| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-31 | |||
| 23:30:33 | openstackgerrit | Eric Fried proposed openstack/nova: PowerVM Driver: console https://review.openstack.org/409402 | |
| 23:35:14 | jaypipes | #success \0/ Finally got a good functional test created that stresses the Ironic and Nova integration and migration from Newton to Ocata. | |
| 23:35:16 | openstackstatus | jaypipes: Added success to Success page | |
| 23:41:35 | openstackgerrit | Eric Fried proposed openstack/nova: PowerVM Driver: config drive https://review.openstack.org/409404 | |
| 23:45:29 | andreykurilin | bnemec: hm | |
| 23:46:17 | andreykurilin | bnemec: is it difficult to add that? | |
| 23:47:11 | bnemec | andreykurilin: I'm not sure how we would. We'd need to look up the id in nova, which we can't do until we can create a client. | |
| 23:48:27 | bnemec | andreykurilin: Can we just switch to using a project name or something? | |
| 23:49:18 | andreykurilin | bnemec: yes, but it will not be compatible with min version in g-r. We can use simple check of novaclient version to support all versions. Are you ok about that solution? | |
| 23:50:32 | openstackgerrit | Scott DAngelo proposed openstack/nova: WIP Support cinder v3 API https://review.openstack.org/420201 | |
| 23:50:35 | bnemec | andreykurilin: I'm not sure we're as concerned about the minimum. Our repos are generated in such a way that I think if we find any way that works with the new novaclient we can start using that. | |
| 23:50:52 | andreykurilin | oh | |
| 23:50:54 | andreykurilin | cool | |
| 23:51:12 | andreykurilin | so it is just +2;-2 change (with tests) | |
| 23:51:12 | bnemec | Anyone on the old novaclient would be using the old instack-undercloud, and when they update to the new novaclient they should also get the new instack-undercloud. | |
| 23:55:11 | andreykurilin | bnemec panda: https://review.openstack.org/427515 | |
| 23:56:03 | andreykurilin | it should fix an issue | |
| 23:56:55 | bnemec | andreykurilin: Cool, thanks. | |
| 23:57:04 | panda | not until we put OS_TENANT_ID in stackrc ? | |
| 23:57:55 | andreykurilin | panda: that solution should work with OS_TENANT_NAME and with latest novaclient without changing stackrc | |
| 23:59:11 | panda | andreykurilin: nice, thanks | |
| #openstack-nova - 2017-02-01 | |||
| 00:07:41 | openstackgerrit | Jay Pipes proposed openstack/nova: placement: RT now adds proper Ironic inventory https://review.openstack.org/404472 | |
| 00:18:29 | openstackgerrit | David Medberry proposed openstack/nova: Explain how allow_resize_to_same_host is useful https://review.openstack.org/426531 | |
| 00:19:41 | openstackgerrit | David Medberry proposed openstack/nova: Explain how allow_resize_to_same_host is useful https://review.openstack.org/426531 | |
| 00:43:26 | andreykurilin | bnemec: for some reasons CI failed. | |
| 00:43:46 | andreykurilin | any other places of novaclient initialization? | |
| 00:46:51 | bnemec | andreykurilin: Crud, maybe I was wrong about not having to worry about older novaclients. | |
| 00:47:22 | bnemec | It's working on our ocata jobs, but they're using the latest novaclient too. | |
| 00:47:59 | andreykurilin | bnemec: u-c is used in all gates, even for those projects who ignores g-r :) so latest novaclient should be there | |
| 00:48:23 | andreykurilin | oh...there is 6.0 | |
| 00:48:29 | andreykurilin | http://logs.openstack.org/15/427515/2/check/gate-tripleo-ci-centos-7-undercloud/962e5b9/logs/pip2-freeze.txt.gz | |
| 00:48:39 | bnemec | andreykurilin: We don't install from source though. We use package repos, and non-tripleo projects are pinned to whatever version last passed our promotion jobs. | |
