| *** tetsuro has joined #openstack-placement | 00:43 | |
| *** takashin has joined #openstack-placement | 01:17 | |
| *** tetsuro has quit IRC | 05:04 | |
| *** e0ne has joined #openstack-placement | 06:48 | |
| *** helenafm has joined #openstack-placement | 07:18 | |
| *** ttsiouts has joined #openstack-placement | 07:30 | |
| *** tetsuro has joined #openstack-placement | 08:21 | |
| *** yikun has quit IRC | 08:22 | |
| *** e0ne has quit IRC | 08:37 | |
| *** takashin has left #openstack-placement | 08:42 | |
| *** e0ne has joined #openstack-placement | 08:44 | |
| *** cdent has joined #openstack-placement | 08:59 | |
| *** rubasov has quit IRC | 09:10 | |
| *** rubasov has joined #openstack-placement | 09:15 | |
| *** helenafm has quit IRC | 09:25 | |
| *** tetsuro has quit IRC | 10:03 | |
| *** ttsiouts has quit IRC | 10:08 | |
| *** tetsuro has joined #openstack-placement | 10:14 | |
| *** tetsuro has quit IRC | 10:16 | |
| *** ttsiouts has joined #openstack-placement | 10:29 | |
| *** helenafm has joined #openstack-placement | 10:39 | |
| *** ttsiouts has quit IRC | 11:02 | |
| *** ttsiouts has joined #openstack-placement | 11:03 | |
| *** ttsiouts has quit IRC | 11:07 | |
| *** ttsiouts has joined #openstack-placement | 11:17 | |
| *** e0ne has quit IRC | 11:18 | |
| *** ttsiouts has quit IRC | 11:31 | |
| *** ttsiouts has joined #openstack-placement | 11:32 | |
| *** e0ne has joined #openstack-placement | 11:33 | |
| *** ttsiouts has quit IRC | 11:34 | |
| *** ttsiouts has joined #openstack-placement | 11:35 | |
| *** ttsiouts has quit IRC | 11:37 | |
| *** ttsiouts has joined #openstack-placement | 11:38 | |
| *** ttsiouts has quit IRC | 11:46 | |
| *** ttsiouts has joined #openstack-placement | 11:47 | |
| *** ttsiouts has quit IRC | 11:47 | |
| *** ttsiouts has joined #openstack-placement | 11:47 | |
| *** ttsiouts has quit IRC | 11:50 | |
| *** ttsiouts has joined #openstack-placement | 11:51 | |
| *** ttsiouts has quit IRC | 11:55 | |
| *** ttsiouts has joined #openstack-placement | 12:00 | |
| *** e0ne has quit IRC | 12:24 | |
| *** e0ne has joined #openstack-placement | 12:27 | |
| *** ttsiouts has quit IRC | 12:32 | |
| *** efried has joined #openstack-placement | 13:10 | |
| *** e0ne has quit IRC | 13:12 | |
| *** e0ne has joined #openstack-placement | 13:16 | |
| *** openstackgerrit has joined #openstack-placement | 13:22 | |
| openstackgerrit | Adam Spiers proposed openstack/nova-specs master: Add spec for libvirt driver launching AMD SEV-encrypted instances https://review.openstack.org/609779 | 13:22 |
|---|---|---|
| *** mriedem has joined #openstack-placement | 13:23 | |
| *** e0ne has quit IRC | 13:53 | |
| efried | bauzas: Is this http://lists.openstack.org/pipermail/openstack-operators/2018-October/016013.html related to what you were seeing the other day? | 13:59 |
| bauzas | efried: yup | 13:59 |
| bauzas | it was for an upgrade | 13:59 |
| bauzas | efried: https://bugzilla.redhat.com/show_bug.cgi?id=1636463 | 14:00 |
| openstack | bugzilla.redhat.com bug 1636463 in openstack-nova "nova-compute's UUID doesn't correlate with nova_api's resource_providers table contents" [High,Closed: notabug] - Assigned to nova-maint | 14:00 |
| bauzas | I closed it as NOTABUG since I wasn't able to reproduce it but I also fixed the problem by asking our customer to just update the DB directly | 14:00 |
| * bauzas tries to not be schziphrenic when it comes to discuss about internal bugs :) | 14:01 | |
| *** e0ne has joined #openstack-placement | 14:01 | |
| bauzas | schizophrenic even :) | 14:01 |
| cdent | I seem to recall an upstream bug that was similar but I'm struggling to find it. The basic idea was that you need to clean up after yourself if you destroy a compute node | 14:02 |
| efried | I don't really understand the issue; will one of you respond to that thread? | 14:02 |
| cdent | I was hoping saying the above would help find the bug and that once found we could then (and only then) respond on the thread | 14:03 |
