| *** ducttape_ has joined #openstack-meeting-cp | 00:06 | |
| *** gouthamr has quit IRC | 00:24 | |
| *** gouthamr has joined #openstack-meeting-cp | 00:32 | |
| *** ducttape_ has quit IRC | 00:37 | |
| *** ducttape_ has joined #openstack-meeting-cp | 00:37 | |
| *** ducttape_ has quit IRC | 00:37 | |
| *** ducttape_ has joined #openstack-meeting-cp | 00:38 | |
| *** ducttape_ has quit IRC | 00:38 | |
| *** ducttape_ has joined #openstack-meeting-cp | 00:42 | |
| *** kbyrne has quit IRC | 01:37 | |
| *** kbyrne has joined #openstack-meeting-cp | 01:40 | |
| *** cebreidian has quit IRC | 02:21 | |
| *** cebreidian has joined #openstack-meeting-cp | 02:30 | |
| *** david-lyle has joined #openstack-meeting-cp | 03:15 | |
| *** david-lyle has quit IRC | 03:21 | |
| *** ducttape_ has quit IRC | 03:28 | |
| *** ducttape_ has joined #openstack-meeting-cp | 03:28 | |
| *** ducttape_ has quit IRC | 03:29 | |
| *** ducttape_ has joined #openstack-meeting-cp | 03:29 | |
| *** ducttape_ has quit IRC | 03:30 | |
| *** cebreidian has quit IRC | 03:42 | |
| *** ducttape_ has joined #openstack-meeting-cp | 04:10 | |
| *** ducttape_ has quit IRC | 04:11 | |
| *** ducttape_ has joined #openstack-meeting-cp | 04:12 | |
| *** ducttape_ has quit IRC | 04:12 | |
| *** ducttape_ has joined #openstack-meeting-cp | 04:12 | |
| *** ducttape_ has quit IRC | 04:13 | |
| *** ducttape_ has joined #openstack-meeting-cp | 04:13 | |
| *** ducttape_ has quit IRC | 04:14 | |
| *** cebreidian has joined #openstack-meeting-cp | 04:45 | |
| *** cebreidian has quit IRC | 05:00 | |
| *** gouthamr has quit IRC | 05:07 | |
| *** cebreidian has joined #openstack-meeting-cp | 05:09 | |
| *** cebreidian has quit IRC | 05:14 | |
| *** ducttape_ has joined #openstack-meeting-cp | 05:23 | |
| *** ducttape_ has quit IRC | 05:29 | |
| *** cebreidian has joined #openstack-meeting-cp | 05:31 | |
| *** ducttape_ has joined #openstack-meeting-cp | 06:54 | |
| *** ducttape_ has quit IRC | 06:59 | |
| *** rarcea has joined #openstack-meeting-cp | 07:12 | |
| *** brault has joined #openstack-meeting-cp | 08:07 | |
| *** ducttape_ has joined #openstack-meeting-cp | 08:25 | |
| *** MarkBaker has joined #openstack-meeting-cp | 08:28 | |
| *** ducttape_ has quit IRC | 08:30 | |
| *** MarkBaker has quit IRC | 09:31 | |
| *** ducttape_ has joined #openstack-meeting-cp | 09:55 | |
| *** ducttape_ has quit IRC | 10:00 | |
| *** MarkBaker has joined #openstack-meeting-cp | 10:24 | |
| *** ducttape_ has joined #openstack-meeting-cp | 11:22 | |
| *** ducttape_ has quit IRC | 11:41 | |
| *** ducttape_ has joined #openstack-meeting-cp | 11:41 | |
| *** ducttape_ has quit IRC | 11:42 | |
| *** ducttape_ has joined #openstack-meeting-cp | 11:42 | |
| *** ducttape_ has quit IRC | 11:42 | |
| *** ducttape_ has joined #openstack-meeting-cp | 11:43 | |
| *** ducttape_ has quit IRC | 11:43 | |
| *** ducttape_ has joined #openstack-meeting-cp | 11:44 | |
| *** ducttape_ has quit IRC | 11:44 | |
| *** ducttape_ has joined #openstack-meeting-cp | 12:12 | |
| *** sdague has joined #openstack-meeting-cp | 12:22 | |
| *** MarkBaker has quit IRC | 12:25 | |
| *** MarkBaker has joined #openstack-meeting-cp | 12:27 | |
| *** ducttape_ has quit IRC | 12:32 | |
| *** _sigmavirus24 is now known as sigmavirus | 13:02 | |
| *** sigmavirus has joined #openstack-meeting-cp | 13:02 | |
| *** MarkBaker has quit IRC | 13:02 | |
| *** MarkBaker has joined #openstack-meeting-cp | 13:04 | |
| *** ducttape_ has joined #openstack-meeting-cp | 13:05 | |
| *** xyang1 has joined #openstack-meeting-cp | 13:12 | |
| *** jaugustine has joined #openstack-meeting-cp | 13:22 | |
| *** ducttape_ has quit IRC | 13:39 | |
| *** gouthamr has joined #openstack-meeting-cp | 13:56 | |
| *** ducttape_ has joined #openstack-meeting-cp | 14:15 | |
| *** ducttape_ has quit IRC | 14:17 | |
