*** takashin has joined #openstack-placement | 00:24 | |
*** efried has quit IRC | 01:49 | |
*** efried has joined #openstack-placement | 02:01 | |
*** tetsuro has joined #openstack-placement | 02:23 | |
*** tetsuro has quit IRC | 04:33 | |
*** tetsuro has joined #openstack-placement | 04:33 | |
*** openstackgerrit has joined #openstack-placement | 05:46 | |
openstackgerrit | Tetsuro Nakamura proposed openstack/placement master: De-nova-ify doc/source/index.rst https://review.openstack.org/613200 | 05:46 |
---|---|---|
*** gibi has joined #openstack-placement | 07:23 | |
mnaser | efried: cdent: the thing we were talking about is def a bug -- https://bugs.launchpad.net/nova/+bug/1799892 | 07:33 |
openstack | Launchpad bug 1799892 in OpenStack Compute (nova) "Placement API crashes with 500s in Rocky upgrade with downed compute nodes" [Undecided,New] | 07:33 |
*** helenafm has joined #openstack-placement | 07:38 | |
*** helenafm has quit IRC | 08:08 | |
*** ttsiouts has joined #openstack-placement | 08:26 | |
*** takashin has left #openstack-placement | 08:31 | |
openstackgerrit | Tetsuro Nakamura proposed openstack/nova-specs master: Spec: Support filtering by forbidden aggregate https://review.openstack.org/603352 | 08:36 |
*** e0ne has joined #openstack-placement | 08:42 | |
*** helenafm has joined #openstack-placement | 08:47 | |
*** ttsiouts has quit IRC | 09:02 | |
*** ttsiouts has joined #openstack-placement | 09:02 | |
*** ttsiouts has quit IRC | 09:21 | |
*** ttsiouts has joined #openstack-placement | 09:22 | |
*** cdent has joined #openstack-placement | 10:10 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Make tox -ereleasenotes work https://review.openstack.org/613118 | 10:11 |
*** ttsiouts has quit IRC | 10:11 | |
*** ttsiouts has joined #openstack-placement | 11:04 | |
*** gibi has quit IRC | 12:03 | |
*** gibi has joined #openstack-placement | 12:16 | |
*** tetsuro has quit IRC | 12:23 | |
*** ttsiouts has quit IRC | 12:47 | |
*** ttsiouts has joined #openstack-placement | 12:48 | |
cdent | edleafe: as promised, but late: https://anticdent.org/quick-placement-development.html | 12:50 |
cdent | (others may find it useful too) | 12:51 |
efried | cdent: Looks like tetsuro has a fix proposed. I've only glanced at it, but I thought we had something deeper in the stack that was supposed to take care of null root provider IDs. Also, mnaser says he was *upgrading* to rocky, and I thought we (jaypipes/dansmith) had upgrade code to make sure these fields got populated; so wouldn't the bug be in that upgrade code? | 12:57 |
jaypipes | efried: there's only inline data migrations for NULL root_provider_id, IIRC. there was never a blocker migration or offline migration added. | 12:59 |
efried | oh, okay. | 12:59 |
cdent | I would have thought so too, but at the same time we probably need to be robust in the face of the database getting out of wack because in the field people _constantly_ mess with the live database | 12:59 |
openstackgerrit | Ed Leafe proposed openstack/placement master: Consider root id is None in the database case https://review.openstack.org/613320 | 13:07 |
edleafe | cdent: ^^ This is Tesuro's bug fix ported to the placement tree | 13:09 |
cdent | edleafe: are you planning to do the recreate test patch too? | 13:12 |
edleafe | I didn't see the need | 13:12 |
edleafe | that adds the failing test, commented out | 13:12 |
cdent | i suppose that makes sense | 13:14 |
openstackgerrit | Adam Spiers proposed openstack/nova-specs master: Add spec for libvirt driver launching AMD SEV-encrypted instances https://review.openstack.org/609779 | 13:15 |
edleafe | I get why separate patches are important in the original branch, but we only need to copy the result. | 13:16 |
edleafe | And if there are tweaks needed to the nova side, I'll copy 'em to the placement side. | 13:17 |
edleafe | Guess I should -W that patch until the nova side merges | 13:17 |
cdent | aye | 13:17 |
cdent | edleafe, efried, jaypipes : is the script in that blog ^ post worth sticking in the repo as a "tool" of some kind? | 13:46 |
* efried clicks | 13:46 | |
jaypipes | cdent: sure. just need to make sure it doesn't get out of date quickly (hopefully it shouldn't, but deployment tooling is a world of shambles and tears, as I've learned, full of dead-end code and outdated configurations) | 13:48 |
cdent | jaypipes: yeah, that's exactly why I'm hesitant to make it a thing. It's supposed to be more of a demo of "look how easy it is for you to do this yourself" not "this is the way to do a thing" | 13:50 |
*** efried has quit IRC | 13:56 | |
*** efried has joined #openstack-placement | 13:57 | |
jaypipes | cdent: ack | 13:57 |
efried | cdent: Typo on L46 of your script: Creat | 13:59 |
efried | Wonder whether it would be better or worse to | 14:00 |
efried | source .tox/py36/bin/activate | 14:00 |
efried | so you don't have to path your uwsgi command | 14:00 |