| bauzas | efried: my assumption is that somehow the compute node name changed | 14:03 |
| bauzas | so it created a new entry | 14:03 |
| bauzas | and then a new UUID | 14:04 |
| bauzas | I asked the reporter to give me the compute_nodes records but he didn't so I still struggle by guessing the problem | 14:04 |
| cdent | yeah, what bauzas is describing sounds what I'm remembering as well | 14:05 |
| * cdent still looking | 14:05 | |
| bauzas | cdent: efried: replying to the thread | 14:06 |
| cdent | thanks | 14:07 |
| bauzas | replied | 14:12 |
| *** munimeha1 has joined #openstack-placement | 14:20 | |
| mriedem | gibi: replies in the grenade placement extraction change https://review.openstack.org/#/c/604454/ - if you're ok with those i'll make the updates | 14:41 |
| edleafe | So I'm working on setting up alembic for the placement database, and was wondering what the consensus is (if any) on moving from the nova-api db: are we copying the entire thing and then removing the tables we don't need, or are we only copying the placement tables? | 15:00 |
| cdent | edleafe: the model in the grenade change is to only copy the relevant tables (with dan's migration script): https://review.openstack.org/#/c/603234/8/tools/mysql-migrate-db.sh | 15:14 |
| cdent | however, if people do choose to copy their entire nova-api, or have been using the optional placement db, the extra tables shouldn't matter. they should be ignored, no? | 15:15 |
| edleafe | cdent: in practice, yes, but alembic will see them in those where they exist and not see them in the rest. Seems like having an inconsistent starting point will make the process more complicated | 15:17 |
| cdent | I guess I'm not understanding | 15:19 |
| edleafe | heh, like I'm any different :) | 15:20 |
| cdent | My understanding/hope is that alembic simply won't care about tables we don't tell it to care about? | 15:21 |
| cdent | also, somewhat related, dansmith and mriedem we're pretty keen about making sure that we document that people do a migration (using the aforementioned script) so that there's a clean break | 15:22 |
| cdent | doing otherwise is an option, of course | 15:22 |
| mriedem | iow don't point placement at the nova_api db | 15:23 |
| mriedem | or nova.conf | 15:23 |
| edleafe | mriedem: no, I'm not thinking about doing that | 15:23 |
| edleafe | just hoping that having two different potential starting points for the alembic migrations doesn't trip us up | 15:24 |
| mriedem | what are the two different starting points? | 15:25 |
| edleafe | mriedem: a DB with all the current nova-api tables, and a DB with just the placement-related tables | 15:25 |
| mriedem | a placement database with no nova_api tables vs a nova_api db with nova tables? | 15:25 |
| edleafe | jinx | 15:25 |
| mriedem | even if you rolled rocky with a placement database, it gets the nova_api schema doesn't it? | 15:25 |
| mriedem | but just doesn't use those non-placement tables | 15:26 |
| openstackgerrit | Sundar Nadathur proposed openstack/nova-specs master: Nova Cyborg interaction specification. https://review.openstack.org/603955 | 15:27 |
| edleafe | I was thinking about whether starting with a db with just the related tables and starting with the whole nova api db would be an issue with alembic | 15:28 |
| edleafe | Alembic is new to me, and based on the docs I've read, it seemed like it *might* be an issue | 15:28 |
| cdent | edleafe: sounds like a job for just trying it | 15:29 |
| edleafe | So I was just seeing if we had agreed on a preferred path for the new db | 15:29 |
| cdent | since we haven't got rid of the unused models yet, might be an option | 15:29 |
| cdent | that create script on https://review.openstack.org/#/c/600161/ only uses the ones that are actually used | 15:29 |
| edleafe | ok, thanks - let me dig further | 15:30 |
| openstackgerrit | Matt Riedemann proposed openstack/placement master: Add nova database migration script for postgresql https://review.openstack.org/604028 | 15:33 |
| mriedem | i don't know anything about alembic, but i'd think it shouldn't care | 15:33 |
