Thursday, 2017-09-07

*** felipemonteiro has quit IRC00:11
*** sdague has quit IRC00:36
*** harlowja has quit IRC00:48
*** aselius has quit IRC01:59
*** markvoelker has quit IRC02:11
*** yamahata has quit IRC02:15
*** iyamahat has quit IRC02:15
*** brault has quit IRC03:09
*** markvoelker has joined #openstack-meeting-cp03:11
*** coolsvap has joined #openstack-meeting-cp03:19
*** markvoelker has quit IRC03:46
*** nhelgeson has quit IRC04:05
*** harlowja has joined #openstack-meeting-cp04:06
*** gouthamr has quit IRC04:31
*** dims has quit IRC04:40
*** rakhmerov has quit IRC04:43
*** markvoelker has joined #openstack-meeting-cp04:43
*** rakhmerov has joined #openstack-meeting-cp04:44
*** reed_ has joined #openstack-meeting-cp04:53
*** reed_ has quit IRC04:54
*** harlowja has quit IRC05:11
*** markvoelker has quit IRC05:16
*** iyamahat has joined #openstack-meeting-cp05:18
*** iyamahat_ has joined #openstack-meeting-cp05:19
*** iyamahat has quit IRC05:21
*** diablo_rojo has joined #openstack-meeting-cp05:48
*** iyamahat_ has quit IRC06:02
*** diablo_rojo has quit IRC06:11
*** markvoelker has joined #openstack-meeting-cp06:13
*** iyamahat has joined #openstack-meeting-cp06:26
*** brault has joined #openstack-meeting-cp06:46
*** markvoelker has quit IRC06:47
*** iyamahat has quit IRC07:41
*** markvoelker has joined #openstack-meeting-cp07:44
*** markvoelker has quit IRC08:17
*** markvoelker has joined #openstack-meeting-cp09:14
*** yamahata has joined #openstack-meeting-cp09:22
*** markvoelker has quit IRC09:48
*** edmondsw has joined #openstack-meeting-cp09:54
*** edmondsw has quit IRC09:58
*** markvoelker has joined #openstack-meeting-cp10:45
*** sdague has joined #openstack-meeting-cp11:06
*** markvoelker has quit IRC11:17
*** MarkBaker has joined #openstack-meeting-cp11:28
*** dims has joined #openstack-meeting-cp12:07
*** markvoelker has joined #openstack-meeting-cp12:15
*** markvoelker has quit IRC12:24
*** markvoelker has joined #openstack-meeting-cp12:24
*** edmondsw has joined #openstack-meeting-cp12:32
*** edmondsw has quit IRC12:43
*** edmondsw has joined #openstack-meeting-cp12:43
*** edmondsw has quit IRC12:48
*** edmondsw has joined #openstack-meeting-cp12:54
*** felipemonteiro has joined #openstack-meeting-cp13:03
*** felipemonteiro_ has joined #openstack-meeting-cp13:05
*** MarkBaker has quit IRC13:06
*** iyamahat has joined #openstack-meeting-cp13:08
*** felipemonteiro has quit IRC13:09
*** edmondsw_ has joined #openstack-meeting-cp13:14
*** edmondsw has quit IRC13:15
*** MarkBaker has joined #openstack-meeting-cp13:19
*** edmondsw_ is now known as edmondsw13:29
*** gouthamr has joined #openstack-meeting-cp13:29
*** felipemonteiro_ has quit IRC13:30
*** f13o has joined #openstack-meeting-cp13:42
*** f13o has quit IRC13:42
*** coolsvap has quit IRC13:57
*** edmondsw has quit IRC14:30
*** edmondsw has joined #openstack-meeting-cp14:37
*** felipemonteiro has joined #openstack-meeting-cp14:58
*** felipemonteiro_ has joined #openstack-meeting-cp14:59
*** felipemonteiro has quit IRC15:03
*** iyamahat has quit IRC15:16
*** MarkBaker has quit IRC15:17
*** aselius has joined #openstack-meeting-cp15:48
*** mriedem has joined #openstack-meeting-cp15:51
ildikov#startmeeting cinder-nova-api-changes16:00
openstackMeeting started Thu Sep  7 16:00:47 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
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.16:00
*** openstack changes topic to " (Meeting topic: cinder-nova-api-changes)"16:00
openstackThe meeting name has been set to 'cinder_nova_api_changes'16:00
ildikovjohnthetubaguy jaypipes e0ne jgriffith hemna mriedem patrickeast smcginnis diablo_rojo xyang1 raj_singh lyarwood jungleboyj stvnoyes16:01
stvnoyeso/16:01
mriedemo.16:01
* jungleboyj is watching out of the corner of my eye16:01
ildikovhi all16:01
ildikovlet's dive in, I think the today's meeting will be a short one16:03
ildikovwe have a bunch of open reviews: https://review.openstack.org/#/q/topic:bp/cinder-new-attach-apis+status:open16:03
ildikovthe gate is in general happy with the patches right now16:04
ildikovwe have the new cinderclient out which fixed a bunch of things for us16:04
ildikovmriedem: do you think we can get these things merged by the end of next week?16:05
johnthetubaguythat big patch on version 133 is scaring me16:05