| *** ducttape_ has joined #openstack-meeting-cp | 14:27 | |
| *** ducttape_ has quit IRC | 14:31 | |
| *** ducttape_ has joined #openstack-meeting-cp | 14:55 | |
| *** _ducttape_ has joined #openstack-meeting-cp | 14:58 | |
| *** MarkBaker has quit IRC | 15:00 | |
| *** ducttape_ has quit IRC | 15:01 | |
| *** _ducttape_ has quit IRC | 15:03 | |
| *** edtubill has joined #openstack-meeting-cp | 15:07 | |
| *** diablo_rojo has joined #openstack-meeting-cp | 15:11 | |
| *** knikolla has quit IRC | 15:20 | |
| *** knikolla has joined #openstack-meeting-cp | 15:20 | |
| *** rarcea has quit IRC | 15:23 | |
| *** MarkBaker has joined #openstack-meeting-cp | 15:29 | |
| *** ducttape_ has joined #openstack-meeting-cp | 15:31 | |
| *** sdague has quit IRC | 15:43 | |
| *** _ducttape_ has joined #openstack-meeting-cp | 15:57 | |
| *** diablo_rojo_phon has joined #openstack-meeting-cp | 15:58 | |
| *** ducttape_ has quit IRC | 16:00 | |
| *** gouthamr has quit IRC | 16:12 | |
| *** gouthamr has joined #openstack-meeting-cp | 16:12 | |
| *** jkomg has joined #openstack-meeting-cp | 16:31 | |
| *** david-lyle has joined #openstack-meeting-cp | 16:32 | |
| *** stvnoyes has joined #openstack-meeting-cp | 16:39 | |
| *** david-lyle_ has joined #openstack-meeting-cp | 16:57 | |
| *** _ducttape_ has quit IRC | 16:58 | |
| *** breitz has joined #openstack-meeting-cp | 16:59 | |
| *** ducttape_ has joined #openstack-meeting-cp | 17:01 | |
| ildikov | #startmeeting cinder-nova-api-changes | 17:01 |
|---|---|---|
| openstack | Meeting started Mon Jan 23 17:01:21 2017 UTC and is due to finish in 60 minutes. The chair is ildikov. Information about MeetBot at http://wiki.debian.org/MeetBot. | 17:01 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 17:01 |
| *** openstack changes topic to " (Meeting topic: cinder-nova-api-changes)" | 17:01 | |
| openstack | The meeting name has been set to 'cinder_nova_api_changes' | 17:01 |
| ildikov | scottda 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 | 17:01 |
| hemna | tough | 17:01 |
| hemna | heh | 17:01 |
| hemna | tough | 17:01 |
| ildikov | hemna: morning :) | 17:01 |
| hemna | fwiw, I'm on a work call now too. | 17:02 |
| ildikov | hemna: I know | 17:02 |
| ildikov | hemna: would this slot work for you on any other day than Monday? | 17:02 |
| scottda | hi. | 17:02 |
| scottda | doubled booked as well | 17:02 |
| hemna | I think so, it seems to change week to week for us | 17:02 |
| ildikov | hemna: I checked the calendar last week and it seemed we have at least one channel available for this slot whole week | 17:02 |
| * patrickeast is sorta here too | 17:03 | |
| diablo_rojo | Hello | 17:03 |
| * breitz breitz is here (sorta too - double booked) | 17:03 | |
| *** david-lyle_ has quit IRC | 17:03 | |
| ildikov | hemna: hmm, I cannot handle random collisions :) or is there any slot that would surely work? :) | 17:03 |
| stvnoyes | steve here... | 17:04 |
| hemna | I think this is fine for me for now | 17:04 |
| ildikov | ok then I will send out a mail with a Doodle poll | 17:04 |
| hemna | I'll just use both halves of my brain, then sleep the rest of the day. | 17:04 |
| ildikov | same slot on some other day of the week will hopefully work | 17:05 |
| ildikov | hemna: it would work for me :) | 17:05 |
| ildikov | breitz: hey, welcome to the club of double booked people :) | 17:06 |
| breitz | thanks! | 17:06 |
| ildikov | we don't have Matt today, he's on vacation | 17:06 |
| ildikov | I couldn't reach johnthetubaguy on the Nova channel earlier either | 17:06 |
| ildikov | so it seems like a Nova-less meeting for today | 17:06 |
| scottda | Good for him. ildikov , please make note of that NOva patch.... | 17:07 |
| ildikov | I updated the etherpad we have with the latest news | 17:07 |
| ildikov | #link https://etherpad.openstack.org/p/cinder-nova-api-changes | 17:07 |
