*** david-lyle has joined #openstack-meeting-cp | 00:08 | |
*** david-lyle has quit IRC | 00:15 | |
*** lamt has quit IRC | 00:18 | |
*** david-lyle has joined #openstack-meeting-cp | 00:54 | |
*** ducttape_ has quit IRC | 00:54 | |
*** david-lyle has quit IRC | 00:54 | |
*** ducttape_ has joined #openstack-meeting-cp | 01:13 | |
*** ducttape_ has quit IRC | 01:35 | |
*** ducttape_ has joined #openstack-meeting-cp | 01:38 | |
*** reed has joined #openstack-meeting-cp | 01:46 | |
*** ducttape_ has quit IRC | 01:49 | |
*** david-lyle has joined #openstack-meeting-cp | 01:52 | |
*** diablo_rojo has quit IRC | 02:22 | |
*** markvoelker has quit IRC | 02:23 | |
*** knangia has quit IRC | 02:31 | |
*** rderose has quit IRC | 02:38 | |
*** ducttape_ has joined #openstack-meeting-cp | 02:50 | |
*** amrith has quit IRC | 02:54 | |
*** ducttape_ has quit IRC | 02:55 | |
*** amrith has joined #openstack-meeting-cp | 02:55 | |
*** benj_ has quit IRC | 03:12 | |
*** benj_ has joined #openstack-meeting-cp | 03:14 | |
*** ducttape_ has joined #openstack-meeting-cp | 03:18 | |
*** ducttape_ has quit IRC | 03:19 | |
*** ducttape_ has joined #openstack-meeting-cp | 03:20 | |
*** ducttape_ has quit IRC | 03:20 | |
*** ducttape_ has joined #openstack-meeting-cp | 03:20 | |
*** markvoelker has joined #openstack-meeting-cp | 03:24 | |
*** diablo_rojo has joined #openstack-meeting-cp | 03:28 | |
*** markvoelker has quit IRC | 03:29 | |
*** diablo_rojo has quit IRC | 04:05 | |
*** KeithMnemonic has quit IRC | 04:14 | |
*** markvoelker has joined #openstack-meeting-cp | 04:25 | |
*** ducttape_ has quit IRC | 04:28 | |
*** markvoelker has quit IRC | 04:29 | |
*** gouthamr has quit IRC | 04:46 | |
*** markvoelker has joined #openstack-meeting-cp | 05:25 | |
*** ducttape_ has joined #openstack-meeting-cp | 05:28 | |
*** markvoelker has quit IRC | 05:30 | |
*** ducttape_ has quit IRC | 05:32 | |
*** ducttape_ has joined #openstack-meeting-cp | 06:29 | |
*** ducttape_ has quit IRC | 06:33 | |
*** ShaneRowe has joined #openstack-meeting-cp | 06:52 | |
*** ShaneRowe has quit IRC | 06:59 | |
*** ducttape_ has joined #openstack-meeting-cp | 07:30 | |
*** ducttape_ has quit IRC | 07:35 | |
*** cartik has joined #openstack-meeting-cp | 08:00 | |
*** ducttape_ has joined #openstack-meeting-cp | 08:30 | |
*** ducttape_ has quit IRC | 08:35 | |
*** markvoelker has joined #openstack-meeting-cp | 08:46 | |
*** cartik has quit IRC | 08:55 | |
*** ducttape_ has joined #openstack-meeting-cp | 09:31 | |
*** ducttape_ has quit IRC | 09:36 | |
*** MarkBaker has joined #openstack-meeting-cp | 11:31 | |
*** markvoelker has quit IRC | 11:33 | |
*** sdague has joined #openstack-meeting-cp | 11:53 | |
*** MarkBaker has quit IRC | 12:09 | |
*** sdague has quit IRC | 12:36 | |
*** ducttape_ has joined #openstack-meeting-cp | 13:41 | |
*** lamt has joined #openstack-meeting-cp | 13:58 | |
*** gouthamr has joined #openstack-meeting-cp | 14:08 | |
*** MarkBaker has joined #openstack-meeting-cp | 14:17 | |
*** ducttape_ has quit IRC | 14:19 | |
*** lamt has quit IRC | 14:27 | |
*** MarkBaker has quit IRC | 14:29 | |
*** lamt has joined #openstack-meeting-cp | 14:31 | |
*** MarkBaker has joined #openstack-meeting-cp | 14:42 | |
*** lamt has quit IRC | 14:53 | |
*** ducttape_ has joined #openstack-meeting-cp | 15:05 | |
*** lamt has joined #openstack-meeting-cp | 15:09 | |
*** ducttape_ has quit IRC | 15:51 | |
*** sdague has joined #openstack-meeting-cp | 15:53 | |
*** KeithMnemonic has joined #openstack-meeting-cp | 15:58 | |
*** raj_singh_ has joined #openstack-meeting-cp | 16:00 | |
*** ducttape_ has joined #openstack-meeting-cp | 16:15 | |
*** ducttape_ has quit IRC | 16:37 | |
*** Rockyg has joined #openstack-meeting-cp | 16:38 | |
*** knangia has joined #openstack-meeting-cp | 16:39 | |
*** Rockyg has quit IRC | 16:45 | |
*** mriedem has joined #openstack-meeting-cp | 16:56 | |
ildikov | #startmeeting cinder-nova-api-changes | 17:00 |
---|---|---|
openstack | Meeting started Thu Mar 9 17:00:08 2017 UTC and is due to finish in 60 minutes. The chair is ildikov. Information about MeetBot at http://wiki.debian.org/MeetBot. | 17:00 |
openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 17:00 |
*** openstack changes topic to " (Meeting topic: cinder-nova-api-changes)" | 17:00 | |
openstack | The meeting name has been set to 'cinder_nova_api_changes' | 17:00 |
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 | 17:00 |
* smcginnis can only lurk for a bit | 17:00 | |
lyarwood | \o | 17:00 |
mriedem | o/ | 17:00 |
jgriffith | 0/ | 17:00 |
breitz | o/ | 17:00 |
ildikov | smcginnis: noted, tnx | 17:00 |
ildikov | hi all :) | 17:01 |
jungleboyj | o/ | 17:01 |
johnthetubaguy | o/ | 17:01 |
ildikov | let's start with the smaller action items before jumping into a philosophical topic | 17:01 |
ildikov | mriedem: johnthetubaguy: have you checked the switch to Cinder v3 patch? | 17:02 |
johnthetubaguy | how did last weeks work items go, I guess? | 17:02 |
johnthetubaguy | I haven't had chance | 17:02 |
mriedem | i haven't seen the latest yet no | 17:03 |
ildikov | johnthetubaguy: yep, basically switch to Cinder v3 and the BDM work | 17:03 |
