvirt: add tests to check multipath in iscsi/fc volume connectors https://review.openstack.org/532272 | |
| 17:47:23 | mriedem | dansmith: per https://review.openstack.org/#/c/517273/39/nova/db/sqlalchemy/api.py@4410 did you want to hold up https://review.openstack.org/#/c/514904/ ? | |
| 17:47:49 | dansmith | no I just wanted to make the comment to sound like I'm paying attention | |
| 17:47:54 | mriedem | heh, ok | |
| 19:15:20 | openstackgerrit | Mark Goddard proposed openstack/nova master: WIP: Send traits to ironic on server boot https://review.openstack.org/508116 | |
| 19:15:20 | openstackgerrit | Mark Goddard proposed openstack/nova master: Add get_traits() method to ComputeDriver https://review.openstack.org/532287 | |
| 19:15:21 | openstackgerrit | Mark Goddard proposed openstack/nova master: Implement get_traits() for the ironic virt driver https://review.openstack.org/532288 | |
| 19:15:21 | openstackgerrit | Mark Goddard proposed openstack/nova master: Add has_any_traits() to provider tree https://review.openstack.org/532289 | |
| 19:15:22 | openstackgerrit | Mark Goddard proposed openstack/nova master: Call get_traits() in the resource tracker https://review.openstack.org/532290 | |
| 19:31:44 | efried | mgoddard Let me know if you want to chat about ^ | |
| 20:13:14 | mriedem | wait wait wait, | |
| 20:13:18 | mriedem | johnthetubaguy has a minion? | |
| 20:15:23 | openstackgerrit | Merged openstack/nova master: libvirt: Provide VGPU inventory for a single GPU type https://review.openstack.org/514763 | |
| 20:18:37 | rybridges | Hey Guys! Have a really quick question. Related to this spec: https://specs.openstack.org/openstack/nova-specs/specs/kilo/implemented/quiesced-image-snapshots-with-qemu-guest-agent.html | |
| 20:19:17 | rybridges | Is that QEMU guest agent that is mentioned something that a user would install on their own instance? Or is that something that we (as operators) would install on the hypervisor itself? | |
| 20:20:29 | mriedem | rybridges: it's in the image | |
| 20:20:44 | mriedem | ‘hw_qemu_guest_agent=yes’ property and ‘hw_require_fsfreeze=yes’ property is set on the image metadata, and QEMU Guest Agent is installed and enabled in the instance | |
| 20:20:54 | mriedem | s/in the instance/in the image/ | |
| 20:21:12 | mriedem | the image metadata is just telling nova that the image has the guest agent and requires fsfreeze | |
| 20:21:14 | mriedem | prior to snapshot | |
| 20:24:09 | rybridges | ok so the part i am confused about is "QEMU Guest Agent is installed and enabled in the instance" | |
| 20:24:20 | rybridges | so this means we would bake the QEMU Guest Agent into the image itself | |
| 20:24:33 | rybridges | such that when we boot an instance, the QEMU Guest Agent will automatically come up and be enabled | |
| 20:25:33 | openstackgerrit | Ildiko Vancsa proposed openstack/nova master: WIP: [api] Allow multi-attach in compute api https://review.openstack.org/271047 | |
| 20:25:44 | ildikov | mriedem: ^^ | |
| 20:26:12 | mriedem | rybridges: yes | |
| 20:26:20 | mriedem | that's my understanding, nova doesn't inject something into the image | |
| 20:26:25 | openstackgerrit | Eric Berglund proposed openstack/nova master: WIP: PowerVM Driver: vSCSI https://review.openstack.org/526094 | |
| 20:33:55 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add cross cell sort support for get_migrations https://review.openstack.org/517273 | |
| 20:33:56 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add pagination and Changes-since filter support for os-migrations. https://review.openstack.org/330406 | |
| 20:33:56 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add index(updated_at) on migrations table. https://review.openstack.org/531132 | |
| 20:34:08 | mriedem | dansmith: i addressed the 2 addressable comments in the migration multi-cell lister patch ^ | |
| 20:38:09 | rybridges | ok thanks mriedem | |
| 20:38:34 | mriedem | rybridges: https://www.sebastien-han.fr/blog/2015/02/09/openstack-perform-consistent-snapshots-with-qemu-guest-agent/ ? | |
| 20:38:42 | mriedem | old blog but probably still good | |
| 20:41:15 | mriedem | stvnoyes: finally got that snapshot test passing http://logs.openstack.org/86/531386/7/check/tempest-full/eca51c5/job-output.txt.gz#_2018-01-09_19_34_20_731609 | |
| 20:41:17 | mriedem | but failed in teardown | |
| 20:41:38 | mriedem | http://logs.openstack.org/86/531386/7/check/tempest-full/eca51c5/job-output.txt.gz#_2018-01-09_20_01_02_052268 | |
| 20:41:41 | mriedem | hmm, failed to delete the volume, | |
| 20:42:20 | stvnoyes | i never like it when problems happen in teardown... | |
| 20:42:25 | mriedem | but the test is explicitly deleting the volume and cascading it's snapshot delete https://review.openstack.org/#/c/266605/24/tempest/api/compute/volumes/test_attach_volume.py@417 | |
| 20:42:36 | mriedem | smcginnis: does the volume cascading delete stuff work? | |
| 20:42:52 | smcginnis | mriedem: Why do I feel like this is a trick question? | |
| 20:43:27 | mriedem | maybe i don't know what's possible, but, | |
| 20:43:38 | mriedem | i create a volume-backed instance and create a snapshot of it, which also creates a snapshot of the volume, | |
| 20:43:57 | mriedem | at the end of the test, i do a delete volume with cascade=True so i don't have to explicitly delete the volume snapshot | |
| 20:44:05 | mriedem | assuming that's OK? | |
| 20:44:37 | smcginnis | mriedem: So from Cinders perspective that's just a volume with a snapshot. So yeah, that should work if you pass cascade=True and it should delete any child snapshots on volume deletion. | |
| 20:45:30 | mriedem | weird, don't even see the error in the c-api logs | |
| 20:45:37 | mriedem | oh here we go | |
| 20:46:45 | mriedem | damn c-api log doesn't really tell me anything | |