| scottda | #link https://review.openstack.org/#/c/422078/ | 17:07 |
| scottda | That should be easy to land | 17:08 |
| scottda | And needed, regardless of cinder-nova api changes | 17:08 |
| ildikov | we need to make Nova to work with Cinder v3, fix for that is ^^ | 17:08 |
| ildikov | without scottda's patch Nova cannot handle the v3 endpoint thanks to some filtering | 17:08 |
| scottda | ildikov: What changes have you made to Nova to test new cinder apis? | 17:09 |
| scottda | ildikov: Or do you just use the cinderclient? | 17:09 |
| ildikov | scottda: I used it with the client, I hardcoded the version to 3.27 in Nova to use that one | 17:09 |
| scottda | ildikov: But what does nova do to make the new Cinder api calls? | 17:10 |
| ildikov | scottda: also used your change and set the catalog_info in the nova.conf to point to v3 | 17:10 |
| ildikov | scottda: it's hardcoded there as well for now | 17:10 |
| scottda | ok. Can you share that? | 17:10 |
| ildikov | scottda: check here: https://review.openstack.org/#/c/330285/13/nova/volume/cinder.py | 17:10 |
| ildikov | scottda: line 533 | 17:10 |
| scottda | ildikov: OK, so you are using jgriffith 's patches. Thanks. | 17:11 |
| ildikov | scottda: so we need to figure out how to negotiate with the client/server on avaliable versions and then I can point this patch to depend on yours or add the changes here | 17:12 |
| * johnthetubaguy wonders in late | 17:12 | |
| ildikov | scottda: yep, fixed a few things | 17:12 |
| ildikov | johnthetubaguy: hi :) | 17:12 |
| ildikov | johnthetubaguy: thanks for joining | 17:12 |
| ildikov | then announcements quickly | 17:12 |
| ildikov | #info we have the new attach/detach flow patch merged for Cinder | 17:12 |
| ildikov | #link https://review.openstack.org/#/c/414632/ | 17:13 |
| johnthetubaguy | ah, sweet | 17:13 |
| ildikov | johnthetubaguy: so this is how it looks now^^ | 17:13 |
| scottda | ildikov: Sure, I need to work on details with nova folks for https://review.openstack.org/420201 | 17:13 |
| ildikov | #info with local testing nova volume-attach works with the new Cinder API's | 17:13 |
| ildikov | #link https://review.openstack.org/#/c/330285/ | 17:13 |
| ildikov | I know the tests are failing that's not the first priority for the PoC now :) | 17:14 |
| ildikov | thanks for all the reviews on the Cinder patch from everyone! | 17:14 |
| scottda | ildikov: What are we thinking is left for Ocata? | 17:14 |
| ildikov | scottda: right, we need to figure out how to make Matt's preferred flow work | 17:15 |
| ildikov | scottda: I would say testing and fix whatever issues we find | 17:15 |
| johnthetubaguy | whats the chances of getting the functional tests passing on that POC? do we know if there are big issues in there? | 17:15 |
| ildikov | scottda: johnthetubaguy we also need ot look into the detach flow to work with Nova | 17:15 |
| johnthetubaguy | ah, the detach flow might be the issue for those tests then | 17:15 |
| ildikov | scottda: and also to figure out how to support multi-attach with the new flow as it does not care about that currently | 17:16 |
| scottda | #link https://review.openstack.org/420201 | 17:16 |
| scottda | johnthetubaguy: Please look at that ^^ | 17:16 |
| scottda | I'm wondering where to put a hardcoded version of cinder/cinderclient that Nova wishes to use | 17:16 |
| scottda | Then I can update that patch | 17:16 |
| johnthetubaguy | OK, so step one is move nova to Cinder v3 API I guess | 17:17 |
| ildikov | johnthetubaguy: the PoC fails with one minor issue I updated some function headers and it's not consistent now | 17:17 |
| ildikov | johnthetubaguy: the other thing that's failing is that I thought to keep the old tests on the old flow and there's one test I couldn't figure out how to mock the version properly... :S | 17:18 |