mriedem | busy with other fires | 17:03 |
jgriffith | https://review.openstack.org/#/c/438744/ | 17:03 |
mriedem | i reviewed the bdm data model patches of lyarwood's last week, | 17:03 |
jgriffith | adn https://review.openstack.org/#/c/438750/ | 17:03 |
mriedem | and they still have some outstanding issues i think | 17:03 |
johnthetubaguy | #link https://etherpad.openstack.org/p/pike-nova-priorities-tracking | 17:03 |
jgriffith | oh... ok; I guess I'll abandon mine | 17:03 |
ildikov | mriedem: johnthetubaguy: https://review.openstack.org/#/c/442004/ | 17:03 |
johnthetubaguy | did we get everything in that etherpad^ | 17:03 |
jgriffith | lyarwood link to your patch (refresh my memory)? | 17:03 |
mriedem | ildikov: wrong patch | 17:03 |
lyarwood | https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/cinder-new-attach-apis | 17:04 |
jgriffith | lyarwood oh... my work is done here :) | 17:04 |
lyarwood | hmm the uuid changes are dropped from that bp | 17:04 |
jgriffith | thanks! | 17:04 |
ildikov | mriedem: johnthetubaguy: https://review.openstack.org/#/c/420201/ | 17:04 |
lyarwood | mriedem: I respun the db and object changes this morning but I think we still have some more work to do | 17:05 |
ildikov | mriedem: tnx for pointing out | 17:05 |
johnthetubaguy | OK I did wonder, thats the one | 17:05 |
mriedem | lyarwood: yeah you updated the attachment_id patch but not the bottom uuid patch | 17:05 |
johnthetubaguy | so for v3, did we say we would deprecate v2 support, or is that coming later? | 17:06 |
smcginnis | SO why do we have multiple patches overlapping? | 17:06 |
jgriffith | so I went ahead and abandoned my versions, I'm happy to hand this off :) | 17:07 |
jgriffith | I just don't want to continue duplicating efforts | 17:07 |
mriedem | smcginnis: the attachment_id duplicates were just from ptg | 17:07 |
mriedem | people didn't know | 17:07 |
lyarwood | pretty much | 17:07 |
jgriffith | mriedem I think there's some other overlap in there, but lyarwood did point out that he had started them and was *changing* a few things | 17:07 |
mriedem | johnthetubaguy: i'd like to say that we're going to deprecate nova's support for cinder v2 in pike | 17:07 |
ildikov | lyarwood: did you fix the comments on the BDM patches? is it ready for review? | 17:08 |
mriedem | johnthetubaguy: but in a separate patch from the one that adds the support for cinder v3 and makes it the default | 17:08 |
*** markvoelker has joined #openstack-meeting-cp | 17:08 | |
jgriffith | mriedem are you (or somebody) working on that? | 17:08 |
lyarwood | ildikov: the attachment_id changes are ready for more reviews yes, I missed mriedem's comments in the original uuid changes | 17:08 |
jungleboyj | mriedem: That is a good thing I think. | 17:08 |
ildikov | mriedem: what would that deprecation mean besides the reno? | 17:08 |
jgriffith | mriedem the "make v3 default in Nova thing" | 17:09 |
mriedem | jgriffith: on which part? | 17:09 |
jgriffith | mriedem "make v3 default in Nova" | 17:09 |
mriedem | jgriffith: that's ildikov/scottda's patch above | 17:09 |
jgriffith | mriedem thanks | 17:09 |
mriedem | https://review.openstack.org/#/c/420201/ | 17:09 |
jungleboyj | Cool. | 17:09 |
johnthetubaguy | mriedem: sounds good | 17:09 |
ildikov | jgriffith: yeap, I uploaded all the 20 lines of change, 18 of which is the reno :) | 17:09 |
jgriffith | ildikov :) | 17:10 |
mriedem | ildikov: it means that if nova is configured to use cinder v2, we log a warning | 17:10 |
mriedem | plus the reno | 17:10 |
ildikov | jgriffith: I need to figure out the micorversion stuff now | 17:10 |
mriedem | just like how we used to log a warning if nova was configured to use cinder v1 | 17:10 |
ildikov | mriedem: ok cool, I'll upload a follow up patch for that | 17:10 |
jgriffith | Oh, I see.. there's the WIP for that (as per mriedem 's comment) | 17:10 |
ildikov | mriedem: got it, I was just wondering what else I might forget | 17:10 |
johnthetubaguy | where is the microversion patch? | 17:11 |
jgriffith | https://review.openstack.org/#/c/385682/ | 17:11 |
johnthetubaguy | but thats not doing what we agreed we would do previously (yet) | 17:11 |
ildikov | johnthetubaguy: it was a previous patch set of the switch one | 17:11 |
johnthetubaguy | do we have that patch up? | 17:11 |
ildikov | johnthetubaguy: the latest non abandoned | 17:11 |
*** diablo_rojo has joined #openstack-meeting-cp | 17:11 | |
ildikov | johnthetubaguy: I need to get the microversion part up | 17:12 |
johnthetubaguy | OK, thats cool, thats still a TODO | 17:12 |
ildikov | johnthetubaguy: we need a cinderclient release for that one | 17:12 |
johnthetubaguy | right | 17:12 |
* ildikov looks at smcginnis on an estimate when that might happen :) | 17:12 | |
johnthetubaguy | are we looking close to getting the client released? | 17:12 |
* jungleboyj was just pinging smcginnis about that. | 17:13 | |
ildikov | lyarwood: cool, sounds good thanks! | 17:13 |
jungleboyj | He was waiting for https://review.openstack.org/#/c/385629/ to merge. | 17:13 |
jungleboyj | It isn't making much progress with scottda being pulled away. | 17:13 |