efried | Also, if it were me, I would capitalize IDENTIFIED BY | 14:01 |
efried | cdent: This is cool. Not sure if it's worth trying to maintain the actual script under source control, but I think including this article in the documentation would be ++ | 14:05 |
edleafe | cdent: won't have time to play with this until this afternoon. But a quick read looks promising | 14:11 |
cdent | efried: I tend to avoid activate because I always forget to deactivate | 14:12 |
edleafe | And yeah, I wouldn't maintain a tool, even with a zillion disclaimers. People will see it and assume "this is the one and only way to do things" | 14:12 |
cdent | and IDENTIFIED BY was copied from devstack, so if you _really_ dont like it you can submit a patch | 14:13 |
efried | shrug, just a preference to make it easy to distinguish the SQL tokens from other stuff in a command. But since in this case the other stuff is uppercase env vars anyway, it wouldn't do that much good. | 14:13 |
edleafe | cdent: add this to your .bashrc: http://paste.openstack.org/show/733048/ | 14:14 |
cdent | edleafe: I have that, makes no difference | 14:14 |
cdent | actually activate does that for you, as I recall | 14:14 |
cdent | anyway, you all have convinced me not to try to put the script in there because this kind of nitpicking is exactly what's _not_ needed, pretty much ever | 14:15 |
cdent | but the document makes some sense | 14:15 |
cdent | Is someone interested in translating it from cdent-ease, or should it just go in as is? | 14:15 |
cdent | is that implicit assent? ;) | 14:18 |
edleafe | Why would you want to translate to something less wonderful? | 14:18 |
*** ttsiouts has quit IRC | 14:24 | |
*** ttsiouts has joined #openstack-placement | 14:26 | |
-openstackstatus- NOTICE: Zuul and Nodepool services are being restarted to migrate them to a new Zookeeper cluster. THis brings us an HA database running on newer servers. | 14:40 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Add a doc describing a quick live environment https://review.openstack.org/613343 | 14:51 |
*** helenafm has quit IRC | 15:18 | |
efried | cdent: Sorry, I missed that conversation. I would take out most of the first paragraph. (Haven't looked at the review yet.) | 15:20 |
cdent | efried: i changed it somewhat, but is still folksy | 15:21 |
efried | yeah, I guess it would probably be better to make it slightly more formal. | 15:21 |
cdent | comment on the review. my preference would be that we get tone tune ups in followups | 15:21 |
cdent | otherwise we'll spin forever | 15:21 |
efried | cdent: It really doesn't need much. | 15:24 |
cdent | then a merge and a fup would be idea, perhaps https://pypi.org/project/git-nit/ | 15:25 |
efried | cdent: How about one respin, and if there's more nits, a fup? | 15:26 |
cdent | if you like, I don't mind _that_ much, I'm simply trying to change the culture a bit. we should be predisposed to merging and fixing later as a habit so that we are more accessible to people who are not around all the time like we are | 15:28 |
cdent | mostly because the number of people is dropping like a stone | 15:30 |
efried | My concern with that kind of culture shift is when the fups get forgotten or lost in the shuffle. Also generating noise in gerrit, the review tree, and git blame. When it's something quick and easy, we should fix in the same commit. | 15:30 |
-openstackstatus- NOTICE: The Zuul and Nodepool database transition is complete. Changes updated during the Zuul outage may need to be rechecked. | 15:31 | |
cdent | If the fups get forgotten or lost then they didn't really need to happen. That's kind of the point. The thing we want to be driving change is the experience of people out in the wild using the stuff, not our desire or ability to predict and overengineer everything we can think of | 15:32 |
edleafe | Perfection is the enemy of the good, or something like that | 15:33 |
openstackgerrit | Chris Dent proposed openstack/placement master: Add a doc describing a quick live environment https://review.openstack.org/613343 | 15:35 |
cdent | did you see https://twitter.com/anticdent/status/1054387689468936192 from a few days ago (I assume not, since you don't watch that show) | 15:35 |
openstackgerrit | Chris Dent proposed openstack/placement master: Add a doc describing a quick live environment https://review.openstack.org/613343 | 15:39 |
openstackgerrit | Merged openstack/nova-specs master: Support initial allocation ratios https://review.openstack.org/552105 | 15:56 |
*** e0ne has quit IRC | 15:59 | |
sean-k-mooney | o/ | 16:10 |
sean-k-mooney | melwitt: cdent im not sure where we stand with https://review.openstack.org/#/c/610034 by the way | 16:13 |
cdent | sean-k-mooney: I think the latest comments are right: we need a small backportable change, followed up by a more complete change | 16:14 |