| ildikov | johnthetubaguy: if we want to keep the old tests until we support the old flow, I might help with that | 17:18 |
| jgriffith | It must be Monday | 17:18 |
| johnthetubaguy | scottda: I think just hard coding the version in a constant is fine | 17:19 |
| ildikov | jgriffith: your instinct work well :) | 17:19 |
| ildikov | *instincts | 17:19 |
| scottda | johnthetubaguy: OK, I'll put that in nova/volume/cinder.py | 17:19 |
| johnthetubaguy | scottda: yeah, I think thats fine. | 17:19 |
| ildikov | johnthetubaguy: hardcoding the one that Nova would wish to use or what Cinder has to offer or both? :) | 17:20 |
| jgriffith | ildikov scottda now that the Cinder side is fairly stable I can rework the Nova patch with ildikov to get the POC of it working again | 17:20 |
| ildikov | jgriffith: I made nova volume-attach work yesterday :) | 17:20 |
| jgriffith | I am not familiar with the reference about the "preferred flow" from Matt R though? | 17:20 |
| johnthetubaguy | so, microversions mean you have to keep all old versions, so what matters is what versions work with Nova | 17:21 |
| ildikov | jgriffith: or well, no failures in the logs and I think I saw some iSCSI stuff there too happening :) | 17:21 |
| scottda | ildikov: Hard code the cinder version that Nova wants to use... | 17:21 |
| johnthetubaguy | for now, I guess we just request the base version? | 17:21 |
| scottda | ildikov: Then get the cinderclient MAX_VERSION and cinder server version dynamically | 17:21 |
| scottda | Then get the MIN of those 3 | 17:21 |
| ildikov | scottda: in my head it looks like we have an if where we want to use stuff that needs a microversion and if we have that then go and if we don't then use the old stuff | 17:21 |
| ildikov | scottda: we need that anyhow | 17:21 |
| jgriffith | oh, ya'll are talking about the micro-version madness; I can't help there | 17:21 |
| ildikov | jgriffith: we just can't get out of that loop... | 17:22 |
| jgriffith | well, I can help | 17:22 |
| johnthetubaguy | so from the nova side it should be fairly simple | 17:22 |
| ildikov | jgriffith: I blame Monday... | 17:22 |
| johnthetubaguy | there is a requirement to use the new flow, else we do the old flow | 17:22 |
| jgriffith | Look, I'll just hack a check like a had before in nova/volume/cinder.py | 17:22 |
| jgriffith | Call the client, check something like "v2_supported" | 17:22 |
| jgriffith | if it raises or the API call is not found, then it doesn't support it | 17:23 |
| jgriffith | else use the new code | 17:23 |
| johnthetubaguy | so if new-flow-micoversion-7 supported, use_new_flow = True or something? | 17:23 |
| jgriffith | johnthetubaguy yes | 17:23 |
| jgriffith | johnthetubaguy it's not elegant but it works | 17:23 |
| johnthetubaguy | surely v3 support the old flow though right? | 17:23 |
| scottda | johnthetubaguy: yes | 17:23 |
| ildikov | jgriffith: you have that check in the PoC decide that it's legacy or new flow | 17:23 |
| jgriffith | johnthetubaguy yes for sure | 17:23 |
| scottda | old flow never goes away | 17:23 |
| ildikov | jgriffith: that works pretty well | 17:23 |
| johnthetubaguy | right, that all makes sense | 17:24 |
| jgriffith | scottda umm, not true! | 17:24 |
| ildikov | we just need the info from the client that what version it can offer | 17:24 |
| johnthetubaguy | so scottda I think there are two changes | 17:24 |
| jgriffith | scottda once we get this in Nova old flow gets deprecated and yes it most certainly does go away | 17:24 |
| scottda | jgriffith: Well, there are those who would yell if we got rid of old apis | 17:24 |
| scottda | jgriffith: OK, well that's not my battle to fight. | 17:24 |
| ildikov | scottda: with proper deprecation it should be able to go away IMHO | 17:24 |
| jgriffith | scottda I fundamentally disagree | 17:24 |