ildikov | lyarwood: is there anything in that chain that needs further discussions/work? | 17:13 |
lyarwood | ildikov: db/object wise no, just more reviews once I've addressed the remaining comments from mriedem | 17:13 |
johnthetubaguy | don't we need this one: https://review.openstack.org/#/c/385641/2 | 17:14 |
johnthetubaguy | ? | 17:14 |
*** raj_singh_ has quit IRC | 17:14 | |
mriedem | johnthetubaguy: yes, that's what they are talking about | 17:14 |
jungleboyj | Yes, but smcginnis Also wanted the other one in before doing a new client release. | 17:15 |
mriedem | unless that's old | 17:15 |
mriedem | jungleboyj: has anyone noticed the state of https://review.openstack.org/#/c/385641/ though? | 17:15 |
johnthetubaguy | jungleboyj: but I think we need this one in the release: https://review.openstack.org/#/c/385641/2 | 17:15 |
ildikov | lyarwood: cool; I guess you're working on getting those addressed and that will be up shortly, right? | 17:15 |
johnthetubaguy | #link https://review.openstack.org/#/c/385641 | 17:15 |
lyarwood | ildikov: yup I'm working on it now | 17:15 |
mriedem | heh, the cinderclient target for nova is https://review.openstack.org/#/c/385641/ which needs work | 17:15 |
ildikov | jungleboyj: I see some activities on it, do we need to look more into who can take care of it? | 17:15 |
johnthetubaguy | mriedem | 17:15 |
ildikov | lyarwood: coolio, tnx! | 17:15 |
johnthetubaguy | +1 | 17:15 |
mriedem | unless https://review.openstack.org/#/c/385629/ supersedes https://review.openstack.org/#/c/385641/ ? | 17:16 |
jungleboyj | johnthetubaguy: Good point. Yes, that needs to get addressed as well I think. | 17:16 |
mriedem | nvm it's a series | 17:16 |
mriedem | so you need both | 17:16 |
jungleboyj | ildikov: I thought we had someone on it but then it got quiet again. | 17:16 |
mriedem | so can the cinder team work on getting https://review.openstack.org/#/c/385641/ merged and released? | 17:16 |
ildikov | hmm, didn't we have another get_highest_version patch merged some time before the last client release? | 17:16 |
mriedem | #info need https://review.openstack.org/#/c/385641/ merged and released in cinderclient before nova can use cinder microversions | 17:17 |
jungleboyj | mriedem: Yeah, I will corner smcginnis and get it worked out. | 17:17 |
ildikov | jungleboyj: tnx! | 17:17 |
jungleboyj | Do we just need https://review.openstack.org/#/c/385641/ ? | 17:17 |
jungleboyj | ildikov: Welcome. | 17:17 |
*** mdbooth has joined #openstack-meeting-cp | 17:18 | |
mriedem | jungleboyj: it's a series | 17:18 |
mriedem | so you need both | 17:18 |
johnthetubaguy | we need to know if we can use the client to talk v3.27 or not | 17:18 |
ildikov | mriedem: so we have this merged: https://review.openstack.org/#/c/425785/ | 17:18 |
mriedem | ildikov: i'm not sure why we can't use that then | 17:18 |
mriedem | is scottda around? | 17:18 |
mriedem | can he explain the state of the changes for cinderclient and what nova needs? | 17:19 |
ildikov | mriedem: so I think we can use that | 17:19 |
mriedem | then what are the other patches for? | 17:19 |
ildikov | mriedem: and this is the bugfix that holds back the client release: https://review.openstack.org/#/c/385629/ | 17:19 |
ildikov | mriedem: so I think we only need the bugfix to land | 17:19 |
mriedem | https://bugs.launchpad.net/python-cinderclient/+bug/1632872 | 17:19 |
openstack | Launchpad bug 1632872 in python-cinderclient "discover_versions is broken" [Undecided,In progress] - Assigned to Justin A Wilson (justin-wilson) | 17:19 |
mriedem | yeah, i pointed that bug out a long time ago i think | 17:19 |
mriedem | october | 17:19 |
mriedem | geez | 17:19 |
mriedem | via code inspection | 17:19 |
mriedem | yeah so we can't use https://review.openstack.org/#/c/425785/ because it's broken | 17:20 |
mriedem | anyway | 17:20 |
mriedem | i think you guys know what you need to focus on for the client release | 17:20 |
jgriffith | what a mess | 17:20 |
* ildikov is in tears :( | 17:20 | |
jungleboyj | So, get https://review.openstack.org/#/c/385629/ and https://review.openstack.org/#/c/385641/2 landed | 17:20 |
jgriffith | jungleboyj we can talk in Cinder channel, I don't think it's that simple | 17:21 |
jgriffith | and those actually have some issues anyway | 17:21 |
jungleboyj | jgriffith: :-( Ok. | 17:21 |
johnthetubaguy | so lets just leave that with you all | 17:21 |
jungleboyj | Sorry, just getting up to speed with that stuff. | 17:21 |
jgriffith | like they'll blow up if you load them :) | 17:21 |
jgriffith | call them I mean | 17:21 |
johnthetubaguy | I think you understand what we need right? | 17:21 |
jungleboyj | jgriffith: Halt and Catch Fire | 17:21 |
ildikov | jungleboyj: get the first landed and see what the other one is I believe that's obsolete and we need to figure out whether the ideas for version discovery are good enough or not | 17:22 |
johnthetubaguy | can I pass "3.27" or should I pass "3.0" into the client factory method thingy | 17:22 |
johnthetubaguy | I think thats the aim | 17:22 |
johnthetubaguy | based on what the server and client both support | 17:22 |