| mriedem | unless it scans the database and compares against the model and complains that you have tables not in the model | 15:33 |
| mriedem | zzzeek probably has the answer off the top of his head | 15:34 |
| dansmith | which is kinda the point of it I think | 15:34 |
| dansmith | I dunno if it cares or not, but I expect it to care more than sqla-migrate | 15:34 |
| openstackgerrit | Matt Riedemann proposed openstack/placement master: Add nova database migration script for postgresql https://review.openstack.org/604028 | 15:36 |
| *** helenafm has quit IRC | 15:46 | |
| *** e0ne has quit IRC | 15:50 | |
| *** tssurya has joined #openstack-placement | 16:31 | |
| cdent | dinner dinner dinner | 16:31 |
| *** e0ne has joined #openstack-placement | 17:08 | |
| *** munimeha1 has quit IRC | 17:10 | |
| mnaser | hi | 17:23 |
| mriedem | read everything from nova for the last 30 minutes | 17:23 |
| mriedem | kthx | 17:23 |
| mriedem | https://github.com/openstack/nova/blob/377921103121bc62a3f7fce60c63e30815406851/nova/api/openstack/placement/objects/resource_provider.py#L1924 is busted | 17:24 |
| mnaser | ..yeah, along those lines | 17:24 |
| mnaser | and breaks upgrades from queens to rocky | 17:24 |
| mriedem | doesn't handle multiple allocations from the same consumer on the same provider | 17:24 |
| mriedem | because the insert_from_select is getting select results with duplicate consumer ids | 17:24 |
| mriedem | 3 instances on the same provider, 3 resource classes per allocation (vcpu, disk, memory_mb) | 17:24 |
| mriedem | https://review.openstack.org/#/c/565405/26/nova/tests/functional/api/openstack/placement/db/test_consumer.py appears to not catch it because it has 3 consumers each with a single allocation record | 17:25 |
| mriedem | i'm trying to recreate | 17:25 |
| mriedem | mnaser: can you report a bug in the interim? | 17:25 |
| mnaser | mriedem: yes | 17:25 |
| mnaser | mriedem: https://bugs.launchpad.net/nova/+bug/1798163 | 17:30 |
| openstack | Launchpad bug 1798163 in OpenStack Compute (nova) "Placement incomplete consumers online migration fails" [Undecided,New] | 17:30 |
| mnaser | mriedem: i think i have figured out how to breaktests | 17:33 |
| mriedem | i think i do too | 17:35 |
| mnaser | trying it | 17:35 |
| mnaser | mriedem: http://paste.openstack.org/show/732266/ to create a more realistic layout of the db | 17:36 |
| mriedem | i got it | 17:37 |
| mnaser | mriedem: as in failed test or failed test + fix? | 17:39 |
| mriedem | failed test | 17:39 |
| mriedem | pushing up in a sec | 17:39 |
| mnaser | mriedem: cool ill have a look at yours and compare to what i have too (i got oslo_db.exception.DBDuplicateEntry locally in tests with what i have too) | 17:40 |
| mriedem | https://review.openstack.org/611113 | 17:49 |
| * mriedem eats lunch | 17:49 | |
| *** e0ne has quit IRC | 17:59 | |
| *** e0ne has joined #openstack-placement | 18:15 | |
| *** e0ne has quit IRC | 18:22 | |
| *** tssurya has quit IRC | 18:52 | |
| *** cdent has quit IRC | 20:24 | |
| openstackgerrit | Sundar Nadathur proposed openstack/nova-specs master: Nova Cyborg interaction specification. https://review.openstack.org/603955 | 20:32 |
| openstackgerrit | Matt Riedemann proposed openstack/placement master: Add recreate test for bug 1798163 https://review.openstack.org/611164 | 20:59 |
| openstack | bug 1798163 in OpenStack Compute (nova) "Placement incomplete consumers online migration fails" [Critical,In progress] https://launchpad.net/bugs/1798163 - Assigned to Matt Riedemann (mriedem) | 20:59 |
| openstackgerrit | Matt Riedemann proposed openstack/placement master: Use unique consumer_id when doing online data migration https://review.openstack.org/611165 | 20:59 |
| mriedem | ^ are the mirror cherry picks of the nova changes | 21:00 |
| *** mriedem is now known as mriedem_away | 21:02 | |
| *** mriedem_away has quit IRC | 21:55 | |
| *** smcginnis is now known as smcginnis_vaca | 21:55 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!