| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-01-31 | |||
| 14:11:12 | lyarwood | yup, looks valid, can you create a bug? | |
| 14:11:12 | jroll | regardless, it's a unicode handling issue in the client, not a library issue | |
| 14:11:25 | bauzas | zigo: strange thing | |
| 14:11:56 | zigo | same stuff with openstackclient: http://paste.openstack.org/show/658154/ | |
| 14:12:01 | zigo | Will do. | |
| 14:12:37 | jaypipes | efried: hey, sorry, was afk yesterday evening... you were asking me something about PUT vs DELETE? | |
| 14:13:14 | bauzas | zigo: any trace we could look at ? | |
| 14:13:21 | bauzas | looks a novaclient issue to me | |
| 14:13:25 | zigo | bauzas: Yeah, just pasted 2 of them. | |
| 14:13:47 | bauzas | since openstackclient uses novaclient, I guess it's from here | |
| 14:13:51 | efried | jaypipes: Yeah, basically is there any reason to prefer a {DELETE <URI> => 204 no content} API over its corresponding {PUT <URI> => 200 <payload>} | |
| 14:13:53 | zigo | jaypipes: Good hey mate ! | |
| 14:15:10 | efried | jaypipes: Specifically for RP inventory, traits, etc. | |
| 14:15:40 | bauzas | zigo: python 2 or 3 ? | |
| 14:15:42 | openstackgerrit | Merged openstack/nova master: Add get_traits() method to ComputeDriver https://review.openstack.org/532290 | |
| 14:15:57 | openstackgerrit | Merged openstack/nova master: libvirt: test to make sure volume_use_multipath is properly used https://review.openstack.org/532270 | |
| 14:16:08 | openstackgerrit | Merged openstack/nova master: libvirt: add tests to check multipath in iscsi/fc volume connectors https://review.openstack.org/532272 | |
| 14:18:34 | zigo | bauzas: py2 | |
| 14:20:50 | ameeda | hello, I am try to find new bugs to resolve, I found this bug review https://review.openstack.org/#/c/212908/ . I think if I create new table with unique constraints then copy the data from old table to new one , so we clear the dirty data. | |
| 14:23:46 | openstackgerrit | Marcin Juszkiewicz proposed openstack/nova master: [dnm] Make sure that we have USB Host controller for tablet https://review.openstack.org/538003 | |
| 14:24:00 | hrw | morning | |
| 14:25:07 | mriedem | ameeda: nova hasn't used quota usages since pike | |
| 14:25:35 | mriedem | ameeda: https://specs.openstack.org/openstack/nova-specs/specs/pike/implemented/cells-count-resources-to-check-quota-in-api.html | |
| 14:25:53 | mriedem | i marked the bug as won't fix | |
| 14:25:53 | hrw | mriedem: can you take a look at https://review.openstack.org/#/c/538003/ patch and tell is it acceptable? | |
| 14:26:10 | ameeda | mriedem: Thank you! | |
| 14:26:26 | mriedem | hrw: should probably have sahid look at that, he worked on the pointer model stuff in the libvirt driver | |
| 14:26:30 | ameeda | mriedem: can you please recheck this https://review.openstack.org/#/c/528385/ | |
| 14:26:55 | hrw | mriedem: will add him, thanks | |
| 14:27:57 | mriedem | ameeda: can you please decouple that patch from the one below it in the series? | |
| 14:28:24 | mriedem | because melwitt's patch https://review.openstack.org/#/c/340614/ is already dealing with the api change (the first patch in your series) | |
| 14:29:00 | ameeda | how can I do that ? | |
| 14:29:45 | mriedem | put them in separate branches | |
| 14:29:45 | ccamacho | hey guys! quick question, Im working with the tripleo upgrades squad and Im trying to analyze the Nova database status prior an upgrade, mostly to run cleanup operation before running an upgrade in nova services. | |
| 14:29:56 | ccamacho | do you know briefly where can I find information about it? | |
| 14:30:16 | openstackgerrit | Matthew Edmonds proposed openstack/nova master: remove unnecessary conf imports https://review.openstack.org/539314 | |
| 14:30:24 | mriedem | ccamacho: find information about what specifically? | |
| 14:30:38 | mriedem | ccamacho: there are a few nova dbs | |
| 14:30:48 | mriedem | nova_api, nova_cell0, nova_cell1 (or just "nova") | |
| 14:31:16 | ameeda | mriedem: I should remove this from my branch https://review.openstack.org/#/c/528069 ? or create new branch for this https://review.openstack.org/#/c/528385/ and resubmit the review with new change id ? | |
| 14:31:17 | ccamacho | mriedan hi, basically about what to check to clean the databases | |
| 14:31:41 | mriedem | ccamacho: like how to archive anything that's already in it but deleted? | |
| 14:31:48 | mriedem | nova-manage db archive_deleted_rows | |
| 14:31:55 | mriedem | will move soft deleted records into the shadow tables | |
| 14:32:09 | mriedem | soft deleted in this case means records where the deleted column != 0 | |
| 14:32:33 | mriedem | we do'nt have a straight up db purge command | |
| 14:32:42 | openstackgerrit | Merged openstack/nova master: trivial: use cn instead of rp https://review.openstack.org/529786 | |
| 14:32:53 | openstackgerrit | Merged openstack/nova master: Remove the inherits parameter for the Resource object https://review.openstack.org/529218 | |
| 14:32:58 | mriedem | ccamacho: https://docs.openstack.org/nova/latest/cli/nova-manage.html#nova-database | |
| 14:32:59 | ccamacho | mrieden oki but that is the only cmd available to do it in a more or less automated way | |