ildikov | johnthetubaguy: I believe we do know that we need a cinderclient that can tell Nova what version Cinder can offer to us | 17:22 |
jungleboyj | Yep. | 17:23 |
ildikov | johnthetubaguy: on the Nova side we should get the switch to base v3 landed as that's a VERY small change and seems working | 17:23 |
mriedem | ildikov: it's in the gate | 17:23 |
ildikov | johnthetubaguy: I will upload the v2 deprecation patch this week | 17:23 |
ildikov | mriedem: \o/ :) thanks! | 17:24 |
ildikov | so when we came up with how the microversion stuff will finally be handled I'll upload the Nova side changes for that | 17:24 |
ildikov | we should also start to look into the detach case code wise | 17:25 |
ildikov | jgriffith has a PoC up for that | 17:25 |
ildikov | and get the BDM patches landed that lyarwood is working on right now | 17:25 |
johnthetubaguy | version = 3.27 ? not client.server_supports_version(3.27) : 3.0 right? | 17:26 |
*** jaugustine has joined #openstack-meeting-cp | 17:26 | |
jgriffith | I just abandoned them, but: https://review.openstack.org/#/c/438744/ and https://review.openstack.org/#/c/438750/ | 17:26 |
mriedem | jgriffith: i'm not sure why you dropped https://review.openstack.org/#/c/438750/ | 17:26 |
mriedem | jgriffith: i don't think there was a duplicate for that? | 17:26 |
jgriffith | because it's part of a chain on the first one :) | 17:26 |
ildikov | mriedem: johnthetubaguy: is there anything with the detach refactor from lyarwood that's questionable? | 17:27 |
mriedem | but you can just rebase that onto the other series from lyarwood right? | 17:27 |
jgriffith | I'll redo it and loose the dep | 17:27 |
mriedem | ildikov: i haven't looked at it | 17:27 |
jgriffith | and I did it slightly differently than lyarwood so the follow on wouldn't work anyway | 17:27 |
johnthetubaguy | ildikov: just needs more review I think | 17:27 |
jgriffith | mriedem yes, I can/will rework it to rebase against lyarwood 's patches | 17:27 |
ildikov | mriedem: it would be good to figure out whether that will work out or not so that we don't get delayed for nothing | 17:27 |
lyarwood | jgriffith: you could rebase on my series and just move everything down into the driver_bdm detach call | 17:27 |
jgriffith | lyarwood yeah, just waiting to see movement on those before doing so :) | 17:28 |
mriedem | lyarwood: has mdbooth gone through your refactor? | 17:28 |
mdbooth | mriedem: Nope | 17:28 |
lyarwood | ah there he is | 17:28 |
mriedem | ok honestly i need more eyes from nova people on this stuff, b/c i'm spread thin right now | 17:28 |
lyarwood | jgriffith: kk understood | 17:28 |
mdbooth | mriedem lyarwood: I can look hard at this next week | 17:29 |
ildikov | mriedem: do you have a list of Nova people who you would like to see these? | 17:29 |
mriedem | thanks | 17:29 |
mriedem | ildikov: mdbooth for one | 17:29 |
johnthetubaguy | so I added these patches into https://etherpad.openstack.org/p/pike-nova-priorities-tracking | 17:29 |
jgriffith | btw the changes lyarwood proposed will greatly simplify a few things for us FWIW | 17:29 |
ildikov | mriedem: I can start to get the patches to their attention | 17:29 |
mriedem | he knows | 17:29 |
ildikov | johnthetubaguy: I saw, looks great, thank you! | 17:29 |
mriedem | so before we get ahead of ourselves, | 17:29 |
mriedem | i'd be happy if the cinder team can get the client stuff done, and the nova team can get the data model changes done | 17:30 |
mriedem | i don't want to lose focus on the first steps here | 17:30 |
johnthetubaguy | +1 | 17:30 |
mriedem | because "review this and this other thing and you probably want to look at this too right now" is just overwhelming atm | 17:30 |
jgriffith | mriedem understood, I'm trying to look at that now; but I'm not sure how *quickly* that's going to go | 17:30 |
ildikov | mriedem: +1, I think we have those items listed as first ones everywhere | 17:30 |
ildikov | mriedem: will annoy the Cinder team with the client part | 17:30 |
jgriffith | mriedem smcginnis ildikov shall we create a sequence list? | 17:30 |
jgriffith | something like: | 17:31 |
jgriffith | 1. cinderclient fixes | 17:31 |
jgriffith | 2. lyarwood 's refactor patches | 17:31 |
mriedem | i believe the sequence on the nova side is just the data model changes, detach refactor, then jgriffith's detach with v3 stuff | 17:31 |
mriedem | right? | 17:31 |
jgriffith | 3. nova default ot V3 | 17:31 |
ildikov | jgriffith: it's roughly on top of this etherpad: https://etherpad.openstack.org/p/cinder-nova-api-changes | 17:31 |
mriedem | jgriffith: yeah we're on the same page | 17:31 |
jgriffith | mriedem correct, but I'm calling out everything so we're all on the same page | 17:31 |
mriedem | ok | 17:31 |
ildikov | mriedem: right, the v3 change is desired independently from the new attach/detach API | 17:31 |
johnthetubaguy | yeah, the micro-version stuff is needed before detach will work, I guess that could be two patches there | 17:32 |
jgriffith | because there's sort of a lot of irons in the fire right now | 17:32 |
ildikov | jgriffith: +1 | 17:32 |