ildikovmriedem: I'm happy to sit together with you and johnthetubaguy and/or other cores interested and get things fixed and merged16:05
mriedemhttps://review.openstack.org/#/c/330285/ will likely not be merged by the end of next week16:05
ildikovjohnthetubaguy: it has a ton of tests if you mean the size of that thing16:06
ildikovmriedem: if we could spend time on figuring out if there's anything fundamentally questionable with that still that would be great16:06
ildikovmriedem: and then I can polish it up as needed that should not be a big issue16:06
johnthetubaguyinitially the size of it, there are some specifics, its more about the context of the changes being hard to reason about16:06
stvnoyeshow about a review for https://review.openstack.org/#/c/463987/ ? (migrate) While I still remember what I did...16:07
*** persia has quit IRC16:07
ildikovjohnthetubaguy: we cannot really split it up to smaller changes honestly16:07
mriedemstvnoyes: i'm runn‌ing tests in another patch on that again today16:08
mriedemhttps://review.openstack.org/#/c/481290/16:08
mriedemi would like to get that one in soon16:08
stvnoyesok great. let me know if there's anything I can do16:09
mriedemlooks like it's passing so that's good16:09
ildikovI checked that briefly too, it looked good16:09
mriedem{0} tempest.api.compute.admin.test_live_migration.LiveMigrationRemoteConsolesV26Test.test_volume_backed_live_migration [26.192704s] ... ok16:09
mriedem{0} tempest.api.compute.admin.test_live_migration.LiveMigrationRemoteConsolesV26Test.test_volume_backed_live_migration [26.192704s] ... ok16:09
* smcginnis catches up on scrollback16:09
mriedemoh those are the same16:09
mriedem{2} tempest.api.compute.admin.test_live_migration.LiveAutoBlockMigrationV225Test.test_volume_backed_live_migration [35.446023s] ... ok16:09
stvnoyesmriedem: i also updated the live migrate tempest test to check that the attachments were done correctly - https://review.openstack.org/#/c/487884/16:10
johnthetubaguynice16:10
mriedemstvnoyes: problem is,16:10
mriedemwe don't ever run test_iscsi_volume16:10
stvnoyeshmmm, I'll need to recheck, I did that test because I was finding that the other tests didn't exercise the new code, but let me go back and recheck that16:11
stvnoyesi'll let you know16:11
stvnoyesif the other tests do what we need i'll move the checks over to them16:12
johnthetubaguyBFV one just have some attachments that need testing, although that is a different path16:12
mriedemhttp://logs.openstack.org/90/481290/3/check/gate-tempest-dsvm-multinode-live-migration-ubuntu-xenial/e8a67e8/console.html#_2017-09-07_15_03_37_33132516:12
johnthetubaguybummer16:12
ildikovyeah :/16:13
mriedemi did have https://review.openstack.org/#/c/459316/16:13
mriedemso, i could roll a dnm change on top of https://review.openstack.org/#/c/481290/ that depends on ^16:14
mriedemand see if the scsi live migration works16:14
mriedemit's just that it intermittenly fails on it's own16:14
johnthetubaguymriedem: do you remember how it failed?16:15
mriedemit was april, so no16:15
mriedemwell, although,16:16
mriedemin https://review.openstack.org/#/c/459316/ it's not the live migration job that's failing16:16
johnthetubaguyyeah, that's all I saw16:16
mriedemso for the scsi one i'd have to recheck those as the logs are gone16:17
mriedemand i think the patch above it failed the scsi test16:17
mriedembut the logs are gone16:17
mriedem"The live migration job failed on the iscsi test which is enabled in the  change below it, so maybe that's not very stable after all."16:17
johnthetubaguyhmm, I see now16:18
johnthetubaguylike you say though, getting a pass from that would be no bad thing16:19
mriedemyeah, so i can stack something up there to run it16:19
mriedemplus stvnoyes's tempest patch as a dependency16:19
mriedemjesus my tab farm is growing16:19
ildikovthat would be great to have those tests in16:21
ildikovand this farm is just crazy I always end up cleaning up around 5-6 merged dependencies, etc...16:21
ildikovmriedem: do you have capacity to do these this week or early next?16:22
mriedemdo how?16:22
mriedemno i don't have capacity, but i'll hump what i can16:22
ildikovok, no worries, I'm just trying to figure out what needs to get done and who can do it16:23
mriedemi will try to stack these iscsi live migration volume test changes to run those against stvnoyes's changes16:24