cdent | whether it is a decorator or not, I'm not sure, but the concept seems right | 16:14 |
melwitt | agree. and the small backportable change should be to avoid calling configure twice | 16:14 |
melwitt | and not examine _started | 16:15 |
sean-k-mooney | well the run once decorator as also something i stared writing and then just made it a flag because that was less code | 16:15 |
sean-k-mooney | ok ill remove the started check | 16:15 |
sean-k-mooney | so just one global _configured flag for now and an if + log messange if we skip reconfiguring? | 16:15 |
sean-k-mooney | i can remove that second unit test then also | 16:16 |
melwitt | sounds OK to me. zzzeek liked the decorator idea but I don't think it matters how you do it, just do something to track whether configure has been called already and don't call it a second time | 16:16 |
sean-k-mooney | melwitt: well i litally started wrigint a call_once decorator and then delted it before the last patch | 16:17 |
sean-k-mooney | so i can do it with a run_once decorator if people want that | 16:17 |
sean-k-mooney | it just was extra code and i assumed people would assume it was over engeirnincg to add the decorator insead of an if | 16:18 |
melwitt | up to you, dunno what others will think | 16:21 |
*** mnaser has quit IRC | 16:26 | |
*** mnaser has joined #openstack-placement | 16:26 | |
*** ttsiouts has quit IRC | 16:36 | |
sean-k-mooney | ok im going to take a break and grab a bite to eat | 16:46 |
sean-k-mooney | cdent: do you want me to move the configure once check external to that module | 16:46 |
sean-k-mooney | or are you ok with me doing it in db_api.py | 16:47 |
cdent | sean-k-mooney: for the backportable thing I think it is fine in db_api | 16:47 |
cdent | but when we do the more complete followup that could go away, and the clean-slate handling ought to be in wsgi.py | 16:47 |
sean-k-mooney | ok cool then for the intiall patch ill do that. | 16:47 |
sean-k-mooney | should i change it to patial bug by the way | 16:47 |
sean-k-mooney | and use the same bug for the follow up or have a specless blueprint for the follow up | 16:48 |
cdent | no, I think closes is fine, because the follow up is a bit more ... yeah that | 16:48 |
sean-k-mooney | cool thanks | 16:48 |
sean-k-mooney | i spoke breifly to pbourke on the kolla channel about moving to uwsgi also | 16:49 |
sean-k-mooney | he seamed open to the idea just an fyi | 16:49 |
sean-k-mooney | melwitt: cdent ill resping the patch and push it up for review in about 2 hours or so | 16:50 |
cdent | yay! | 16:51 |
cdent | (to both) | 16:51 |
* cdent goodnights | 17:56 | |
*** cdent has quit IRC | 17:56 | |
efried | jaypipes, edleafe: Is parent_provider_id NULL when I'm a root, or is it == my ID == root_provider_id? | 18:54 |
edleafe | efried: IIRC, it's NULL | 18:56 |
jaypipes | NULL | 19:06 |
openstackgerrit | Matt Riedemann proposed openstack/osc-placement stable/queens: Do not depend on jenkins user in devstack gate https://review.openstack.org/556666 | 19:23 |
openstackgerrit | Matt Riedemann proposed openstack/osc-placement stable/queens: Migrate legacy-osc-placement-dsvm-functional job in-tree https://review.openstack.org/556635 | 19:23 |
efried | jaypipes, edleafe: Is it possible for internal db ID to be zero? | 19:33 |
openstackgerrit | Ed Leafe proposed openstack/placement master: Added alembic environment https://review.openstack.org/611441 | 19:33 |
openstackgerrit | Ed Leafe proposed openstack/placement master: Clean up the old migrations and models https://review.openstack.org/611440 | 19:33 |
edleafe | efried: sure, why not? | 19:34 |
efried | just didn't know if we started 'em at 1 or something. | 19:34 |
edleafe | We might, but why would that matter? | 19:34 |
edleafe | They are defined like this: id = Column(Integer, primary_key=True, nullable=False) | 19:35 |
edleafe | So I guess it would depend on the database implementation | 19:35 |
efried | It would make it possible to do | 19:37 |
efried | foo or bar | 19:37 |
efried | instead of | 19:37 |
efried | foo if foo is not None else bar | 19:37 |
efried | But we shouldn't do that anyway. | 19:37 |
edleafe | It shouldn't have any meaning | 19:38 |
edleafe | so no, don't do that :) | 19:38 |
jaypipes | ++ | 19:42 |
*** irclogbot_2 has joined #openstack-placement | 20:01 | |
*** edleafe_ has joined #openstack-placement | 20:01 | |
*** mriedem has joined #openstack-placement | 20:03 | |
*** irclogbot_2 has quit IRC | 20:22 | |
*** takashin has joined #openstack-placement | 20:50 | |
*** edleafe_ has left #openstack-placement | 21:06 | |
openstackgerrit | Merged openstack/placement master: De-nova-ify doc/source/index.rst https://review.openstack.org/613200 | 21:09 |
*** irclogbot_2 has joined #openstack-placement | 21:14 | |
*** efried is now known as pot | 21:43 | |
*** pot is now known as efried | 21:43 | |
*** mriedem has quit IRC | 22:21 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!