mriedem | johnthetubaguy: how close do you feel the spec is at this point? i know mdbooth has been reviewing, but i haven't gone through it since before the ptg | 17:32 |
johnthetubaguy | so the list is in the spec work items list: https://review.openstack.org/373203 | 17:32 |
jgriffith | #action jgriffith take a shot at the version discovery stuff in the client | 17:32 |
jgriffith | can't believe I'm saying that | 17:32 |
johnthetubaguy | it feels almost there, but mdbooth and I found a... nit shall we say | 17:33 |
jgriffith | micro-versions mumble mumble grumble | 17:33 |
mriedem | jgriffith: from scratch, or take over scottda's changes? | 17:33 |
johnthetubaguy | jgriffith: all as we need is, can we use 3.27 or not | 17:33 |
mdbooth | johnthetubaguy: Probably something to be consciously swept under the rug, I think | 17:33 |
ildikov | jgriffith: I'm blushing reading that :) | 17:33 |
jgriffith | mriedem we'll see, right now considering "loosely" based on scottda 's stuff | 17:33 |
mriedem | jgriffith: ok whatever works for you guys | 17:33 |
johnthetubaguy | mdbooth: yeah, my update basically did just that | 17:33 |
mdbooth | johnthetubaguy: I'd be happy documenting that we can't do that | 17:34 |
mdbooth | Cool, haven't looked yet | 17:34 |
jgriffith | I'd like to model more closely to what exists on the Nova side... anyway, we'll see | 17:34 |
johnthetubaguy | so I don't feel the spec has had many cinder eyes on it post PTG | 17:34 |
johnthetubaguy | I feel like lyarwood and mdbooth have had a good look at many of the details in the spec | 17:34 |
ildikov | johnthetubaguy: I remember you were asking about volume migration or smth like on the Cinder channel today | 17:35 |
johnthetubaguy | feels more like polish and re-writing now | 17:35 |
johnthetubaguy | ildikov: its part of swap volume in the spec | 17:35 |
ildikov | johnthetubaguy: ok cool | 17:35 |
ildikov | johnthetubaguy: is there anything would worth discussing here so you can progress? | 17:35 |
johnthetubaguy | ildikov: basically there is no way migrate works with multi-attach, but it doesn't block the new API | 17:35 |
johnthetubaguy | the key point is attachments I think | 17:36 |
ildikov | johnthetubaguy: fair enough, that should not block the new API work | 17:36 |
johnthetubaguy | I am assuming that migrate_volume_completion deletes attachments as required | 17:36 |
ildikov | we can write up a separate spec with multi-attach corner cases and things that not going to work | 17:36 |
johnthetubaguy | as long as that is true, I think we are good | 17:36 |
johnthetubaguy | ildikov: yup | 17:36 |
ildikov | johnthetubaguy: and are we sure that's deleting? | 17:37 |
mdbooth | johnthetubaguy: Have you proposed any of the changes to the cinder api we discussed in the nova spec? | 17:37 |
*** ducttape_ has joined #openstack-meeting-cp | 17:37 | |
mdbooth | the migration completion callback | 17:37 |
*** jaugustine has quit IRC | 17:37 | |
johnthetubaguy | mdbooth: no, I am basically going to ignore the problem for now, or try to | 17:37 |
johnthetubaguy | lets take that conversation to a speparate spec | 17:38 |
*** ducttape_ has quit IRC | 17:38 | |
mdbooth | johnthetubaguy: Ok, will look later. That's going to require changes on both sides in any case. | 17:38 |
johnthetubaguy | ildikov: thats my question really | 17:38 |
*** ducttape_ has joined #openstack-meeting-cp | 17:38 | |
ildikov | johnthetubaguy: if there's anything we could/should do in Cinder we can look into that | 17:38 |
johnthetubaguy | so basically, how does migrate work with the new API | 17:38 |
johnthetubaguy | I wrote up in the spec how I think it works | 17:39 |
ildikov | jgriffith: jungleboyj: smcginnis: do you have answer to that migrate_volume_completion question? | 17:39 |
johnthetubaguy | but I dunno if thats how its implemented | 17:39 |
johnthetubaguy | line 251 in here: https://review.openstack.org/#/c/373203/19/specs/pike/approved/cinder-new-attach-apis.rst | 17:39 |
jgriffith | sorry.. | 17:39 |
jgriffith | scrolling back | 17:39 |
mdbooth | ildikov: tl;dr we want to pass attachment ids instead of volume ids everywhere in the volume migration flow | 17:39 |
johnthetubaguy | mdbooth: sweet, that would be good, I added a comment on why I put the instance into the ERROR state | 17:39 |
jgriffith | ildikov I'll have to look, that was vish and avishay that put those together | 17:40 |
johnthetubaguy | I am hoping basically it does the delete attachment stuff for us | 17:40 |
ildikov | mdbooth: I think in general that should be the desire from the Cinder side as well to use attachment_id | 17:40 |
johnthetubaguy | but something tells me it just breaks in a heap right now | 17:40 |
mdbooth | ildikov: Sweet | 17:40 |
ildikov | mdbooth: need to look into the current state of things tough | 17:40 |
johnthetubaguy | yeah, I don't think we should change the API as part of this spec, it covers enough changes as it is | 17:41 |
jgriffith | johnthetubaguy I'll look at your spec today as well if that helps | 17:41 |
johnthetubaguy | jgriffith: that would be great, need lots of eyes on that | 17:41 |
