| *** diablo_rojo has quit IRC | 00:04 | |
| *** ediardo has quit IRC | 00:37 | |
| *** Rockyg has quit IRC | 03:38 | |
| *** lamt has joined #openstack-meeting-cp | 03:53 | |
| *** lamt has quit IRC | 06:22 | |
| *** lamt has joined #openstack-meeting-cp | 06:32 | |
| *** knangia has joined #openstack-meeting-cp | 06:36 | |
| *** rarcea has joined #openstack-meeting-cp | 06:44 | |
| *** lamt has quit IRC | 07:32 | |
| *** DFFlanders has quit IRC | 08:15 | |
| *** knangia has quit IRC | 09:01 | |
| *** sdague has joined #openstack-meeting-cp | 10:10 | |
| *** lamt has joined #openstack-meeting-cp | 12:21 | |
| *** lamt has quit IRC | 12:22 | |
| *** lamt has joined #openstack-meeting-cp | 12:22 | |
| *** lamt has quit IRC | 12:22 | |
| *** lamt has joined #openstack-meeting-cp | 12:23 | |
| *** lamt has quit IRC | 12:23 | |
| *** Daviey has quit IRC | 12:28 | |
| *** diablo_rojo has joined #openstack-meeting-cp | 12:54 | |
| *** Daviey has joined #openstack-meeting-cp | 12:56 | |
| *** gouthamr has joined #openstack-meeting-cp | 13:28 | |
| *** lamt has joined #openstack-meeting-cp | 14:00 | |
| *** jaugustine has joined #openstack-meeting-cp | 14:56 | |
| *** mtreinish has quit IRC | 14:57 | |
| *** mtreinish has joined #openstack-meeting-cp | 14:58 | |
| *** zerick_ has joined #openstack-meeting-cp | 15:05 | |
| *** lifeless_ has joined #openstack-meeting-cp | 15:07 | |
| *** homerp_ has joined #openstack-meeting-cp | 15:09 | |
| *** melwitt_ has joined #openstack-meeting-cp | 15:09 | |
| *** gnarld_ has joined #openstack-meeting-cp | 15:09 | |
| *** dstanek_ has joined #openstack-meeting-cp | 15:09 | |
| *** lbragstad_ has joined #openstack-meeting-cp | 15:09 | |
| *** zigo_ has joined #openstack-meeting-cp | 15:09 | |
| *** EmilienM_ has joined #openstack-meeting-cp | 15:09 | |
| *** kencjohnston_ has joined #openstack-meeting-cp | 15:10 | |
| *** stvnoyes has quit IRC | 15:10 | |
| *** zerick has quit IRC | 15:10 | |
| *** lbragstad has quit IRC | 15:10 | |
| *** melwitt has quit IRC | 15:10 | |
| *** TheJulia has quit IRC | 15:10 | |
| *** kencjohnston has quit IRC | 15:10 | |
| *** homerp has quit IRC | 15:10 | |
| *** lifeless has quit IRC | 15:10 | |
| *** jgriffith has quit IRC | 15:10 | |
| *** cFouts has quit IRC | 15:10 | |
| *** dstanek has quit IRC | 15:10 | |
| *** EmilienM has quit IRC | 15:10 | |
| *** phealy has quit IRC | 15:10 | |
| *** zigo has quit IRC | 15:10 | |
| *** dstanek_ is now known as dstanek | 15:10 | |
| *** EmilienM_ is now known as 17WAAOC7S | 15:11 | |
| *** 21WAAA2JF has joined #openstack-meeting-cp | 15:11 | |
| *** stvnoyes has joined #openstack-meeting-cp | 15:12 | |
| *** zigo_ is now known as zigo | 15:12 | |
| *** zigo is now known as Guest52586 | 15:12 | |
| *** lbragstad_ is now known as lbragstad | 15:14 | |
| *** lamt has quit IRC | 15:15 | |
| *** TheJulia has joined #openstack-meeting-cp | 15:18 | |
| *** lamt has joined #openstack-meeting-cp | 15:25 | |
| *** tommylikehu has joined #openstack-meeting-cp | 15:36 | |
| ildikov | #startmeeting cinder-nova-api-changes | 16:00 |
|---|---|---|
| openstack | Meeting started Thu Apr 13 16:00:13 2017 UTC and is due to finish in 60 minutes. The chair is ildikov. Information about MeetBot at http://wiki.debian.org/MeetBot. | 16:00 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 16:00 |
| *** openstack changes topic to " (Meeting topic: cinder-nova-api-changes)" | 16:00 | |
| openstack | The meeting name has been set to 'cinder_nova_api_changes' | 16:00 |
| ildikov | DuncanT ameade cFouts johnthetubaguy jaypipes takashin alaski e0ne jgriffith tbarron andrearosa hemna erlon mriedem gouthamr ebalduf patrickeast smcginnis diablo_rojo gsilvis xyang1 raj_singh lyarwood breitz jungleboyj | 16:00 |
| jungleboyj | o/ | 16:00 |
| smcginnis | ./ | 16:00 |
| tommylikehu | o/ | 16:00 |
| hemna | o/ | 16:01 |
| stvnoyes | o/ | 16:01 |
| ildikov | hi all | 16:02 |
| ildikov | let me check whether we have anyone for today from Nova | 16:02 |