| johnthetubaguy | jgriffith: but thats how microversions works, all your old API live for ever..., well until you raise the min_support API version I guess | 17:24 |
| johnthetubaguy | anyways, thats a different debate | 17:24 |
| jgriffith | johnthetubaguy and that's one of the flaws in microversions | 17:24 |
| jgriffith | but ok | 17:24 |
| jgriffith | I'm out of it | 17:25 |
| johnthetubaguy | jgriffith: we call it a feature, but yeah, depends what camp you are in | 17:25 |
| jgriffith | johnthetubaguy :) | 17:25 |
| johnthetubaguy | scottda: so maybe you have two changes there, support v3, and decide when its safe to use the new flow? | 17:25 |
| scottda | johnthetubaguy: yes | 17:26 |
| scottda | they are 2 patch sets | 17:26 |
| johnthetubaguy | cools | 17:26 |
| jgriffith | johnthetubaguy ildikov scottda I said this before, and I'll say it again; if there's going to be a requirement that no flow ever gets taken away then revert the change that I implemented and we go back to reworking the existing API's | 17:26 |
| jgriffith | becuase keeping both in there is madness | 17:26 |
| scottda | jgriffith: I am happy to deprecate stuff. | 17:26 |
| jgriffith | and there's no way you're going to be able to continue to support, test and maintain both | 17:26 |
| scottda | jgriffith: I won't argue against it. | 17:26 |
| ildikov | johnthetubaguy: we have checks like this in line 2086 here: https://review.openstack.org/#/c/330285/13/nova/compute/api.py | 17:27 |
| ildikov | johnthetubaguy: in the PoC | 17:27 |
| ildikov | jgriffith: I'm a deprecation fan for sure! | 17:27 |
| johnthetubaguy | ildikov: yeah, makes sense | 17:27 |
| ildikov | johnthetubaguy: scottda so we only need to replace that hardcoded number with the version the client gives back and if it's >= 3.27 than we're good... | 17:28 |
| jgriffith | ildikov +1 | 17:28 |
| ildikov | jgriffith: maintaining both would be a real nightmare and I don't think we want that, I surely don't | 17:29 |
| scottda | ildikov: Sure. | 17:29 |
| ildikov | scottda: so if you can make the negotiation work in your patch for a version that would be great and we can use that as a base for the PoC | 17:29 |
| ildikov | jgriffith: so just get rid of the old flow and blame me, I can take it :) | 17:30 |
| scottda | ildikov: Yes. And we need to decide how to get that info. A global? a getter? | 17:30 |
| scottda | I can code it up any way we want | 17:31 |
| johnthetubaguy | SUPPORTS_NEW_FLOW=None and if its None, you set that before you return the client? | 17:31 |
| ildikov | scottda: honestly I personally don't have a strong preference on that at this point | 17:31 |
| scottda | Sure. I'll set a Global for now. I'll update that patch set today. | 17:31 |
| johnthetubaguy | my preference is for simple (but also cached) | 17:32 |
| ildikov | johnthetubaguy: let's stick with a version number, people will get confused if we have a variable like that IMHO | 17:32 |
| johnthetubaguy | but then you need to spread the version number everywhere? | 17:32 |
| johnthetubaguy | its hard to know without seeing it all in gerrit, if I am honest | 17:32 |
| johnthetubaguy | its probably there, I just haven't read it | 17:33 |
| ildikov | you need spread either this info or the version number everywhere anyway :) | 17:33 |
| jgriffith | johnthetubaguy yeah, I like the idea of just checking the version number on init and storing as part of the cinder object in Nova | 17:33 |
| ildikov | and I think the version number is crystal clear :) | 17:33 |
| jgriffith | ez/pz and cached in memory | 17:33 |
| scottda | ildikov: Version number spread throughout nova won't be clear to everyone... | 17:33 |
| johnthetubaguy | yeah, the cached in memory bit is I think worth it | 17:33 |
| johnthetubaguy | beyond that, just keep it simple | 17:33 |
| johnthetubaguy | so going back to this patch: | 17:34 |