ildikov | jgriffith: I think it's definitely good to have the Cinder side view on it | 17:42 |
mdbooth | Incidentally, I still think that for multi-attach with shared connections we're going to end up locking exclusively on the Nova side | 17:42 |
johnthetubaguy | mriedem: to answer your question, my main worry is how we keep swap volume working, other than that we are good, so its tempting to merge it with open questions at this point | 17:42 |
mdbooth | Don't know if that's worth bringing up here | 17:42 |
johnthetubaguy | mdbooth: even when its in the API saying what should happen, the lock is still held on the nova side | 17:43 |
ildikov | mdbooth: eventually we will need to talk about that | 17:43 |
johnthetubaguy | I want to leave that till post this API move stuff | 17:43 |
johnthetubaguy | its totally a requirement to get multi-attach working | 17:44 |
mdbooth | johnthetubaguy: Ok, sure. But basically the problem is ordering. The only thing which can know everything about call order is the nova compute process. | 17:44 |
johnthetubaguy | mdbooth: +! | 17:44 |
johnthetubaguy | oops | 17:44 |
johnthetubaguy | +1 | 17:44 |
mdbooth | So the api can say: "... and you're the last one", but by the time it executes it might not be. | 17:44 |
ildikov | mdbooth: if you could list concerns, ideas, anything here that would be great: https://etherpad.openstack.org/p/cinder-nova-api-changes | 17:45 |
johnthetubaguy | mdbooth: correct, thats why I proposed to re-query that fact with the lock held | 17:45 |
johnthetubaguy | mdbooth: the lock has to be nova side | 17:45 |
mdbooth | johnthetubaguy: I might have to squint at that again too. | 17:45 |
* johnthetubaguy always fails at not talking about something | 17:45 | |
* mdbooth too, sorry | 17:46 | |
ildikov | mdbooth: that's the etherpad to get multi-attach work so we're trying to add tasks there and action items too so we don't miss anything | 17:46 |
johnthetubaguy | mdbooth: no worries, its so tempting to try fix this stuff | 17:46 |
johnthetubaguy | so... can we get folks to review the spec | 17:46 |
johnthetubaguy | make sure it says what we said at the PTG, and matches the cinder view of the world? | 17:46 |
ildikov | mdbooth: I agree on that part re the last attachment, we need to figure that out | 17:46 |
johnthetubaguy | #help please review the nova spec: https://review.openstack.org/#/c/373203 | 17:47 |
ildikov | johnthetubaguy: +1 | 17:47 |
ildikov | smcginnis: jungleboyj: please take a look on the Nova spec ^^ | 17:47 |
* mdbooth is still trying to pimp this change: https://review.openstack.org/#/c/383859/ | 17:47 | |
*** KeithMnemonic has quit IRC | 17:47 | |
ildikov | as we have jgriffith signed up already :) | 17:47 |
mdbooth | It will require reworking for the new api, but I think the approach is solid | 17:47 |
ildikov | mdbooth: reworking the new API? | 17:48 |
jgriffith | ildikov "reworking *for* the new API" | 17:48 |
jungleboyj | johnthetubaguy: Adding it to my list. | 17:49 |
jgriffith | mad rush to get every change related to attach/detach possible in NOW before we change the API version it seems | 17:49 |
ildikov | jgriffith: ah, ok, tnx :) | 17:49 |
jgriffith | not a great idea IMO | 17:49 |
*** ducttape_ has quit IRC | 17:49 | |
jgriffith | but haven't looked at patch yet either so... | 17:50 |
ildikov | jgriffith: do you mean from testing perspective? | 17:50 |
johnthetubaguy | mdbooth: thats not applying to cinder volumes yet, its just nova ephemeral? or did I miss read that? | 17:50 |
mdbooth | johnthetubaguy: No, that's cinder volumes on shared filesystems | 17:50 |
jgriffith | I mean from too many moving levers perspective, we'll never get things settled enough to actually finish the new API work if we keep adding "one more thing" | 17:50 |
johnthetubaguy | mdbooth: OK, so I guess I don't understand it then, I need to read more | 17:51 |
ildikov | jgriffith: ah, ok | 17:51 |
mdbooth | johnthetubaguy: I believe it's race free and supports multi-attach (at least notionally). | 17:51 |
ildikov | jgriffith: knowing about issue and have them on the radar is good IMHO | 17:51 |
johnthetubaguy | mdbooth: my worry is the locking is too technology specific, the previous proposal at the PTG was technology agnostic, I should get on a hangout one morning to work out whats different between the two approaches | 17:52 |
jgriffith | ildikov sure, always | 17:52 |
mdbooth | Effectively does reference counting with additional sanity checks on the compute side. | 17:52 |
johnthetubaguy | mdbooth: they might actually be the same | 17:52 |
ildikov | jgriffith: but we need to get progress on the items you listed earlier before moving on to solve others | 17:52 |
mdbooth | johnthetubaguy: Sounds like a plan. | 17:52 |
johnthetubaguy | mdbooth: both approaches did that | 17:52 |
johnthetubaguy | mdbooth: you around Monday? | 17:52 |
mdbooth | johnthetubaguy: Tuesday is better. Lets take that offline. | 17:53 |
johnthetubaguy | #action mdbooth and johnthetubaguy to work out how their approaches are the same and how they are different | 17:53 |
johnthetubaguy | mdbooth: ack | 17:53 |