| * johnthetubaguy lurks | 16:02 | |
| johnthetubaguy | so its spec freeze day today | 16:02 |
| *** mriedem has joined #openstack-meeting-cp | 16:03 | |
| mriedem | o/ | 16:03 |
| ildikov | johnthetubaguy: ah, I see, I knew there was something :) | 16:03 |
| lyarwood | o/ | 16:03 |
| ildikov | I have only a few items for today | 16:03 |
| ildikov | johnthetubaguy: I uploaded a new version of check detach with an updated commit message | 16:04 |
| ildikov | I believe that's where we got last week with that discussion | 16:05 |
| ildikov | I tried to add more info about the change | 16:05 |
| ildikov | I also started to look into the detach patch | 16:05 |
| ildikov | as it's an older one we need to look into cases like swap and post migrate | 16:06 |
| ildikov | I thought to move those changes to a follow up patch and have those changes included in the first one that are ready to merge | 16:06 |
| ildikov | mriedem: johnthetubaguy: lyarwood: would that make sense? ^ | 16:07 |
| mriedem | i think we said we'd handle the different operations in different patches | 16:07 |
| mriedem | since none of it turns on until we do attach with new flow at the end | 16:07 |
| mriedem | so let's just handle normal volume detach in that change | 16:07 |
| ildikov | ok, sounds good, I will separate those then | 16:08 |
| lyarwood | agreed, this should also help spread the load here a little, I wasn't sure who was working on this detach change tbh | 16:08 |
| mriedem | i would definitely move this pattern into a private method in the compute manager | 16:09 |
| mriedem | if bdm.is_volume: | 16:09 |
| ildikov | lyarwood: we try to work together with John (Griffith) except when we get overloaded at the same time... | 16:09 |
| mriedem | if not bdm.attachment_id: | 16:09 |
| mriedem | else: | 16:09 |
| mriedem | self.volume_api.attachment_update | 16:09 |
| mriedem | or whatever | 16:09 |
| mriedem | that's duplicated everywhere | 16:09 |
| mriedem | let's do that in a single private method | 16:09 |
| ildikov | lyarwood: as these changes can be independent it's easy to spread the load | 16:09 |
| lyarwood | ildikov: yup agreed | 16:10 |
| ildikov | mriedem: that one BTW will be a delete - create combo, but got your point | 16:10 |
| johnthetubaguy | mriedem: +1 | 16:11 |
| ildikov | lyarwood: would your planned change help with what mriedem is pointing out above? ^ | 16:11 |
| mriedem | i think just using a simple private method to contain that if/else/if logic will help | 16:11 |
| johnthetubaguy | mriedem: I was wondering if we want a DNM patch at the end that does that new style attach, in an attempt to get test coverage sooner? Maybe that just makes things messier? | 16:12 |
| mriedem | rather than overhauling and moving everything to the block_device module before this | 16:12 |
| mriedem | johnthetubaguy: idk at this point | 16:12 |
| ildikov | mriedem: ok, got it | 16:13 |
| lyarwood | I can look into a DNM attach change next week if there are no other takers | 16:13 |
| ildikov | johnthetubaguy: I can try to look into how much the earlier attach PoC works still | 16:13 |
| ildikov | johnthetubaguy: we had it working at some point, but I believe that was before merging the new Cinder API | 16:14 |
| mriedem | i guess we don't detach on shelve offload huh | 16:14 |
| mriedem | i suppose that makes sense | 16:14 |
| mriedem | we need the ports and volumes when we unshelve | 16:14 |
| mriedem | although, when you unshelve, you're building on a new host (maybe) | 16:14 |
| ildikov | johnthetubaguy: there's also some bug fixing activity driven by tommylikehu that has API changes, like rename instance_id to server_id | 16:14 |
| mriedem | API changes? | 16:15 |
| tommylikehu | ildikov: yes | 16:15 |
| ildikov | mriedem: if we have an empty attachment there that's just reserving the volume I guess that would do | 16:15 |
| mriedem | so another microversion for key changes in the payload? | 16:15 |
