| *** tetsuro has joined #openstack-placement | 06:39 | |
| *** tetsuro has quit IRC | 06:48 | |
| openstackgerrit | Brin Zhang proposed openstack/nova-specs master: Support delete_on_termination in volume attach api https://review.openstack.org/612949 | 08:23 |
|---|---|---|
| *** bauzas is now known as bauwser | 08:52 | |
| *** lyarwood is now known as lyarpwood | 08:53 | |
| *** cdent has joined #openstack-placement | 09:04 | |
| *** cdent has quit IRC | 09:06 | |
| *** cdent has joined #openstack-placement | 09:10 | |
| *** ttsiouts has joined #openstack-placement | 09:17 | |
| openstackgerrit | Jie Li proposed openstack/nova-specs master: Support volume-backed server rebuild https://review.openstack.org/532407 | 09:26 |
| *** e0ne has joined #openstack-placement | 09:51 | |
| *** ttsiouts has quit IRC | 10:16 | |
| openstackgerrit | Jie Li proposed openstack/nova-specs master: Support volume-backed server rebuild https://review.openstack.org/532407 | 10:20 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Move ensure_consumer to a new placement.handlers.util https://review.openstack.org/614307 | 10:30 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Harden placement init under wsgi https://review.openstack.org/612444 | 10:30 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Added alembic environment https://review.openstack.org/614350 | 10:30 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Delete the old migrations https://review.openstack.org/611440 | 10:30 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Add a placement-manage CLI https://review.openstack.org/600161 | 10:30 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Remove sqlalchemy-migrate from requirements.txt https://review.openstack.org/614539 | 10:31 |
| *** ttsiouts has joined #openstack-placement | 10:32 | |
| openstackgerrit | Chris Dent proposed openstack/placement master: Add a document for creating DB revisions https://review.openstack.org/614024 | 10:32 |
| openstackgerrit | Chris Dent proposed openstack/placement master: WIP - Show an alembic migration https://review.openstack.org/614025 | 10:32 |
| cdent | edleafe: I made some fixes on https://review.openstack.org/#/c/614350/6 but it still has some warts, at least with pep8, maybe more | 10:38 |
| *** stephenfin is now known as finucannot | 10:44 | |
| *** tssurya has joined #openstack-placement | 10:54 | |
| openstackgerrit | Chris Dent proposed openstack/placement master: Added alembic environment https://review.openstack.org/614350 | 11:21 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Delete the old migrations https://review.openstack.org/611440 | 11:21 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Add a document for creating DB revisions https://review.openstack.org/614024 | 11:21 |
| openstackgerrit | Chris Dent proposed openstack/placement master: WIP - Show an alembic migration https://review.openstack.org/614025 | 11:21 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Remove sqlalchemy-migrate from requirements.txt https://review.openstack.org/614539 | 11:22 |
| openstackgerrit | Chris Dent proposed openstack/placement master: Add a placement-manage CLI https://review.openstack.org/600161 | 11:22 |
| *** cdent has quit IRC | 12:06 | |
| *** ttsiouts has quit IRC | 12:39 | |
| *** jaypipes is now known as leakypipes | 12:56 | |
| *** mriedem has joined #openstack-placement | 13:28 | |
| *** ttsiouts has joined #openstack-placement | 13:32 | |
| *** efried is now known as fried_rice | 13:35 | |
| *** ttsiouts_ has joined #openstack-placement | 14:09 | |
| *** ttsiouts has quit IRC | 14:12 | |
| *** dansmith is now known as SteelyDan | 14:13 | |
| *** tssurya has quit IRC | 14:18 | |
| *** ttsiouts_ has quit IRC | 14:25 | |
| *** ttsiouts has joined #openstack-placement | 14:35 | |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs master: Add support for emulated virtual TPM https://review.openstack.org/571111 | 15:03 |
| *** ttsiouts has quit IRC | 15:04 | |
| *** ttsiouts has joined #openstack-placement | 15:05 | |
| openstackgerrit | Ed Leafe proposed openstack/placement master: Added alembic environment https://review.openstack.org/614350 | 15:18 |
| openstackgerrit | Ed Leafe proposed openstack/placement master: Delete the old migrations https://review.openstack.org/611440 | 15:18 |
| openstackgerrit | Ed Leafe proposed openstack/placement master: Add a document for creating DB revisions https://review.openstack.org/614024 | 15:18 |
| openstackgerrit | Ed Leafe proposed openstack/placement master: WIP - Show an alembic migration https://review.openstack.org/614025 | 15:18 |
| openstackgerrit | Ed Leafe proposed openstack/placement master: Add a placement-manage CLI https://review.openstack.org/600161 | 15:24 |
| *** fried_rice is now known as fried_rolls | 16:30 | |
| *** e0ne has quit IRC | 16:33 | |
| *** ttsiouts has quit IRC | 16:38 | |
| *** ttsiouts has joined #openstack-placement | 16:39 | |
| *** ttsiouts has quit IRC | 16:43 | |
| *** cdent has joined #openstack-placement | 17:29 | |
| *** mriedem has quit IRC | 17:30 | |