| 00:48:46 | bnemec | And this bug is blocking the promotion jobs. :-) | |
| 00:49:25 | bnemec | andreykurilin: I have to get going. I'll pick this back up tomorrow though. | |
| 00:52:11 | andreykurilin | bnemec: give me a sec, I'll resubmit a patch | |
| 00:53:12 | andreykurilin | bnemec: I submitted a new patch-set with hack for old release | |
| 00:54:02 | openstackgerrit | Ildiko Vancsa proposed openstack/nova: Implement new attach/detach Cinder flow https://review.openstack.org/330285 | |
| 00:54:06 | openstackgerrit | Ildiko Vancsa proposed openstack/nova: Remove check_attach https://review.openstack.org/335358 | |
| 01:38:51 | inc0 | hello guys, we (kolla) started to experience issues with nova - https://bugs.launchpad.net/kolla-ansible/+bug/1660761 | |
| 01:38:51 | openstack | Launchpad bug 1660761 in kolla-ansible "VM stuck in scheduling" [Critical,Confirmed] - Assigned to Michał Jastrzębski (inc007) | |
| 01:39:15 | inc0 | cells v2 landed? I assume there is new step durng deployment process right? | |
| 01:39:47 | inc0 | and we *have to* configure cells now? | |
| 01:41:43 | diana_clarke | inc0: yes, cells v2 setup is required as of ocata | |
| 01:42:31 | inc0 | http://docs.openstack.org/developer/nova/cells.html#cells-v2 this is correct doc right? | |
| 01:42:57 | diana_clarke | inc0: I'm just pulling up some links for you (sorry if that sounded short/terse a moment ago). | |
| 01:43:21 | diana_clarke | inc0: yup, that a good one to read | |
| 01:43:24 | diana_clarke | inc0: https://review.openstack.org/#/c/426864/5/releasenotes/notes/ocata_prelude-cfa8793d07f963e7.yaml | |
| 01:43:26 | inc0 | diana_clarke, it did not:) it sounded nothing but helpful:) | |
| 01:43:53 | inc0 | thank you, I'll get to writing deployment code for that | |
| 01:45:00 | diana_clarke | inc0: http://docs.openstack.org/developer/nova/man/nova-manage.html#nova-cells-v2 | |
| 01:45:57 | inc0 | thanks, I'll check if other deployment tools got that in their deploy as well | |
| 01:46:13 | inc0 | this will have to be part of upgrade task too I assume? | |
| 01:46:43 | diana_clarke | inc0: Finally, I find the codesearch tool for "cells_v2" helpful to see how others have done it: http://codesearch.openstack.org/?q=cell_v2&i=nope&files=&repos= | |
| 01:47:31 | diana_clarke | inc0: yup, upgrade too | |
| 01:48:02 | inc0 | once I'll write code for it, I'll leave links here too so you could double check | |
| 01:48:10 | inc0 | thank you very much | |
| 01:49:24 | diana_clarke | inc0: Good luck! I'm still learning the ins and outs of it, but I can try to answer any questions you have. | |
| 03:07:10 | mriedem | yay new gate bugs http://logs.openstack.org/91/426991/1/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/f218227/logs/screen-n-api.txt.gz#_2017-02-01_01_25_02_367 | |
| 03:15:49 | mriedem | melwitt: dansmith: for debugging pleasure tomorrow, looks like we have a race after moving instance creation to conductor causing this failure | |
| 03:15:51 | mriedem | https://bugs.launchpad.net/nova/+bug/1660878 | |
| 03:15:51 | openstack | Launchpad bug 1660878 in OpenStack Compute (nova) "test_reboot_deleted_server fails with 409 "Cannot 'reboot' instance while it is in vm_state building"" [High,Confirmed] | |
| 03:16:18 | mriedem | the test creates a server, then deletes it and waits for the 404, then tries to reboot it and expects a 404 but gets a 409 because it's now building | |
| 03:16:37 | mriedem | so my guess is we get a 404 when the build request is gone but before the instance is created, something like that | |
| 03:18:07 | openstackgerrit | Oliver Walsh proposed openstack/nova: DNM - TripleO CI test https://review.openstack.org/427536 | |