| tommylikehu | I am not sure whether we need another micrioversion if the attachment APis haven't been truely used | 16:16 |
| ildikov | as we don't use any microversion at this point I would think to just use the latest that contains what we need? | 16:16 |
| mriedem | tommylikehu: you do | 16:16 |
| mriedem | ildikov: we use 3.27 or whatever | 16:16 |
| ildikov | mriedem: not at this point | 16:17 |
| mriedem | tommylikehu: it's unsafe to make assumptions about who is using an api once you release it | 16:17 |
| mriedem | ildikov: i mean, that's the minimum we can use, from ocata | 16:17 |
| tommylikehu | mriedem: ok | 16:17 |
| johnthetubaguy | mriedem: +1, sounds like a new microversion to me | 16:17 |
| mriedem | seems like a waste of a microversion for a cosmetic change like that | 16:18 |
| ildikov | mriedem: BTW, is that configurable? | 16:18 |
| mriedem | ildikov: is what configurable? | 16:18 |
| ildikov | what microversion Nova is using | 16:18 |
| mriedem | no | 16:18 |
| ildikov | ok | 16:18 |
| ildikov | that's what I thought | 16:18 |
| mriedem | remember, configuration changes which impact how apis work are the devil | 16:18 |
| johnthetubaguy | I think we should support just one micro-version for the new flow, its complicated if not | 16:19 |
| mriedem | johnthetubaguy: yes | 16:19 |
| mriedem | 3.27 or bust | 16:19 |
| mriedem | until we need something newer | 16:19 |
| ildikov | I know, I just got lost on the another microversion in the payload comment | 16:19 |
| mriedem | anyway, we're not there yet | 16:19 |
| mriedem | i need to remind myself at some point why we don't terminate connections during shelve offload | 16:19 |
| mriedem | because that seems important when you unshelve and the instance is built on another host | 16:20 |
| ildikov | so if we would end up do the cosmetics we just bump the minimum microversion in Nova | 16:20 |
| ildikov | I guess | 16:20 |
| mriedem | 1) i wouldn't do the cosmetics personally | 16:20 |
| smcginnis | I'm a little confused on that now though. Do we need a bump? | 16:20 |
| mriedem | 2) nova doesn't care about the cosmetics | 16:20 |
| johnthetubaguy | mriedem: oh, that sounds like a bug, we shouldn't detach the volume for sure, but I thought we should terminate... ? | 16:20 |
| mriedem | smcginnis: if you change a key in a request body, you need a microversion change | 16:20 |
| tommylikehu | request and response | 16:21 |
| johnthetubaguy | FWIW, we haven't bothered doing the cosmetic changes in Nova | 16:21 |
| mriedem | johnthetubaguy: i don't see where we terminate on shelve offload | 16:21 |
| smcginnis | OK, so Pike Nova should be able to work with Ocata Cinder with these changes, correct? | 16:21 |
| mriedem | smcginnis: yes | 16:21 |
| smcginnis | That was my concern. | 16:21 |
| johnthetubaguy | mriedem: your probably correct, just agreed it sounds wrong | 16:21 |
| mriedem | smcginnis: so you guys understand on the cinder side, | 16:21 |
| mriedem | if you change a request or response payload, | 16:21 |
| mriedem | it's a microversion bump right? | 16:21 |
| smcginnis | Yes, definitely. | 16:21 |
| johnthetubaguy | right, and the old microversion doesn't change | 16:21 |
| ildikov | personally I'm less enthusiastic about the change if it causes too much headache | 16:22 |
| mriedem | ok, because i'm a bit worried by the fact that cinder doesn't using jsonschema validation for requests | 16:22 |
| ildikov | I get the microversion part | 16:22 |
| mriedem | and i don't know if tempest is validating your microversioned responses with jsonschema yet | 16:22 |
| smcginnis | mriedem: I don't believe os. | 16:22 |
| smcginnis | *so | 16:22 |
| mriedem | really kind of need a tight jsonschema validation on request and response when you start supporting microversions | 16:23 |
| mriedem | because checks in python code aren't going to catc hstuff | 16:23 |
| smcginnis | mriedem: Right, seems like the only way to validate nothing slips through. | 16:23 |
| mriedem | https://review.openstack.org/#/c/454436/ | 16:24 |
