| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-04-30 | |||
| 17:04:41 | dansmith | especially since they can fix one-off issues with osc-placement quite easy, and there are community tools to do it | |
| 17:04:45 | openstackgerrit | OpenStack Release Bot proposed openstack/os-vif stable/ussuri: Update .gitreview for stable/ussuri https://review.opendev.org/724712 | |
| 17:04:46 | openstackgerrit | OpenStack Release Bot proposed openstack/os-vif stable/ussuri: Update TOX_CONSTRAINTS_FILE for stable/ussuri https://review.opendev.org/724713 | |
| 17:04:48 | openstackgerrit | OpenStack Release Bot proposed openstack/os-vif master: Update master for stable/ussuri https://review.opendev.org/724714 | |
| 17:04:49 | openstackgerrit | OpenStack Release Bot proposed openstack/os-vif master: Add Python3 victoria unit tests https://review.opendev.org/724715 | |
| 17:04:49 | artom | dansmith, so, full disclosure, I don't really care one or another :P | |
| 17:04:53 | dansmith | the stein one, while self-contained is already pretty smelly | |
| 17:05:06 | artom | dansmith, as I said, it was purely procedural for me - I wanted to do it one release at a time, and upstream CI on it | |
| 17:05:07 | dansmith | and beyond stein I think it's unreasonable without a major issue, personally | |
| 17:05:17 | dansmith | artom: ack | |
| 17:05:34 | bauzas | lemme look why we needed this command | |
| 17:05:40 | bauzas | ie. which bugs leaked the allocations | |
| 17:05:58 | bauzas | some of them were old and are fixed, some of them still exist IIRC | |
| 17:06:05 | artom | dansmith, so I'd be perfectly comfortable doing it downstream-only | |
| 17:06:17 | artom | dansmith, assuming you don't destroy the case for that as well ;) | |
| 17:06:33 | artom | I probably didn't do enough "due diligence", and just blindly went ahead and started cherry-picking | |
| 17:06:48 | dansmith | artom: you know I just love crushing hopes and dreams right? | |
| 17:06:53 | dansmith | (I really don't despite what you probably think) | |
| 17:07:03 | artom | dansmith, I know. How do you think I became such an empty husk? :) | |
| 17:07:53 | bauzas | https://review.opendev.org/#/c/670112/14/nova/tests/functional/test_nova_manage.py are the cases we verify | |
| 17:16:14 | bauzas | so I've tested three cases | |
| 17:16:25 | openstackgerrit | Stephen Finucane proposed openstack/nova master: docs: Note the ``hw_numa_nodes`` image property https://review.opendev.org/683849 | |
| 17:16:32 | bauzas | 1/ deleted instances leak allocations | |
| 17:16:50 | bauzas | 2/ resize confirm can also leak allocations | |
| 17:17:02 | bauzas | 3/ deleted compute evacuate too | |
| 17:17:22 | bauzas | now, the question is, which bugs are those 3 checks related with ? | |
| 17:17:46 | bauzas | the last one is easy | |
| 17:18:01 | bauzas | https://bugs.launchpad.net/nova/+bug/1829479 | |
| 17:18:01 | openstack | Launchpad bug 1829479 in OpenStack Compute (nova) "The allocation table has residual records when instance is evacuated and the source physical node is removed" [Medium,In progress] - Assigned to Matt Riedemann (mriedem) | |
| 17:22:04 | bauzas | dansmith: artom: just to finish, I think https://bugs.launchpad.net/nova/+bug/1829479 impacts all releases up to now | |
| 17:22:04 | openstack | Launchpad bug 1829479 in OpenStack Compute (nova) "The allocation table has residual records when instance is evacuated and the source physical node is removed" [Medium,In progress] - Assigned to Matt Riedemann (mriedem) | |
| 17:22:19 | bauzas | which makes the audit command a good workaround for it | |
| 17:22:46 | bauzas | anyway, /me disappears | |
| 17:22:50 | dansmith | okay I'm confused.. are we still arguing to backport this or what? | |
| 17:23:53 | dansmith | did we fix the leak due to evacuation or are we saying the fix for that is just to have the audit command? | |
| 17:24:05 | dansmith | obviously for things that happened before such a fix, audit could help | |
| 17:26:35 | bauzas | dansmith: nope, we haven't fixed it yet | |
| 17:26:50 | dansmith | okay I would be in favor of backporting *that* :) | |
| 17:26:53 | bauzas | the audit command just cleans the mess | |
| 17:28:00 | bauzas | anyway, I need a beer | |
| 17:49:36 | mriedem | https://review.opendev.org/#/c/678100/ | |
| 17:57:25 | openstackgerrit | sean mooney proposed openstack/nova master: [DNM] testing with force_legacy_port_binding workaround https://review.opendev.org/724387 | |
| 18:03:19 | melwitt | mriedem: want an e-r stat? https://review.opendev.org/724719 | |
| 18:04:18 | mriedem | bauzas: no just linking in bugs that are assigned to me | |
| 18:08:41 | mriedem | melwitt: commented | |
| 18:08:59 | mriedem | it's weird but the before and after on that query are on the same change, and in the new query they are all coming from the same file (n-cpu in the subnode) | |
| 18:09:02 | mriedem | is that expected? | |
| 18:09:15 | mriedem | was qemu or libvirt bumped in devstack on master? | |