| 14:33:07 | ccamacho | mrieden thanks for the link | |
| 14:33:37 | ccamacho | mrieden are the shadow tables there for ever? | |
| 14:33:39 | mriedem | there has been a spec and code up before for a command to do similar to archive_deleted_rows but actually hard delete the records rather than move them | |
| 14:33:45 | mriedem | ccamacho: yeah | |
| 14:33:52 | mriedem | so there is still bloat in the db for deleted stuff | |
| 14:34:49 | mriedem | https://blueprints.launchpad.net/nova/+spec/purge-deleted-instances-cmd | |
| 14:35:36 | mriedem | https://blueprints.launchpad.net/nova/+spec/purge-deleted-instances-cmd | |
| 14:35:38 | mriedem | oops | |
| 14:35:42 | mriedem | this is the spec https://specs.openstack.org/openstack/nova-specs/specs/mitaka/approved/purge-deleted-instances-cmd.html | |
| 14:35:57 | mriedem | and a poc from someone at red hat https://review.openstack.org/#/c/330791/ | |
| 14:36:15 | ccamacho | mrieden thanks, the idea is to write +- a procedure to verify and clean the DBs before the upgrade and be sure they are cleaned.. hopefully we can try to implement it after | |
| 14:36:18 | ccamacho | thanks man | |
| 14:36:42 | ameeda | mriedem: I should remove this from my branch https://review.openstack.org/#/c/528069 ? or create new branch for this https://review.openstack.org/#/c/528385/ and resubmit the review with new change id ? | |
| 14:36:56 | mriedem | ameeda: i think you should abandon https://review.openstack.org/#/c/528069 | |
| 14:37:14 | mriedem | ameeda: to do that locally, rebase your series and drop that change from the local branch, | |
| 14:37:18 | mriedem | git rebase -i master | |
| 14:37:23 | mriedem | <remove that commit> | |
| 14:37:30 | mriedem | git review | |
| 14:37:39 | ameeda | Ok let me check | |
| 14:37:40 | ameeda | Thanks | |
| 14:38:28 | mriedem | ccamacho: the man page doesn't list the option, but nova-manage db archive_deleted_rows [--max_rows <number>] [--verbose] has an option for running until complete | |
| 14:38:47 | mriedem | https://github.com/openstack/nova/blob/master/nova/cmd/manage.py#L494 | |
| 14:39:00 | ccamacho | mrieden ack thanks! | |
| 14:39:19 | efried | jaypipes: I opened some bugs on the subject. https://bugs.launchpad.net/nova/+bug/1746075 https://bugs.launchpad.net/nova/+bug/1746373 https://bugs.launchpad.net/nova/+bug/1746374 | |
| 14:39:21 | openstack | Launchpad bug 1746075 in OpenStack Compute (nova) "Report client placement cache consistency is broken" [Undecided,New] | |
| 14:39:22 | openstack | Launchpad bug 1746373 in OpenStack Compute (nova) "Placement APIs with missing conflict detection" [Undecided,New] | |
| 14:39:23 | openstack | Launchpad bug 1746374 in OpenStack Compute (nova) "Report client _delete_inventory violates generation consistency" [Undecided,New] | |
| 14:39:34 | efried | At a minimum, that last one needs to be fixed. | |
| 14:40:01 | efried | ...whenever we get to a point where there's actually a possibility of concurrency. | |
| 14:40:13 | mriedem | ameeda: it's all yours https://bugs.launchpad.net/nova/+bug/1746531 | |
| 14:40:14 | openstack | Launchpad bug 1746531 in OpenStack Compute (nova) "nova-manage db archive_deleted_rows man page does not mention --until-complete option" [Medium,Confirmed] - Assigned to Ameed Ashour (ameeda) | |
| 14:40:16 | jaypipes | efried: well, DELETE <uri> and PUT <uri> have different semantics, no? | |
| 14:40:38 | jaypipes | efried: I mean, PUT <uri> can do more than DELETE some object in a collection... | |
| 14:41:13 | jaypipes | efried: you are getting gibi-like bug-happy. | |
| 14:41:45 | efried | jaypipes: Correct. I'm saying PUT encompasses what DELETE does. But in these particular cases, DELETE doesn't actually provide us with enough functionality to allow us to ensure consistency. | |
| 14:41:54 | jaypipes | efried: how impactful are those bugs? do they need to be addressed for queens, in your opinion? | |
| 14:42:19 | efried | jaypipes: Probably not, since we have that big ol' lock preventing concurrency in the first place. | |
| 14:42:39 | jaypipes | efried: ack. can you add this to the list of discussion topics for Dublin? | |
| 14:44:10 | ameeda | mriedem: Thank you very much !! | |
| 14:45:43 | openstackgerrit | Ameed Ashour proposed openstack/nova master: detach instance volumes when VM creation fails https://review.openstack.org/528385 | |
| 14:46:25 | ameeda | mriedem: patch uploaded https://review.openstack.org/#/c/528385/ | |
| 14:46:50 | efried | jaypipes: https://etherpad.openstack.org/p/nova-ptg-rocky L50-59 | |
| 14:47:13 | jaypipes | efried: danke | |
| 14:47:28 | efried | jaypipes: To put possibilities in your head, we could have a DELETE that accepts a payload and returns a payload. | |
| 14:47:55 | efried | or just use PUT :) | |
| 14:48:26 | jaypipes | cdent might shit a brick, though. | |
| 14:48:57 | cdent | i've laid so many bricks in the past few days I'm not sure I have any left | |
| 14:48:59 | mriedem | if another core can look at this, i think we should get it in before rc1 https://review.openstack.org/#/c/522112/ beause backports with new config options are generally gross | |
| 14:49:09 | jaypipes | cdent: poor guy. still under the weather? | |
| 14:49:17 | jaypipes | mriedem: on it. | |