| mriedem | you guys don't require a spec for api changes? | 16:24 |
| tommylikehu | ... | 16:25 |
| tommylikehu | releasenote is required | 16:25 |
| mriedem | release notes are fine, from a process standpoint we've found that we need a spec for api changes | 16:26 |
| smcginnis | Not always. Usually only need a spec if it's a big enough change that we want to make sure everyone understands what's being done. | 16:26 |
| mriedem | ok, that's fine, it's your world - fwiw i've -1ed that proposal since i don't see the value | 16:26 |
| tommylikehu | I am ok with that | 16:27 |
| ildikov | fair enough | 16:27 |
| *** openstack has joined #openstack-meeting-cp | 16:33 | |
| *** ChanServ sets mode: +o openstack | 16:33 | |
| stvnoyes | i'm at oracle and they're very interested in this for RAC. what is the expectation for how much of this is going to be completed for Pike? | 16:34 |
| ildikov | but as that will be a separate patch anyhow we can deal with that on the review, I would guess we don't want the volume connected to the instance as a principle | 16:35 |
| ildikov | stvnoyes: we would love to get the base functionality working | 16:35 |
| mriedem | stvnoyes: which thing specifically? | 16:35 |
| stvnoyes | and for RAC, we can probably get away with not worrying about migration/swap/bfv | 16:35 |
| mriedem | nova using the new attach/detach APIs? | 16:35 |
| ildikov | stvnoyes: we need to deal with a few things for multi-attach still | 16:35 |
| mriedem | or volume multiattach? | 16:35 |
| stvnoyes | twe'd need multi-attach working for at least the zfssa | 16:36 |
| *** scottda has joined #openstack-meeting-cp | 16:36 | |
| mriedem | i expect we'll get the nova using the new api flow with cinder in pike | 16:36 |
| mriedem | i don't know about multiattach support honestly | 16:36 |
| mriedem | depends on the progress made | 16:36 |
| ildikov | I don't think we will get to that if I would need to be brutally honest | 16:36 |
| ildikov | not with the current progress at least | 16:37 |
| jungleboyj | We have been working toward this for so long that at least having the API improvements is very encouraging. | 16:37 |
| ildikov | the more hands on the keyboard the better | 16:37 |
| ildikov | we could definitely use more atm | 16:37 |
| stvnoyes | if would you feel more confident if it vm's with ma disks could not be migrated, disks couldn't be swapped, bfv not supported. that might simplify things for pike for a first implementation | 16:38 |
| mriedem | yeah we'll have to see | 16:38 |
| mriedem | first step is using the new apis, | 16:39 |
| mriedem | because once we attach with the new style, we have to detach with the new style everywhere | 16:39 |
| stvnoyes | i am (slowly) coming up to speed on nova and we have someone else joining us, and hopefully another person in a few more weeks. | 16:39 |
| stvnoyes | so we should be able to help | 16:39 |
| mriedem | and once you support multiattach (not bfv) then you need to consider what that means for those other operations | 16:39 |
| ildikov | stvnoyes: +1, thanks! | 16:39 |
| mriedem | unless you just completely punt and fail saying it's not supported | 16:39 |
| mriedem | anyway, that's a ways down the road | 16:40 |
| stvnoyes | i was thinking that you would throw an api exception if you try to migrate an instance with a ma disk for example | 16:40 |
| mriedem | maybe, but we're nowhere near that point yet | 16:40 |
| mriedem | so let's not rathole on it :) | 16:40 |
| stvnoyes | kk understood | 16:40 |
| mriedem | we already did that about a year ago | 16:40 |
| mriedem | ildikov: ok done for the meeting? | 16:41 |
| ildikov | mriedem: I think we are | 16:41 |
| ildikov | stvnoyes: we can catch up after the meeting if you need more info or anytime if you have further questions | 16:42 |
| ildikov | stvnoyes: feel free to ping me | 16:42 |
| stvnoyes | ok thanks! | 16:42 |
| *** melwitt_ is now known as melwitt | 16:43 | |
| ildikov | is there anything else from anyone for today? | 16:43 |
