| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-03-14 | |||
| 14:30:37 | mriedem | i'm back, caught up | |
| 14:30:45 | jaypipes | mriedem: I'm strongly opposed to a separate client for placement. Having an openstackclient module for resource providers and inventories/allocations/aggregate associations is A-OK, but not a separate python client. | |
| 14:30:50 | mriedem | so no to the client for APIs, and no to CLI for osc? | |
| 14:30:57 | bauzas | it's a valid concern, and if people don't really want to have that common client, plus the concerns you mention, then I'm okay | |
| 14:31:27 | mriedem | i'm good with not having a python API client | |
| 14:31:37 | bauzas | jaypipes: OSC being only a CLI wrapper ? | |
| 14:31:38 | mriedem | i was wondering more about CLI | |
| 14:31:48 | mriedem | osc is a cli wrapper, not an api library | |
| 14:31:54 | mriedem | it uses the other client API libraries and the sdk | |
| 14:31:57 | bauzas | yeah, just wanted to reassure that | |
| 14:32:08 | bauzas | that we won't push smart code in OSC | |
| 14:32:40 | bauzas | again, if people feel confident with that, let's stick with dansmith's plan given in the ML | |
| 14:32:53 | bauzas | I don't want to divert too much on that | |
| 14:33:08 | bauzas | and we could raise a priority on writing a python client module if we need to | |
| 14:35:19 | jaypipes | mriedem: OSC doesn't *have* to import a placementclient lib, though. for example, we could just have make_client() return something similar to the keystone session object in the scheduler report client. | |
| 14:35:30 | jaypipes | mriedem: in other words, here's the compute example make_client() https://github.com/openstack/python-openstackclient/blob/master/openstackclient/compute/client.py#L38 | |
| 14:35:55 | jaypipes | mriedem: we could just as easily change that code to just make a "client" for placement be the requests session and keystoneauth session | |
| 14:38:41 | bauzas | jaypipes: given how we heavily mention OSC in our docs, and given we would need that for shared storage stories, is there someone actively working on that? | |
| 14:38:45 | bauzas | for Pike, I mean | |
| 14:38:48 | jroll | jaypipes: your ironic get_inventory() thing lgtm, nice work | |
| 14:39:28 | jaypipes | bauzas: pretty sure we can find folks, yes. | |
| 14:40:08 | bauzas | cool :) | |
| 14:40:24 | mriedem | jaypipes: i know, i was actually thinking we'd have the osc plugin in nova, | |
| 14:40:26 | openstackgerrit | Alex Szarka proposed openstack/nova master: Transform instance-live_migration_abort notification https://review.openstack.org/420453 | |
| 14:40:32 | mriedem | and that gets split out with the server api stuff when it all happens | |
| 14:40:54 | mriedem | i thought projects were supposed to host osc plugins now, rather than try and put them all in osc itself and make that core team own them, | |
| 14:41:01 | mriedem | that's what i remember from a session in austin anyway | |
| 14:41:15 | jaypipes | mriedem: that's fine by me. | |
| 14:41:39 | mriedem | sdague: dansmith: are you against an osc plugin living in nova purely for CLI? | |
| 14:41:46 | mriedem | or meh? | |
| 14:41:59 | dansmith | against existence, or against living in nova? | |
| 14:42:04 | jaypipes | mriedem: probably a question for dtroyer too? | |
| 14:42:08 | mriedem | the former | |
| 14:42:11 | dansmith | no | |
| 14:42:29 | openstackgerrit | Béla Vancsics proposed openstack/nova master: Transform instance.trigger_crash_dump notification https://review.openstack.org/396225 | |
| 14:42:40 | dansmith | there is that thing already yes? | |
| 14:42:48 | mriedem | dansmith: i don't see one | |
| 14:42:58 | mriedem | mlavalle's is a full blown client, like novaclient | |
| 14:43:03 | dansmith | oh, I thought it was a plugin | |
| 14:43:15 | mriedem | https://github.com/miguellavalle/python-placementclient/tree/adding-grp-support | |
| 14:44:07 | sdague | mriedem: in the nova tree? | |
| 14:44:30 | mriedem | sdague: yes, where placement currently lives | |
| 14:44:33 | sdague | I'm definitely -1 on in the nova tree, those things shouldn't be coupled that way | |
| 14:44:44 | mriedem | so openstack/placement-osc-plugin? | |
| 14:44:52 | sdague | mriedem: sure | |
| 14:45:05 | sdague | it can be under nova control, but you don't want to pip install all of nova | |
| 14:45:14 | sdague | to get 200 lines of osc plugin | |
| 14:45:14 | mriedem | oh right i didn't think about that | |
| 14:45:20 | bauzas | good point | |
| 14:45:21 | sean-k-mooney | mriedem: would it not make sense to have the placement osc plugin live in the placement repo when that is created | |
| 14:45:29 | mriedem | sean-k-mooney: but that's not pike, | |
| 14:45:32 | mriedem | or maybe even queens | |
| 14:45:39 | sdague | sean-k-mooney: no, it would not | |
| 14:45:51 | sdague | osc plugins should not be in the trees of larger projects | |
| 14:45:55 | bauzas | yeah | |