mriedemand go through the live migration change again16:24
ildikovthat sounds great16:24
ildikovthank you16:24
ildikovdoes anything else need more testing BTW?16:25
ildikovwe have coverage in Grenade and stvnoyes made sure we have updates in tempest for a few items16:25
ildikovI added unit and functional tests to the attach patch16:25
johnthetubaguycan I ask a dumb question?16:26
ildikovshoot :)16:26
johnthetubaguythe live-migrate patch, we are only testing the old code path isn't broken, because the attach new flow patch is separate?16:26
johnthetubaguyi.e. bdm.attachment_id is always empty?16:27
ildikovjohnthetubaguy: the attach patch depends on the live migrate one16:27
mriedemjohnthetubaguy: i'm testing those together here https://review.openstack.org/#/c/481290/16:28
stvnoyescurrently the ci is testing the patch with empty attach id's since the live migrate patch will go in forst16:28
ildikovjohnthetubaguy: and I think the DNM patch also has the Depends-On tags to get the new code in16:28
stvnoyesfirst16:28
ildikovthe attach patch has the tests running with the full code base16:28
johnthetubaguyOK, got it16:28
johnthetubaguythanks16:28
johnthetubaguytotally didn't see that depends-on in there16:28
johnthetubaguyI just wonder if we want live-migrate patch passing those extra tests before and after16:29
johnthetubaguythe new attach flow16:29
ildikovjohnthetubaguy: not your fault we have a ton everywhere that I clean up occasionally, so it's hard to follow16:29
stvnoyesjohnthetubaguy- not sure what you mean, which tests?16:30
mriedemthe live migration patch itself is passing live migration with the old flow16:30
mriedemmy patch with the depends-on is passing it with the new flow16:31
mriedemsince it depends on both the live migration change and the api change to use the new flow16:31
johnthetubaguyah, sweet16:31
mriedemwhich is why i was making sure it was working first16:31
ildikovthe api change depends-on the live migrate one too to pass the functional test for it...16:31
mriedemand i'll stack something on top of that to run the iscsi test16:31
johnthetubaguycool, sorry I wasn't following those moving bits before, makes sense to me now16:32
mriedemit's hard to follow16:33
ildikovwe have less moving bits now, so it's getting there16:33
johnthetubaguyyeah, there is some light at the end of the tunnel16:34
ildikovshould we try to get together for a short group review next week while we're all at the same place?16:35
ildikovor we now manage without that?16:35
johnthetubaguyI like the idea of trying that16:35
ildikovI think we plan to update the teams on this activity, but that will not get things merged16:35
ildikovand I don't think we would want to go into that deep details on the joint session16:36
johnthetubaguyI think I get the new attaching -> attached stuff, is there context on what changed peoples minds on that?16:36
ildikovso I'm happy to block some time next week to get a better progress on these patches and sort out anything that's still questionable16:36
ildikovjohnthetubaguy: we needed to have a final 'OK' from the consumer, in this case Nova to tell Cinder the connection is all built up on the host16:37
johnthetubaguyildikov: its for the API state I guess, it went attached to early without that?16:38
ildikovjohnthetubaguy: otherwise the volume would get to 'in-use' before brick finished it's business and you can start a detach too early which messed up the things on the compute16:38
ildikov*its16:38
johnthetubaguyah... you can start detach too early, got it16:38
ildikovyeah, we had broken iscsi connections16:38
ildikovin the test logs as they execute the calls in a sequence, so we realized we need the extra state change16:39
johnthetubaguyI remember wanting it before, I couldn't remember why, thats it16:39
ildikovwe got there eventually :)16:39
stvnoyesi wonder if that had something to do with the failing iscsi test16:39
johnthetubaguyI think I commented in one patch, delete is called before create, during a move16:39
ildikovit's quite complicated especially with having the two flows existing in parallel, so it just makes it harder to get a clean slate to re-design it or even just to see the big picture...16:40
johnthetubaguyI thought it was going to the other way around to avoid a window where the volume gets detached16:40