| openstackgerrit | Chris Dent proposed openstack/placement master: Remove sqlalchemy-migrate from requirements.txt https://review.openstack.org/614539 | 17:34 |
| *** mriedem has joined #openstack-placement | 17:35 | |
| cdent | edleafe: those tests that I commented out because they don't work are still not working: http://logs.openstack.org/50/614350/8/check/openstack-tox-functional/5f5b7dc/testr_results.html.gz | 17:42 |
| openstackgerrit | Jim Rollenhagen proposed openstack/nova-specs master: Use conductor groups to partition nova-compute services for Ironic https://review.openstack.org/609709 | 17:43 |
| edleafe | cdent: what are you doing here? You're supposed to be off | 17:43 |
| cdent | I was travelling for work, I'm home now | 17:43 |
| edleafe | I wanted to see if they would pass on zuul, as they were passing locally | 17:44 |
| cdent | were you using mysql or postgresql as your openstackcitest database? | 17:45 |
| edleafe | mysql | 17:45 |
| cdent | too much magic | 17:45 |
| cdent | but it creeps ever closer | 17:45 |
| edleafe | I have to run out to a doctor appt now, but I will try to play a bit more when I get back | 17:47 |
| -openstackstatus- NOTICE: OpenStack infra's mirror nodes stopped accepting connections on ports 8080, 8081, and 8082. We will notify when this is fixed and jobs can be rechecked if they failed to communicate with a mirror on these ports. | 18:09 | |
| *** cdent has quit IRC | 18:24 | |
| *** e0ne has joined #openstack-placement | 18:34 | |
| *** fried_rolls is now known as fried_rice | 18:46 | |
| -openstackstatus- NOTICE: The firewall situation with ports 8080, 8081, and 8082 on mirror nodes has been resolved. You can recheck jobs that have failed to communicate to the mirrors on those ports now. | 18:54 | |
| *** e0ne has quit IRC | 19:05 | |
| *** finucannot is now known as stephenfin | 19:35 | |
| openstackgerrit | Jim Rollenhagen proposed openstack/nova-specs master: Use conductor groups to partition nova-compute services for Ironic https://review.openstack.org/609709 | 19:45 |
| *** cdent has joined #openstack-placement | 20:24 | |
| fried_rice | cdent: Move here, since it's a placement discussion? | 20:47 |
| cdent | sure | 20:47 |
| fried_rice | cdent: Yeah, I worked hard to eliminate all the generation comparison in the report client code, for this very reason. | 20:47 |
| cdent | but we seem to have a new understanding of what a generation needs to mean, and the cost of that removal | 20:47 |
| fried_rice | anyway, let's go with it for the moment, and assume I can look at it on the client side and (at least) compare for equality and have that be meaningful. | 20:48 |
| cdent | (as I said on the review of your POC, I think the POC is the right thing for a subset of use cases, but the nested situation goes awry, and it would be good to resolve that in a fairly "light" way) | 20:48 |
| fried_rice | You're talking about extending the meaning of "this thing has changed" to "this thing or any of its descendants have changed" | 20:49 |
| fried_rice | btw, are you percolating the gen bump all the way up the branch, or just effecting it on the root? Either is fraught. | 20:49 |
| cdent | more subtle than that: this thing includes its descendants | 20:49 |
| cdent | I'm saying: if I change, my parents changes, thus it cascades all the way up | 20:49 |
| cdent | s/parents/parent/ | 20:49 |
| fried_rice | But my main beef with this is that now, every time cyborg modifies a device provider, whatever you have on the client side is going to be stale and you'll get a 409 bounce/retry the next time you try to use it. | 20:50 |
| cdent | which is what _should_ happen | 20:50 |
| cdent | if the nested-tree changes, our view of reality is _wrong_ | 20:50 |
| fried_rice | I would have to see a convincing use case whereby a device RP changing its inventory of FPGA regions makes it matter when I change a trait on the compute RP. | 20:51 |
| fried_rice | or whatever. | 20:51 |
| fried_rice | They're separate providers. They're associated with each other, but they're separate. | 20:51 |
| fried_rice | Are you going to kick generation bumps across aggregates too? | 20:52 |
| cdent | I guess I don't understand why there should be such a thing as a ProviderTree object, if the nested tree isn't a "block of reality". Then again, I've tried to argue that there shouldn't be a ProviderTree. | 20:52 |
| fried_rice | By ProviderTree you mean local cache? | 20:53 |
| cdent | I mean the class | 20:53 |
| fried_rice | By itself, ProviderTree is just a convenient way to handle the view of what's in placement. | 20:53 |
| fried_rice | Even if we didn't cache anything, using the ProviderTree class to sling that info around would be a good thing imo. | 20:53 |
| cdent | In a structure that says there is a thing which is a root and has children in a very formal relationship | 20:54 |
| fried_rice | otherwise we're just mucking with raw dicts, which is dangerous. | 20:54 |
| fried_rice | yes, it reflects the relationship that placement itself contains. | 20:54 |
| cdent | re: aggregates, don't know, hadn't thought that far, but first thought is "no" | 20:55 |