| ildikov | scottda: but if we will have something new in the future that's connected to a particular higher version number we will have a zillion boolean variables for all if we follow this model | 17:34 |
| johnthetubaguy | https://review.openstack.org/#/c/420201 | 17:34 |
| jgriffith | ildikov oh dear, please no :( | 17:35 |
| johnthetubaguy | ildikov: I think the idea is we evenutally delete the ifs, and get back to one path though | 17:35 |
| jgriffith | ildikov I'd hope the interface won't change again so there wouldn't be a risk of that | 17:35 |
| jgriffith | johnthetubaguy +1 | 17:35 |
| ildikov | johnthetubaguy: sure, if this boolean suggestion brings us closer to that I'm all in! | 17:35 |
| johnthetubaguy | looking at https://review.openstack.org/#/c/420201 | 17:35 |
| johnthetubaguy | there is a chunk in there about the microversion | 17:35 |
| ildikov | johnthetubaguy, jgriffith: I'm always paranoid on Mondays, please forgive me :) | 17:36 |
| johnthetubaguy | its tempting to leave that to a later patches | 17:36 |
| jgriffith | LOL | 17:36 |
| jgriffith | I"m always paranoid in general :) | 17:36 |
| johnthetubaguy | ildikov: a good dose of paranoia can be helpful | 17:36 |
| scottda | johnthetubaguy: Well, we can get the cinder api version from the server before we instantiate the cinderclient | 17:36 |
| johnthetubaguy | what john said | 17:36 |
| ildikov | jgriffith: maybe not a bad strategy around here :) | 17:36 |
| johnthetubaguy | scottda: ah, I see what you are wanting now | 17:37 |
| johnthetubaguy | scottda: maybe just call it MAX_SUPPORT_CINDER_API_VERSION? | 17:37 |
| scottda | johnthetubaguy: That patch needs an update...to use Nova's idea of which max cinder version it can use | 17:37 |
| johnthetubaguy | so I think there are two bits here | 17:38 |
| scottda | johnthetubaguy: sure, I'll call it that. And set it before nova instantiates cinderclient | 17:38 |
| johnthetubaguy | whats the max version | 17:38 |
| johnthetubaguy | is that high enough for us to use the new flow? | 17:38 |
| johnthetubaguy | most calls can just default to not specifying a micro version and using the base version, I guess? | 17:38 |
| scottda | johnthetubaguy: yes, but 3 bits: max cinder server version, max cinderclient version, nova's desired cinder version to use new stuff | 17:39 |
| johnthetubaguy | cinderclient version? | 17:39 |
| ildikov | johnthetubaguy: base version? | 17:39 |
| johnthetubaguy | ildikov: the version you get without requesting a micro version, its possible you don't have that | 17:39 |
| scottda | johnthetubaguy: Maybe cinder server has new APIs, but you haven't updated your client | 17:39 |
| johnthetubaguy | scottda: yeah, gotcha, just being slow | 17:40 |
| ildikov | johnthetubaguy: ah, ok | 17:40 |
| scottda | johnthetubaguy: Trivial case, but possible. So needs checking. | 17:40 |
| johnthetubaguy | does the client tell us what is the max version it and the server speaks? | 17:40 |
| johnthetubaguy | that sounds like the one we want to record | 17:40 |
| scottda | It could... | 17:40 |
| ildikov | scottda: I guess the client can return the lower from those two versions or smth like? | 17:40 |
| scottda | Right now, client has MAX_VERSION | 17:40 |
| ildikov | scottda: as if the client does not have the code snippet Nova tries to call than it doesn't matter what the server has | 17:41 |
| scottda | And I have patch in flight for static method that takes endpoint url and gets version info from tthe server.. | 17:41 |
| scottda | Client could get MIN of those 2 , and return in a method. Or nova could do the work. | 17:41 |
| scottda | Either way | 17:42 |
| johnthetubaguy | oh wait, we pass the microversion into the client constructor? | 17:42 |
| ildikov | I don't think Nova should know that | 17:42 |
| scottda | I'll put up a cinderclient patch to return the MIN of (cinderclient_MAX_VERSION, cinder_server_VERSION) | 17:42 |