*** MarkBaker has joined #openstack-meeting-cp16:40
johnthetubaguyildikov: upgrade certainly spoils almost every party16:41
ildikovgrenade seems fine with the simple attach-detach at least16:42
ildikovwe tried to cover the create/delete path, although without re-writing the whole code it's tough to keep orders16:42
ildikovalthough I think we didn't really leave a big window anywhere16:42
johnthetubaguyits line 2951 in here: https://review.openstack.org/#/c/330285/133/nova/compute/manager.py16:43
johnthetubaguyit might be correct, just not got my head around that bit16:43
johnthetubaguyI am struggling to remember what part of the flow all this little helper methods are, probably have to go re-read that16:45
ildikovI need to double check that part as we had a few rebase rounds and we might left it there by mistake16:45
ildikovjohnthetubaguy: I have to re-understand those helper methods every second week...16:46
johnthetubaguyildikov: might be worth adding some comments on those methods, if you re-discover what they are really doing?16:46
ildikovjohnthetubaguy: partially why I asked whether we want to get together next week and go through these16:46
mriedemstvnoyes: here is the change to test the iscsi tempest thing https://review.openstack.org/#/c/501805/16:46
ildikovjohnthetubaguy: will try, although sometimes they are used by multiple things and I didn't feel that I want to be responsible for a not fully correct note... :)16:47
stvnoyesok great, I had just been running it locally.16:47
ildikovjohnthetubaguy: as letting people to figure it out is still better than misleading them16:47
johnthetubaguyildikov: yeah, I think we might want to bit the bullet on understanding those16:47
johnthetubaguyildikov: I am happy to review the comments and share the blame, normally I am totally against such comments because of the missleading over time thing16:48
ildikovjohnthetubaguy: same here16:48
johnthetubaguyjust to go back to a think16:48
ildikovjohnthetubaguy: we can try to draw some high level diagrams just to visualize the flows16:48
johnthetubaguyassuming its needed, can we do attach before the detach there?16:49
ildikovjohnthetubaguy: which one do you mean?16:49
johnthetubaguyI suspect it may fail if the rebuild lands on the same host like it does in the gate16:49
*** edmondsw has quit IRC16:49
ildikovjohnthetubaguy: do you mean the prep_block_device one now?16:50
johnthetubaguyboth of them16:50
johnthetubaguywell, the detach followed by attach_create16:50
johnthetubaguyI mean attachment_delete followed by create16:51
johnthetubaguythe reverse is less racey16:51
ildikovyeah, we reserve the volume with create there16:51
johnthetubaguytotally need the create, I just want it before the delete, ideally16:51
*** edmondsw has joined #openstack-meeting-cp16:51
ildikovI need to check the Cinder API as I remember we said that we will allow things for the same instance, just need to double check on where things evolved in practice16:52
ildikovit's also tough with these helper methods to ensure we get the right order of things, etc.16:53
ildikovjohnthetubaguy: please add all questions/comments you have to the review and I will ensure we catch mriedem and jgriffith next week to go through the questionable ones16:54
johnthetubaguyyeah, they should all be in comments on the review now16:54
ildikovok, we have 6 more minutes left16:54
ildikovjohnthetubaguy: +1, thanks16:55
*** edmondsw_ has joined #openstack-meeting-cp16:55
ildikovis there anything else to discuss so we're all prepared for next week?16:55
johnthetubaguymulti-attach?16:56
* johnthetubaguy ducks16:56
ildikovjohnthetubaguy: I uploaded the spec for review16:56
*** edmondsw has quit IRC16:56
ildikovjohnthetubaguy: it's in WIP, I will add a few small changes to it, so we kind of have the base for discussions16:56
johnthetubaguyI should totally re-read that, lots of the folks I work with really want that16:56
ildikovjohnthetubaguy: there were a lot of people from a lot of areas who want that16:57
*** edmondsw_ has quit IRC16:57
*** MarkBaker has quit IRC16:57
ildikovjohnthetubaguy: that's why I'm pushing for it to talk about it next week and get progress16:57
*** edmondsw has joined #openstack-meeting-cp16:57
johnthetubaguytotally, just hinting I should be able to make time to help keep moving that, not 100% sure16:57