| 14:45:56 | jroll | mriedem: how we do osc stuff is the osc plugin goes in the python-ironicclient tree | |
| 14:46:08 | sdague | because osc plugins should be super minimal | |
| 14:46:21 | jroll | seems to work well, and you'll need the ironicclient dep anyway | |
| 14:46:34 | mriedem | jroll: we wouldn't have a dep on a placementclient though | |
| 14:46:49 | sean-k-mooney | sdague: oh ok. i guess you can use depends-on to track patches to both but i just assumend it would be eaisier to keep them in sync if the clients and apis were in the same repo | |
| 14:46:55 | openstackgerrit | Alex Szarka proposed openstack/nova master: Transform instance-exists notification https://review.openstack.org/403660 | |
| 14:47:07 | sdague | sean-k-mooney: the whole point of osc is to optimize for consumption | |
| 14:47:32 | sean-k-mooney | sdague: ah so you want them to be seperate to minimies the code footprint | |
| 14:47:34 | jroll | mriedem: ah, yeah I guess just it's own repo makes sense to me then | |
| 14:47:43 | sdague | and, honestly, I'm not really sure why we would do things that way vs. just put it into osc directly | |
| 14:47:55 | mriedem | sdague: i'm talking with dtroyer in -dev | |
| 14:48:00 | mriedem | if you want to ask about it living in osc directly | |
| 14:48:08 | mriedem | because i thought they didn't want more stuff living in osc directly, | |
| 14:48:09 | openstackgerrit | Alex Szarka proposed openstack/nova master: Transform instance.resize.error notifications https://review.openstack.org/417882 | |
| 14:48:13 | mriedem | just like tempest wants to move things to plugins | |
| 14:51:01 | dansmith | I guess there's no support for generating multiple packages from one source, like deb and rpm can do? | |
| 14:51:51 | jroll | dansmith: pbr doesn't do that, no | |
| 14:53:49 | mriedem | melwitt: did you happen to get a functional test started for https://review.openstack.org/#/c/444106/ ? if not, i could try starting something there | |
| 14:54:16 | dansmith | jroll: yeah, that's a bummer | |
| 14:54:23 | jroll | indeed | |
| 15:10:03 | mriedem | cdent: bauzas: dansmith: sdague: jaypipes: ok here is a simple blueprint to start tracking the osc plugin stuff https://blueprints.launchpad.net/nova/+spec/placement-osc-plugin | |
| 15:10:22 | mriedem | with enough detail to hopefully get someone started, maybe someone from osic - they are an infinite resource right? | |
| 15:11:13 | cdent | ✔ | |
| 15:18:59 | johnthetubaguy | mriedem: I wish we had that ;) | |
| 15:19:18 | johnthetubaguy | raj_singh: I guess everyone is busy right now right? | |
| 15:20:39 | johnthetubaguy | mriedem: but +1 documenting it so someone can take that on | |
| 15:20:49 | sdague | dansmith / jroll it's not just a pbr limitation, there are other substantial issues there with the python packaging toolchain | |
| 15:21:13 | raj_singh | johnthetubaguy: yea running out of bandwidth for this cycle | |
| 15:22:05 | jroll | sdague: that's the understatement of the day :) | |
| 15:22:17 | jroll | I know pbr has hard blockers for it, that's all I wanted to point out | |
| 15:22:20 | mriedem | Kevin_Zheng: i know it's late for you, just fyi that https://review.openstack.org/#/c/429569/ could use some small improvements and then i'm +2 | |
| 15:25:51 | openstackgerrit | Dan Smith proposed openstack/nova master: Make conductor ask scheduler to limit migrates to same cell https://review.openstack.org/438025 | |
| 15:25:55 | dansmith | bauzas: can you look at this ^ ? | |
| 15:26:23 | dansmith | bauzas: I think that's what you're going for.. I still need to do some tests and fix the commit message and stuff, but I think that will efficiently make the caching scheduler behave the same as the filter scheduler | |
| 15:26:41 | bauzas | dansmith: sure, will do | |
| 15:26:47 | dansmith | i.e. filter based on cell per request, from the global collected list | |
| 15:26:58 | bauzas | dansmith: I need to go to the garage soon, but I'll look at it quickly | |
| 15:27:13 | dansmith | just a brief look to see if it's what you meant would be good | |
| 15:27:20 | dansmith | then I'll release you to the garage | |
| 15:28:55 | bauzas | dansmith: so https://review.openstack.org/#/c/438025/23/nova/scheduler/caching_scheduler.py | |
| 15:29:22 | bauzas | dansmith: lgtm, but I was also thinking about _get_by_hosts() returning a dict instead of a list | |
| 15:29:38 | bauzas | oops _get_up_hosts | |
| 15:29:53 | bauzas | dansmith: so we could just lookup by the cell name | |
| 15:30:00 | hamzy | hey mriedem I think that I am running into https://bugs.launchpad.net/nova/+bug/1670627 is there a way to clean up the instances from the CLI? I guess that I need to do some DB cleanup? | |
| 15:30:00 | openstack | Launchpad bug 1670627 in OpenStack Compute (nova) ocata "quota is always in-use after delete the ERROR instances " [Critical,In progress] - Assigned to Matt Riedemann (mriedem) | |