| 03:35:07 | dansmith | mriedem: that's not supposed to happen because of the way we do the create of the instance and build request | |
| 03:35:19 | dansmith | but okay, let's look tomorrow | |
| 03:35:33 | mriedem | i'm going to put up a change with a bunch of debug logging | |
| 03:35:47 | mriedem | since the delete and get paths in the api are pretty hairy | |
| 03:58:48 | smcginnis | Since this afternoon I'm getting stacking failures with "No host found for cellv1" errors. | |
| 03:59:00 | smcginnis | Is there something I need to do to get this to work right? | |
| 04:00:21 | mriedem | smcginnis: latest devstack? | |
| 04:00:26 | mriedem | any special localrc? | |
| 04:00:34 | smcginnis | mriedem: Freshly cloned. | |
| 04:00:42 | smcginnis | mriedem: Nothing special in local.conf. | |
| 04:00:53 | mriedem | got a paste of the rough failure log? | |
| 04:01:52 | mriedem | are you sure that's the same error message? because i'm not finding that | |
| 04:02:16 | smcginnis | mriedem: local.conf file: http://paste.openstack.org/show/597124/ | |
| 04:02:53 | smcginnis | mriedem: Shoot, looks like I'll have to try to stack again to get the exact text. | |
| 04:02:55 | openstackgerrit | Matt Riedemann proposed openstack/nova: WIP: Add debug logging for delete/query race bug https://review.openstack.org/427544 | |
| 04:03:26 | mriedem | smcginnis: http://lists.openstack.org/pipermail/openstack-dev/2017-January/111295.html ? | |
| 04:03:32 | mriedem | you don't have placement-api in ENABLED_SERVICES | |
| 04:03:58 | smcginnis | mriedem: Mmmm, nope, I don't. | |
| 04:04:17 | smcginnis | mriedem: Guess I should have read that one. | |
| 04:04:28 | mriedem | dansmith: melwitt: just fyi for the morning https://review.openstack.org/#/c/427544/ | |
| 04:04:39 | mriedem | this isn't a super high failure rate so i might have some rechecking to do | |
| 04:04:43 | mriedem | but it's high enough | |
| 04:04:50 | mriedem | 164 hits since 1/27 | |
| 04:07:00 | melwitt | mriedem: thanks. I'm looking at it now. nothing jumping out at me yet | |
| 04:07:31 | mriedem | the mess of conditionals all over create/delete/get in api and conductor makes me think something can slip in here | |
| 04:07:41 | mriedem | and we're probably doing a pass or some other assumption in one of them | |
| 04:07:54 | mriedem | like, | |
| 04:07:58 | mriedem | well, | |
| 04:08:06 | mriedem | nvm | |
| 04:08:52 | mriedem | i was thinking, before the api casts to conductor to create the instance in the cell, the build request could be deleted, and then we create the instance from stale build request info, but then in conductor when we go to delete the build request, it should fail with a 404 and then we should cleanup the instance too | |
| 04:09:24 | mriedem | however, i wonder if the GET request is hitting right in between the time that the instance is created and the time that we try to delete the build request and find out it's already deleted, then cleanup | |
| 04:09:32 | melwitt | hmm | |
| 04:09:38 | mriedem | because there is no mutex there at all from what i can tell, | |
| 04:09:55 | mriedem | and after we cast to conductor, we never refresh the build request from the db to get it's current state before we actually create the instance in the cell, from what i can tell anyway | |
| 04:12:56 | melwitt | yeah, that sounds reasonable. I was thinking along those lines of somehow the instance getting resurrected | |
| 04:17:02 | openstackgerrit | Matt Riedemann proposed openstack/nova: WIP: Add debug logging for delete/query race bug https://review.openstack.org/427544 | |