ildikovjohnthetubaguy: feel free to add a ton of comments/questions to the spec review16:58
ildikovjohnthetubaguy: that would be great!16:58
johnthetubaguyyou got the link for that spec?16:58
ildikovhttps://review.openstack.org/#/c/499777/16:58
johnthetubaguy#link https://review.openstack.org/#/c/499777/16:58
johnthetubaguysweet16:59
ildikovok, we hit the top of the hour, but we can continue on the channels if there's more to discuss for prep16:59
ildikovjohnthetubaguy: good to have you back :)16:59
johnthetubaguynice to be around again17:00
ildikov:)17:00
ildikovthank you all for today17:00
ildikovand see you all next week! :)17:00
*** mriedem has left #openstack-meeting-cp17:00
ildikov#endmeeting17:00
*** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings"17:00
openstackMeeting ended Thu Sep  7 17:00:31 2017 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)17:00
openstackMinutes:        http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-09-07-16.00.html17:00
openstackMinutes (text): http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-09-07-16.00.txt17:00
openstackLog:            http://eavesdrop.openstack.org/meetings/cinder_nova_api_changes/2017/cinder_nova_api_changes.2017-09-07-16.00.log.html17:00
*** edmondsw has quit IRC17:00
*** iyamahat has joined #openstack-meeting-cp17:08
*** edmondsw has joined #openstack-meeting-cp17:10
*** iyamahat_ has joined #openstack-meeting-cp17:13
*** edmondsw has quit IRC17:14
*** persia has joined #openstack-meeting-cp17:16
*** iyamahat has quit IRC17:16
*** harlowja has joined #openstack-meeting-cp17:19
*** edmondsw has joined #openstack-meeting-cp17:21
*** edmondsw has quit IRC17:25
*** edmondsw has joined #openstack-meeting-cp17:33
*** gouthamr has quit IRC17:33
*** edmondsw has quit IRC17:37
*** edmondsw has joined #openstack-meeting-cp17:37
*** edmondsw has quit IRC17:49
*** edmondsw has joined #openstack-meeting-cp17:59
*** gouthamr has joined #openstack-meeting-cp18:01
*** edmondsw has quit IRC18:18
*** edmondsw has joined #openstack-meeting-cp18:24
*** kbyrne has joined #openstack-meeting-cp18:30
*** dims has quit IRC18:34
*** dims has joined #openstack-meeting-cp18:36
*** dims has quit IRC18:41
*** sdague has quit IRC18:42
*** yamahata has quit IRC18:42
*** edmondsw has quit IRC18:44
*** dims has joined #openstack-meeting-cp18:45
*** edmondsw has joined #openstack-meeting-cp18:48
*** edmondsw has quit IRC18:48
*** edmondsw has joined #openstack-meeting-cp18:54
*** iyamahat_ has quit IRC18:56
*** edmondsw has quit IRC18:57
*** edmondsw has joined #openstack-meeting-cp18:57
*** edmondsw has quit IRC18:58
*** edmondsw_ has joined #openstack-meeting-cp19:05
*** edmondsw_ has quit IRC19:09
*** iyamahat has joined #openstack-meeting-cp19:12
*** xyang1 has joined #openstack-meeting-cp19:15
*** yamahata has joined #openstack-meeting-cp19:31
*** edmondsw has joined #openstack-meeting-cp19:37
*** nhelgeson has joined #openstack-meeting-cp19:46
*** edmondsw has quit IRC19:53
*** edmondsw has joined #openstack-meeting-cp19:53
*** edmondsw has quit IRC19:58
*** iyamahat has quit IRC20:03
*** iyamahat has joined #openstack-meeting-cp20:03
*** edmondsw has joined #openstack-meeting-cp20:14
*** diablo_rojo has joined #openstack-meeting-cp20:16
*** iyamahat has quit IRC20:47
*** iyamahat_ has joined #openstack-meeting-cp20:47
*** scottda_ has joined #openstack-meeting-cp20:59
*** edmondsw has quit IRC21:32
*** edmondsw has joined #openstack-meeting-cp21:34
*** edmondsw has quit IRC21:38
*** felipemonteiro_ has quit IRC21:46
*** diablo_rojo has quit IRC21:56
*** diablo_rojo has joined #openstack-meeting-cp22:07
*** diablo_rojo_ has joined #openstack-meeting-cp22:21
*** diablo_rojo has quit IRC22:24
*** diablo_rojo__ has joined #openstack-meeting-cp22:39
*** diablo_rojo_ has quit IRC22:42
*** diablo_rojo_ has joined #openstack-meeting-cp22:43
*** diablo_rojo__ has quit IRC22:46
*** iyamahat_ has quit IRC22:53
*** iyamahat has joined #openstack-meeting-cp22:53
*** xyang1 has quit IRC23:06
*** gouthamr has quit IRC23:09
*** diablo_rojo_ has quit IRC23:21
*** diablo_rojo_ has joined #openstack-meeting-cp23:22
*** diablo_rojo_ has quit IRC23:43
*** gouthamr has joined #openstack-meeting-cp23:58

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!