| *** rarcea has quit IRC | 16:44 | |
| ildikov | ok I take it as a no | 16:44 |
| ildikov | as a summary I will split the detach patch to smaller ones and let's try to merge what we can from it | 16:44 |
| ildikov | see you all next week! | 16:44 |
| ildikov | thanks for today | 16:44 |
| ildikov | #endmeeting | 16:44 |
| *** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings" | 16:44 | |
| openstack | Meeting ended Thu Apr 13 16:44:51 2017 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 16:44 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-04-13-16.00.html | 16:44 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-04-13-16.00.txt | 16:44 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-04-13-16.00.log.html | 16:44 |
| jungleboyj | Thanks. | 16:44 |
| *** mriedem has left #openstack-meeting-cp | 16:45 | |
| *** lamt has quit IRC | 17:01 | |
| *** jgriffith has joined #openstack-meeting-cp | 17:02 | |
| *** 21WAAA2JF is now known as EmilienM | 17:05 | |
| *** EmilienM has joined #openstack-meeting-cp | 17:05 | |
| *** jaugustine has quit IRC | 17:06 | |
| *** knangia has joined #openstack-meeting-cp | 17:06 | |
| *** jgriffith has quit IRC | 17:13 | |
| *** jgriffith has joined #openstack-meeting-cp | 17:35 | |
| *** lamt has joined #openstack-meeting-cp | 18:05 | |
| *** jaugustine has joined #openstack-meeting-cp | 18:14 | |
| *** lamt has quit IRC | 18:27 | |
| *** lamt has joined #openstack-meeting-cp | 18:33 | |
| *** jaugustine has quit IRC | 18:37 | |
| *** ying_zuo has joined #openstack-meeting-cp | 18:53 | |
| *** gouthamr has quit IRC | 19:17 | |
| *** jaugustine has joined #openstack-meeting-cp | 19:33 | |
| lbragstad | robcresswell o/ | 20:05 |
| *** gouthamr has joined #openstack-meeting-cp | 20:10 | |
| *** diablo_rojo has quit IRC | 20:29 | |
| robcresswell | lbragstad: Aw crap, my turn to forget | 20:32 |
| lbragstad | robcresswell no worries :) | 20:33 |
| robcresswell | Sorry, didn't realise the time | 20:33 |
| lbragstad | robcresswell does look like many folks were here anyway - so maybe we just pick up next week? | 20:33 |
| lbragstad | robcresswell unless you have something specific to discuss? | 20:33 |
| robcresswell | lbragstad: Nothing specific. I think we should also stop having this meeting weekly IMO | 20:34 |
| lbragstad | yeah - i'd be fine with that so long as we encourage folks to call for it if we need the communication | 20:34 |
| lbragstad | does that sound fair? | 20:34 |
| robcresswell | lbragstad: I was just thinking monthly perhaps :) | 20:35 |
| robcresswell | So there is a sync point | 20:35 |
| lbragstad | robcresswell oh - even better | 20:36 |
| lbragstad | robcresswell i like that | 20:36 |
| robcresswell | lbragstad: I think it'll encourage people to attend the specific meetings more readily I hope. ATM its a bit dead. | 20:37 |
| *** ediardo has joined #openstack-meeting-cp | 20:37 | |
| robcresswell | I'll send out an email next week, its holidays in the UK till Tuesday | 20:37 |
| lbragstad | robcresswell true - that sounds good | 20:38 |
| lbragstad | robcresswell if you want me to take a stab at the email i can, since you're on holiday | 20:38 |
| robcresswell | lbragstad: We'll need to patch it too, but yeah if you don't mind, that'd be great | 20:54 |
| lbragstad | robcresswell done | 21:06 |
| lbragstad | robcresswell if we nail down a day of the month in the thread, I'll submit a patch for it | 21:06 |
| *** sdague has quit IRC | 21:11 | |
| *** diablo_rojo has joined #openstack-meeting-cp | 21:26 | |
| *** jaugustine has quit IRC | 21:31 | |
| *** diablo_rojo has quit IRC | 21:51 | |
| *** lamt has quit IRC | 21:55 | |
| *** lamt has joined #openstack-meeting-cp | 21:56 | |
| *** lamt has quit IRC | 22:00 | |
| *** gouthamr has quit IRC | 23:24 | |
| *** gouthamr has joined #openstack-meeting-cp | 23:35 | |
| *** diablo_rojo has joined #openstack-meeting-cp | 23:48 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!