| cdent | because there the association has less apparent strength | 20:55 |
| cdent | but I dunno, I need to think about it more, it just came to me in the last few minutes | 20:56 |
| fried_rice | I haven't yet grokked what it would accomplish wrt what my email/WIP is trying to accomplish, though. | 20:56 |
| cdent | it's a follow up to your WIP, that fixes it for nested: it makes it possible to check for tree comparisons by getting just one resource provider (by uuid, which is one of the fastest operations that placement has) | 20:57 |
| cdent | because at some point, with clustered hypervisors and with nested with multiple agents involved, we have to compare and update | 20:58 |
| fried_rice | I have no data, but my gut says GET /rps?in_tree=$u isn't any slower than get /rps/$u for trees under dozens or hundreds of nodes. | 20:59 |
| fried_rice | But one of the things my patch is based on, which was part of the discussion yesterday, is that *nova* doesn't care about updates to any part of the tree except the ones that *nova* makes. | 21:00 |
| cdent | anyway, I need to not be on the puter at 9pm on a friday, I just want to get that idea out there before I forgot it because I want to be sure that we do not fall prey to fixing an already complicated situation without first reviewing the original complexity | 21:00 |
| fried_rice | E.g. if cyborg adds a child provider representing a device, nova doesn't need to know about it. | 21:00 |
| cdent | right: thus my comment above, your patch is right in nearly every situation | 21:00 |
| openstackgerrit | Ed Leafe proposed openstack/placement master: Added alembic environment https://review.openstack.org/614350 | 21:01 |
| openstackgerrit | Ed Leafe proposed openstack/placement master: Delete the old migrations https://review.openstack.org/611440 | 21:01 |
| openstackgerrit | Ed Leafe proposed openstack/placement master: Add a document for creating DB revisions https://review.openstack.org/614024 | 21:01 |
| openstackgerrit | Ed Leafe proposed openstack/placement master: WIP - Show an alembic migration https://review.openstack.org/614025 | 21:01 |
| fried_rice | If you get a chance at some point to scribble down an example of a situation where it's definitely *not* right, that would be appreciated. | 21:01 |
| cdent | from a purely SQL standpoint (which is where we do a lot of the placement work), a select for a single row by primary key is massively more performant than a multi row join | 21:02 |
| cdent | matt made some comments in his followup email that may apply here, I have it on my todo list to review the code and respond to that email properly | 21:03 |
| cdent | so I'll do that, but not right now | 21:03 |
| fried_rice | Again, just gut, but I think the sql is a negligible part of the request | 21:04 |
| fried_rice | in this case | 21:04 |
| fried_rice | Yes, you should go not work right now, tempting though this topic is. | 21:04 |
| cdent | actually, it's not tempting to me at all. I just feel obliged to make sure that we explore the options. | 21:05 |
| fried_rice | I had actually sent out the f'ing email a couple hours earlier right before I hit the mats, but forgot to change my origin address so I had f'ing bounce waiting for me when I got back. | 21:05 |
| cdent | I wish we never did nested providers | 21:05 |
| cdent | But I've come to accept reality as well as I can. But tempting? No. | 21:05 |
| fried_rice | I mean tempting to discuss. | 21:05 |
| fried_rice | clearly | 21:05 |
| fried_rice | gotta run. ttyl. | 21:06 |
| edleafe | cdent: while you're here... | 21:06 |
| cdent | hi edleafe | 21:06 |
| edleafe | I played with a few things and didn't see it make any difference | 21:06 |
| cdent | edleafe: I think it is what I said on my comments that I left as TODO(you): those tests can't work if there aren't >1 revision | 21:07 |
| edleafe | One idea was for the "upgrade twice" test was to insert a blank revision *before* the initial revision | 21:07 |
| cdent | (fried_rice I'd prefer to not even talk about it, but I (by some definition of) have to) | 21:07 |
| edleafe | so the first version is the blank, and the head version is the intiial | 21:08 |
| cdent | edleafe: you could just leave a todo to add those tests in after the first migration. The only reason there are there at all is because you copied them in from somewhere else | 21:08 |
| edleafe | But it doesn't make a difference locally, so I pushed it to see the results on zuul | 21:08 |
| cdent | seems a reasonable plan | 21:09 |
| cdent | Do you know the state of your test database at the beginning of things? Maybe some kind of leftovers? | 21:10 |
| edleafe | cdent: so why don't you bugger off now and enjoy your Friday evening | 21:10 |
| cdent | meh | 21:10 |
| * cdent buggers off | 21:13 | |
| *** cdent has quit IRC | 21:13 | |
| *** ttsiouts has joined #openstack-placement | 21:38 | |
| *** ttsiouts has quit IRC | 21:43 | |
| *** ttsiouts has joined #openstack-placement | 21:53 | |
| *** ttsiouts has quit IRC | 23:17 | |
| *** mriedem has quit IRC | 23:18 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!