| scottda | johnthetubaguy: Yes | 17:42 |
| ildikov | johnthetubaguy: currently we do, I would be happier to just get it TBH, but I can easily miss smth here | 17:43 |
| scottda | johnthetubaguy: We did this prior to microversions | 17:43 |
| johnthetubaguy | scottda: I was kinda expecting to just pass in "3" | 17:43 |
| scottda | It just used to me major version (1|2|3) and now can take minor version (microversion) | 17:43 |
| * ildikov is not a microversion specialist on any level in any parallel universe... | 17:43 | |
| scottda | microversion == minor version | 17:43 |
| scottda | ildikov: That's all you need to know | 17:43 |
| scottda | APIs are versioned. | 17:43 |
| johnthetubaguy | I can't remember what we do for nova right now | 17:43 |
| scottda | So you know what you need, and are getting | 17:44 |
| johnthetubaguy | novaclient I mean | 17:44 |
| johnthetubaguy | so for this first patch, should we just pass 3.0? | 17:45 |
| johnthetubaguy | if version == "3" | 17:46 |
| scottda | no | 17:46 |
| scottda | johnthetubaguy: Because this is a much higher version | 17:46 |
| scottda | johnthetubaguy: cinder version 3.0 does not have the new APIs for attach/detach | 17:46 |
| johnthetubaguy | so first version is just calling the old flow, on v2.0 | 17:46 |
| johnthetubaguy | I mean v3.0 | 17:47 |
| scottda | yes | 17:47 |
| johnthetubaguy | oh..... | 17:47 |
| scottda | v3.0 == v2 | 17:47 |
| johnthetubaguy | yeah | 17:47 |
| johnthetubaguy | so thats what we want for the first patch right? | 17:47 |
| johnthetubaguy | just use the old APIs on the new v3 end point? | 17:47 |
| scottda | "that" is what? | 17:47 |
| scottda | ahh..right | 17:47 |
| johnthetubaguy | I am thinking pure in isolation enable the use of v3 | 17:48 |
| johnthetubaguy | then second step is determine if can use new flow (v3.42 say) vs old flow (v3.0) | 17:48 |
| scottda | johnthetubaguy: Yes, as of today, we cannot use v3 | 17:49 |
| johnthetubaguy | oh dear, now I get you | 17:49 |
| scottda | That's why we need https://review.openstack.org/422078 | 17:49 |
| scottda | That is a crappy bug because of how cinder has 3 things in the keystone catalog... | 17:49 |
| scottda | "volume", "volume2", "volume3" | 17:49 |
| * scottda would like to change that | 17:50 | |
| ildikov | scottda: oh, that's kinda terrible | 17:50 |
| johnthetubaguy | volume4 that just points to the route? | 17:50 |
| scottda | we've no volume4 ATM | 17:50 |
| scottda | (and hopefully never will) | 17:50 |
| johnthetubaguy | yeah, volume4 points to v2 and v3's route | 17:50 |
| johnthetubaguy | or something | 17:50 |
| ildikov | scottda: although after fixing the context list the 'catalog_info' still needs to be set | 17:50 |
| scottda | Well, how we fix it, I'm not sure... | 17:50 |
| ildikov | scottda: that's what we need the devstack config for we talked about a few weeks ago | 17:51 |
| scottda | I'm a ML post and plans at PTG to talk about the service catalog. Please attend. | 17:51 |
| scottda | ildikov: Sorry, which config was that? | 17:51 |
| scottda | ildikov: you mean to use v3 endpoint? | 17:51 |
| scottda | [cinder] catalog_info = volumev3:cinderv3:publicURL | 17:52 |
| ildikov | scottda: yeah | 17:52 |
| ildikov | Matt said that some post Devstack config can fix that so we can test v3 in the placement_api job | 17:52 |
| scottda | yup, that and my bug fix | 17:52 |
| ildikov | yep | 17:52 |
| johnthetubaguy | I think I am better understanding the earlier questions around the version number now, which is good | 17:53 |
| johnthetubaguy | we totally need to sit down and talk service catalog standardisation again, I remember sdague started some good conversations around that | 17:54 |
| ildikov | johnthetubaguy: cool, then this meeting was useful :) | 17:54 |