johnthetubaguy | mdbooth: so the instance going into ERROR state, I think that was the only other thing you mentioned | 17:54 |
johnthetubaguy | in the spec | 17:54 |
ildikov | we have 6 minutes from the today's meeting | 17:54 |
mdbooth | johnthetubaguy: Yeah, I think that's a detail that will come out in the wash. | 17:54 |
johnthetubaguy | I forgot to say the migration object should also go into the error state, as needed | 17:54 |
johnthetubaguy | but anyways, updated the spec with that | 17:54 |
mdbooth | I was more observing the current behaviour, rather than commenting on what it should be. | 17:54 |
johnthetubaguy | ah, got ya | 17:54 |
mdbooth | I didn't see where we set the instance to an error state. | 17:55 |
johnthetubaguy | yeah, I think we need to do better there, and put the instance into ERROR, if we have a failed rollback, if rollback works, then you go back to ACTIVE | 17:55 |
johnthetubaguy | we already do that for certain other exceptions | 17:55 |
johnthetubaguy | like say you can't talk to the other compute node over RPC and it never responds, etc | 17:55 |
johnthetubaguy | basically any failed rollback should go into the error state | 17:56 |
johnthetubaguy | (there is a bug where sometimes it gets stuck in a migrating state, but it should be in the ERROR state) | 17:56 |
johnthetubaguy | ildikov: what else do we need to cover, we have the work for this week I think | 17:56 |
johnthetubaguy | lets recap that I guess | 17:56 |
johnthetubaguy | 1) cinder client release | 17:56 |
ildikov | johnthetubaguy: I think we're good with the action items | 17:56 |
ildikov | johnthetubaguy: please review the v3 switch patch, it's really tiny | 17:57 |
johnthetubaguy | 2) BDM patches | 17:57 |
johnthetubaguy | ildikov: its already +Wed and in the gate I think | 17:57 |
ildikov | johnthetubaguy: I'm not myself today :) | 17:57 |
johnthetubaguy | ildikov: no worries | 17:57 |
ildikov | so yep | 17:57 |
johnthetubaguy | 3) review nova spec | 17:58 |
ildikov | and 4) is detach refactor | 17:58 |
johnthetubaguy | 4) patches to use new cinder client release, + detatch | 17:58 |
johnthetubaguy | yeah, | 17:58 |
johnthetubaguy | detach depends on the new cinder client release | 17:58 |
johnthetubaguy | I think | 17:58 |
ildikov | the refactor work should not | 17:58 |
johnthetubaguy | right, I guess I called them BDM patches | 17:58 |
johnthetubaguy | so +1 | 17:59 |
ildikov | ah ok, I thought those are more the attachment_id and UUID changes | 17:59 |
ildikov | it looks good then | 17:59 |
johnthetubaguy | theres a few, its in the etherpad | 18:00 |
johnthetubaguy | I guess we are out of time | 18:00 |
ildikov | yep, I think we have enough work to get done until the next meeting | 18:00 |
ildikov | anything else from anyone that's urgent for today? | 18:00 |
ildikov | I take this as a no :) | 18:01 |
* johnthetubaguy goes for his dinner | 18:01 | |
ildikov | thanks everyone! please do reviews! | 18:01 |
ildikov | have a good rest of the day! | 18:02 |
ildikov | #endmeeting | 18:02 |
*** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings" | 18:02 | |
openstack | Meeting ended Thu Mar 9 18:02:14 2017 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 18:02 |
openstack | Minutes: http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-03-09-17.00.html | 18:02 |
openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-03-09-17.00.txt | 18:02 |
openstack | Log: http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-03-09-17.00.log.html | 18:02 |
jungleboyj | Thanks! | 18:02 |
*** breitz has left #openstack-meeting-cp | 18:02 | |
*** mriedem has left #openstack-meeting-cp | 18:03 | |
*** stvnoyes has joined #openstack-meeting-cp | 18:10 | |
*** ducttape_ has joined #openstack-meeting-cp | 18:50 | |
*** ducttape_ has quit IRC | 18:55 | |
*** david-lyle has quit IRC | 19:14 | |
*** sdague has quit IRC | 19:26 | |
*** sdague has joined #openstack-meeting-cp | 19:28 | |
*** alij has joined #openstack-meeting-cp | 19:29 | |
*** ducttape_ has joined #openstack-meeting-cp | 19:51 | |
*** alij has quit IRC | 19:52 | |
*** david-lyle has joined #openstack-meeting-cp | 19:53 | |
*** rderose has joined #openstack-meeting-cp | 19:55 | |
*** ducttape_ has quit IRC | 19:56 | |
*** rockyg has joined #openstack-meeting-cp | 19:58 | |
robcresswell | #startmeeting keystone_horizon | 20:00 |
openstack | Meeting started Thu Mar 9 20:00:06 2017 UTC and is due to finish in 60 minutes. The chair is robcresswell. Information about MeetBot at http://wiki.debian.org/MeetBot. | 20:00 |
openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 20:00 |
*** openstack changes topic to " (Meeting topic: keystone_horizon)" | 20:00 | |
openstack | The meeting name has been set to 'keystone_horizon' | 20:00 |
robcresswell | o/ | 20:00 |
dstanek | ehlo | 20:00 |
lbragstad | o/ | 20:00 |
* lbragstad is on another call - will be multi-tasking :-/ | 20:00 | |
cmurphy | o/ | 20:01 |
david-lyle | o/ | 20:01 |
robcresswell | No problem, I think it'll be a quick one today, still havent had much time to get things moving | 20:01 |
robcresswell | #link https://etherpad.openstack.org/p/keystone-horizon | 20:01 |