| 18:11:17 | melwitt | mriedem: ah dangit ... I focused on the error log message and didn't notice the warning log is what's in the e-r query. so the query doesn't need an update. sorry :( | |
| 18:13:49 | openstackgerrit | sean mooney proposed openstack/nova master: silence amqp heartbeat warning https://review.opendev.org/724188 | |
| 18:13:50 | openstackgerrit | sean mooney proposed openstack/nova master: remove support of oslo.messagiang 9.8.0 warning message https://review.opendev.org/724725 | |
| 18:13:59 | melwitt | and since the bug was greyed out on http://status.openstack.org/elastic-recheck/#1840159 at the time, that reinforced my thinking that the query didn't pick it up. sigh | |
| 18:21:40 | mriedem | \o/ | |
| 18:33:54 | gmann | dansmith: gibi this is ready - https://review.opendev.org/#/c/723645/ | |
| 18:49:04 | dansmith | gmann: a couple more wording clarification tings | |
| 18:49:06 | dansmith | *things | |
| 19:18:18 | openstackgerrit | Ghanshyam Mann proposed openstack/nova master: Add nova-status upgrade check and reno for policy new defaults https://review.opendev.org/723645 | |
| 19:18:21 | gmann | dansmith: updated ^^ | |
| 19:19:43 | dansmith | gmann: okay lemme just fix another thing in-line to avoid you having to do it | |
| 19:20:00 | gmann | dansmith: sure, thanks | |
| 19:20:56 | dansmith | gmann: is it "recommended" or "required" to enable enforce_scope if you have scope checking rules? | |
| 19:21:02 | dansmith | I thought required | |
| 19:21:22 | gmann | dansmith: this is separate flaf not scope one - enforce_new_defaults | |
| 19:21:24 | gmann | flag | |
| 19:21:42 | gmann | to switch to new default without overwriting the file | |
| 19:22:01 | dansmith | gmann: right, but if you have scope-checking rules and want to keep them, then you need enforce_scope=True right? | |
| 19:22:14 | dansmith | you're checking enforce_scope, but if that's disabled, you tell them to enable enforce_new_defaults | |
| 19:23:07 | gmann | dansmith: ok, so enforce_scope=True is required and (enforce_new_defaults=True OR file overwrite) to move to new things | |
| 19:23:24 | dansmith | right, I think you're telling them the wrong thing | |
| 19:23:33 | dansmith | let me comment for context and you can see if I'm right | |
| 19:23:39 | gmann | in that sentence i was targeting only file overwrite alternate via flag | |
| 19:24:00 | gmann | but i agree we can add enforce_scope=True also for full context | |
| 19:25:19 | dansmith | but the sentence is about keeping the new format | |
| 19:25:24 | dansmith | U cinnebted | |
| 19:25:26 | dansmith | whoa | |
| 19:25:28 | dansmith | "I commented" | |
| 19:33:26 | gmann | dansmith: replied, https://review.opendev.org/#/c/723645/11/nova/cmd/status.py@358 | |
| 19:33:38 | gmann | added scope and new flag way. | |
| 19:34:24 | dansmith | gmann: ++ | |
| 19:34:28 | gmann | my intention is they are aware of new flag enforce_new_defaults t o avoid legacy way of overwriting the file to new default | |
| 19:34:30 | dansmith | gmann: I will try really hard to +2 the next one | |
| 19:34:39 | gmann | ok, updating. | |
| 19:34:53 | dansmith | gmann: ack, I just think the message should include the error, and "how to fix the error" | |
| 19:40:54 | openstackgerrit | Ghanshyam Mann proposed openstack/nova master: Add nova-status upgrade check and reno for policy new defaults https://review.opendev.org/723645 | |
| 19:41:17 | gmann | dansmith: ^^ updated | |
| 19:44:49 | dansmith | gmann: I think you forgot to remove the old sentence | |
| 19:45:03 | gmann | ohh :( | |
| 19:45:08 | dansmith | otherwise that looks good to me | |
| 19:46:05 | dansmith | melwitt: you around to help me put gmann out of his misery? | |
| 19:47:29 | openstackgerrit | Ghanshyam Mann proposed openstack/nova master: Add nova-status upgrade check and reno for policy new defaults https://review.opendev.org/723645 | |
| 19:47:52 | gmann | dansmith: done, sorry for that. | |
| 19:47:56 | dansmith | gmann: thanks for your patience with me, +2d | |
| 19:47:57 | dansmith | melwitt: ^ | |
| 19:48:10 | dansmith | I gotta stretch my legs, bbl | |
| 19:49:48 | gmann | dansmith: thanks | |
| 19:51:31 | melwitt | yeah I'll get it | |
| 20:13:27 | melwitt | gmann: I found some misspellings and grammar things and what seem to be a few bugs, but I also suggested some additional info. let me know what you think. if you agree with them, I can update and approve if you want. I know you have been through a lot of updates with the policy docs things | |
| 20:13:56 | gmann | melwitt: ok, checking.. | |
| 20:15:13 | melwitt | the main thing I'm worried about is if operators just enable enforce_scope = True when they see the status check, without realizing that means end users will have to request scoped tokens from keystone in order to access the formerly admin, now system scoped admin APIs. if I understand correctly | |
| 20:33:58 | gmann | melwitt: thanks. replied. i linked the new doc there which i can move on those line so that they have complete steps info. | |