| johnthetubaguy | it was for me | 17:54 |
| ildikov | we have 5 minutes left | 17:55 |
| ildikov | I guess we're more or less on track regarding what we want to do with the versions | 17:55 |
| ildikov | I will send out a Doodle poll to the list trying to find another slot so we don't have that many collisions as we have now | 17:56 |
| *** david-lyle_ has joined #openstack-meeting-cp | 17:56 | |
| *** david-lyle_ has quit IRC | 17:56 | |
| ildikov | starting from next week I hope we can sort out what we need to make the PTG discussions good and productive | 17:56 |
| *** david-lyle_ has joined #openstack-meeting-cp | 17:57 | |
| scottda | ildikov: Yeah. I've a placeholder in cinder etherpad... | 17:57 |
| ildikov | also we need to figure out what we need on the Cinder side for the multi-attach bits | 17:57 |
| scottda | For meeting with Nova | 17:57 |
| ildikov | jgriffith: ^^ | 17:57 |
| scottda | We need to perhaps schedule a time for that. | 17:57 |
| ildikov | scottda: I think Matt added a similar placeholder to the Nova etherpad too, so we should be covered from this perspective | 17:57 |
| ildikov | scottda: yeah, I'll try to ping smcginnis and mriedem about timing this when both are available | 17:58 |
| ildikov | ok, anything else for today? :) | 17:59 |
| ildikov | all right, thanks everyone! | 18:00 |
| ildikov | see you all next week at this or at another slot :) | 18:00 |
| ildikov | #endmeeting | 18:00 |
| *** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings" | 18:01 | |
| openstack | Meeting ended Mon Jan 23 18:00:59 2017 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 18:01 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-01-23-17.01.html | 18:01 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-01-23-17.01.txt | 18:01 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-01-23-17.01.log.html | 18:01 |
| *** david-lyle has quit IRC | 18:05 | |
| *** david-lyle_ is now known as david-lyle | 18:05 | |
| *** breitz has left #openstack-meeting-cp | 18:27 | |
| *** diablo_rojo_phon has quit IRC | 19:20 | |
| *** gouthamr has quit IRC | 19:20 | |
| *** sdague has joined #openstack-meeting-cp | 19:22 | |
| *** knikolla has quit IRC | 19:23 | |
| *** knikolla has joined #openstack-meeting-cp | 19:23 | |
| *** knikolla has quit IRC | 19:23 | |
| *** knikolla has joined #openstack-meeting-cp | 19:24 | |
| *** knikolla has quit IRC | 19:24 | |
| *** knikolla has joined #openstack-meeting-cp | 19:25 | |
| *** knikolla has quit IRC | 19:25 | |
| *** knikolla has joined #openstack-meeting-cp | 19:25 | |
| *** knikolla has quit IRC | 19:25 | |
| *** sdague has quit IRC | 20:00 | |
| *** _ducttape_ has joined #openstack-meeting-cp | 20:30 | |
| *** stvnoyes has quit IRC | 20:33 | |
| *** ducttape_ has quit IRC | 20:33 | |
| *** stvnoyes has joined #openstack-meeting-cp | 20:33 | |
| *** jkomg has quit IRC | 20:40 | |
| *** jkomg has joined #openstack-meeting-cp | 20:40 | |
| *** _ducttape_ has quit IRC | 21:19 | |
| *** ducttape_ has joined #openstack-meeting-cp | 21:20 | |
| *** ducttape_ has quit IRC | 21:20 | |
| *** ducttape_ has joined #openstack-meeting-cp | 21:20 | |
| *** markvoelker has quit IRC | 21:47 | |
| *** ducttape_ has quit IRC | 21:56 | |
| *** harlowja has quit IRC | 22:03 | |
| *** jaugustine has quit IRC | 22:12 | |
| *** jkomg has quit IRC | 22:13 | |
| *** jkomg has joined #openstack-meeting-cp | 22:22 | |
| *** jkomg has quit IRC | 22:26 | |
| *** jkomg has joined #openstack-meeting-cp | 22:35 | |
| *** edtubill has quit IRC | 22:35 | |
| *** jkomg has quit IRC | 22:39 | |
| *** diablo_rojo has quit IRC | 22:39 | |
| *** xyang1 has quit IRC | 23:25 | |
| *** harlowja has joined #openstack-meeting-cp | 23:34 | |
| *** harlowja has quit IRC | 23:35 | |
| *** harlowja has joined #openstack-meeting-cp | 23:48 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!