rderose | o/ | 20:02 |
robcresswell | Most of this is waiting on Horizon, though I might be pestering keystone for some expertise too. Spent a bunch of time last week properly wrapping my head around scopy-ness | 20:02 |
robcresswell | I was hoping someone could offer some insight on https://review.openstack.org/#/c/339487/ | 20:02 |
*** ayoung has quit IRC | 20:03 | |
robcresswell | cmurphy noted that the issue couldnt be easily reproduced, so I was wondering if it was fixed by some of the domain handling patches that went in fairly recently, or if this is a sepatate issue | 20:03 |
robcresswell | separate* | 20:03 |
david-lyle | I haven't tried to repro yet | 20:04 |
david-lyle | not sure if it's related or not | 20:05 |
robcresswell | Yeah, a lot of Horizons policy handling was written by a madman before my time :p | 20:05 |
david-lyle | but if domain context was being messed up, it's highly likely to have effected these tasks | 20:05 |
rdopiera | o/ | 20:05 |
robcresswell | Yeah, that was my thinking, and I expect rdopiera's patches may have fixed it, or at least affected it | 20:06 |
robcresswell | The second question I had was regarding PCI-DSS 8.2.6 - there's a comment on the keystone side saying "We still need a way to express that the 401 was the result of an expired password" | 20:07 |
robcresswell | rderose: ^^ do you know if that was resolved on the keystone side, before we start trying to use it on the Horizon side? | 20:08 |
rdopiera | I'm not sure, but I don't think it touched any related code | 20:08 |
rderose | robcresswell: not yet | 20:08 |
robcresswell | rderose: Any relevant patches I could link to? | 20:08 |
rderose | robcresswell: yeah, we still need to do this, so that you know auth was success, but password is expired | 20:08 |
knikolla | o/ sorry i'm late. | 20:09 |
rderose | robcresswell: not yet, but I'll try to get something up soon | 20:09 |
robcresswell | rderose: Sure no rush, it's mainly us holding things up as it is; just wanted to be able to track status accurately | 20:09 |
robcresswell | I've made a note to wait for keystone to do that bit | 20:09 |
rderose | robcresswell: cool | 20:10 |
rderose | robcresswell: the other PCI stuff is ready to go | 20:10 |
robcresswell | Awesome. I've been trying to get a couple of our easier blueprints reviewed the past week, so havent really written much code myself. | 20:10 |
robcresswell | Should start getting some time to move things soon | 20:11 |
robcresswell | So those are both of things I wanted to bring up | 20:11 |
robcresswell | Any other topics for discussion? | 20:11 |
lbragstad | I don't think i have anything | 20:12 |
david-lyle | I've been off in the weeds | 20:12 |
robcresswell | haha | 20:12 |
robcresswell | No problem, we can call it early and give everyone their time back | 20:13 |
lbragstad | woo - thanks robcresswell | 20:13 |
david-lyle | will try to target some time on finishing is_admin_project support | 20:13 |
robcresswell | If anyone has any concerns or discussion points please feel free to add to the etherpad | 20:13 |
robcresswell | Things like specs that we could weigh in on, I know lbragstad has pinged me about a couple | 20:14 |
robcresswell | Thanks everyone! | 20:14 |
robcresswell | #endmeeting | 20:14 |
*** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings" | 20:14 | |
openstack | Meeting ended Thu Mar 9 20:14:36 2017 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 20:14 |
cmurphy | \o | 20:14 |
openstack | Minutes: http://eavesdrop.openstack.org/meetings/keystone_horizon/2017/keystone_horizon.2017-03-09-20.00.html | 20:14 |
openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/keystone_horizon/2017/keystone_horizon.2017-03-09-20.00.txt | 20:14 |
openstack | Log: http://eavesdrop.openstack.org/meetings/keystone_horizon/2017/keystone_horizon.2017-03-09-20.00.log.html | 20:14 |
david-lyle | thanks | 20:14 |
*** r1chardj0n3s has joined #openstack-meeting-cp | 20:34 | |
*** rockyg has quit IRC | 20:41 | |
*** sdague has quit IRC | 20:41 | |
*** ayoung has joined #openstack-meeting-cp | 20:42 | |
*** r1chardj0n3s has left #openstack-meeting-cp | 20:43 | |
*** MarkBaker has quit IRC | 20:50 | |
*** ducttape_ has joined #openstack-meeting-cp | 20:52 | |
*** MarkBaker has joined #openstack-meeting-cp | 20:53 | |
*** ducttape_ has quit IRC | 20:56 | |
*** ayoung has quit IRC | 21:06 | |
*** MarkBaker has quit IRC | 21:10 | |
*** ayoung has joined #openstack-meeting-cp | 21:40 | |
*** ducttape_ has joined #openstack-meeting-cp | 21:52 | |
*** ducttape_ has quit IRC | 21:57 | |
*** ayoung has quit IRC | 22:03 | |
*** ducttape_ has joined #openstack-meeting-cp | 22:28 | |
*** gouthamr has quit IRC | 22:48 | |
*** ducttape_ has quit IRC | 22:59 | |
*** ducttape_ has joined #openstack-meeting-cp | 23:04 | |
*** gouthamr has joined #openstack-meeting-cp | 23:10 | |
*** MarkBaker has joined #openstack-meeting-cp | 23:16 | |
*** ayoung has joined #openstack-meeting-cp | 23:37 | |
*** ducttape_ has quit IRC | 23:48 | |
*** ducttape_ has joined #openstack-meeting-cp | 23:58 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!