| *** VanL has joined #openstack-defcore | 00:04 | |
| *** VanL has quit IRC | 00:31 | |
| *** VanL has joined #openstack-defcore | 01:42 | |
| *** VanL has quit IRC | 01:46 | |
| *** fallenpegasus has joined #openstack-defcore | 02:03 | |
| *** markvoelker has quit IRC | 03:31 | |
| *** markvoelker has joined #openstack-defcore | 04:32 | |
| *** markvoelker has quit IRC | 04:37 | |
| *** markvoelker has joined #openstack-defcore | 05:55 | |
| *** markvoelker has quit IRC | 06:00 | |
| *** alex_klimov has joined #openstack-defcore | 07:26 | |
| *** markvoelker has joined #openstack-defcore | 07:44 | |
| *** markvoelker has quit IRC | 07:49 | |
| *** markvoelker has joined #openstack-defcore | 09:33 | |
| *** markvoelker has quit IRC | 09:37 | |
| *** alex_klimov has quit IRC | 10:13 | |
| *** kbaikov has quit IRC | 10:28 | |
| *** kbaikov has joined #openstack-defcore | 10:35 | |
| *** alex_klimov has joined #openstack-defcore | 10:49 | |
| *** markvoelker has joined #openstack-defcore | 11:22 | |
| *** markvoelker has quit IRC | 11:27 | |
| *** fallenpegasus has quit IRC | 11:56 | |
| *** markvoelker has joined #openstack-defcore | 12:05 | |
| *** fallenpegasus has joined #openstack-defcore | 12:57 | |
| *** fallenpegasus has quit IRC | 13:02 | |
| *** alex_klimov has quit IRC | 13:54 | |
| *** VanL has joined #openstack-defcore | 14:01 | |
| *** VanL_ has joined #openstack-defcore | 14:02 | |
| *** VanL has quit IRC | 14:06 | |
| *** VanL_ has quit IRC | 14:17 | |
| *** VanL has joined #openstack-defcore | 14:20 | |
| *** VanL has quit IRC | 14:22 | |
| *** alex_klimov has joined #openstack-defcore | 14:25 | |
| *** galstrom_zzz is now known as galstrom | 14:31 | |
| *** fallenpegasus has joined #openstack-defcore | 14:45 | |
| *** fallenpegasus has quit IRC | 14:50 | |
| *** zehicle_irl has joined #openstack-defcore | 14:50 | |
| zehicle_irl | meeting starts in 8 mins on #openstack-meeting-4 | 14:53 |
|---|---|---|
| eglute | meeting started in openstack-meeting-4 | 15:03 |
| *** VanL has joined #openstack-defcore | 15:04 | |
| *** openstackgerrit has quit IRC | 15:05 | |
| *** openstackgerrit has joined #openstack-defcore | 15:05 | |
| openstackgerrit | Mark T. Voelker proposed openstack/defcore: Create a prelimary flag guidelines list https://review.openstack.org/188661 | 15:30 |
| *** VanL has quit IRC | 15:48 | |
| *** VanL has joined #openstack-defcore | 16:02 | |
| *** johnthetubaguy has joined #openstack-defcore | 16:03 | |
| *** VanL has quit IRC | 16:13 | |
| johnthetubaguy | hi, so is there anyone around to answer questions about the images stuff? | 16:15 |
| johnthetubaguy | bit confused by the questions from hogepodge last night | 16:16 |
| markvoelker | johnthetubaguy: I have to run, but I could probably dig up some context for you real quick if you need it | 16:16 |
| hogepodge | johnthetubaguy: I'm a bit concerned that there's a huge drift in glance v1/v2 support. | 16:16 |
| hogepodge | Like nova can proxy v1, but not v2, which is causing problems for testing. | 16:16 |
| * markvoelker lets hogepodge handle this and runs out for lunch meeting | 16:17 | |
| johnthetubaguy | hogepodge: so I don't get that implication arrow basically | 16:17 |
| hogepodge | Defcore is using the nova api to manage images, and does not have a hard requirement on image service | 16:17 |
| johnthetubaguy | the current state is basically Nova requires glance v1, we want glance v2 support, but no one has stepped up to complete the work that got half started in kilo | 16:17 |
| hogepodge | So implicitly v1 becomes required by defcore | 16:18 |
| johnthetubaguy | right | 16:18 |
| johnthetubaguy | so v2 was apparently "completed" at the end of kilo | 16:18 |
| hogepodge | it makes me think that v2 does not have the wide support that v1 has. | 16:18 |
| johnthetubaguy | I thought it was juno actually, but I got told otherwise | 16:18 |
| johnthetubaguy | right, so thats why Nova only works with glance v1 right now | 16:18 |
| hogepodge | which creates tensions, causing tests to be flagged and weaking our interop standards | 16:19 |
| johnthetubaguy | glance v2 doesn't yet have enough features for us to implement our images API on top of it, which is a problem | 16:19 |
| johnthetubaguy | hogepodge: so my point is basically, the issue is, glance doesn't have a stable API yet | 16:19 |
| johnthetubaguy | Nova requires glance v1, but that doesn't mean deployers can't also deploy glance v2 | 16:19 |
| johnthetubaguy | and the glance v1 API, can be internally only | 16:20 |
| hogepodge | that's good information to have, and I would argue that required v1 for interop is an ok thing based on that. So one of my aims is to help v2 have wider support. | 16:20 |
| johnthetubaguy | which is what it was designed to be | 16:20 |
| hogepodge | but it also means they can't drop v1 | 16:20 |
| hogepodge | which at least one vendor has done | 16:20 |
| johnthetubaguy | so, Nova will never drop our v1 proxy, we would like to, but that ship has sailed | 16:20 |
| johnthetubaguy | now Nova might be talking to glance v2, but our API will till be a glance v1 API | 16:21 |
| johnthetubaguy | (thats what this is taking time) | 16:21 |
| hogepodge | yeah, understand it's not an easy problem | 16:21 |
| johnthetubaguy | Nova has no plans to add a glance v2 proxy, and glance v2 was deigned so it could be exposed to end users | 16:21 |
| hogepodge | Ok, have to run, hard deadline for me. I need to start a ml thread about this. | 16:22 |
| johnthetubaguy | hogepodge: so basically the person working on it on it in kilo never got it finished, and now is too busy to work on it | 16:22 |
| johnthetubaguy | hogepodge: OK, no problems | 16:22 |
| hogepodge | johnthetubaguy: hm, :-( | 16:22 |
| hogepodge | johnthetubaguy: thanks for chatting with us about it. Really appreciate the input. This has been on my mind a lot over the last month and it's good to start talking about it | 16:23 |
| johnthetubaguy | now apparently we have some folks to take it over, but thats like a start over thing | 16:23 |
| johnthetubaguy | hogepodge: so there is a bigger issue | 16:23 |
| johnthetubaguy | hogepodge: glance v2 has two upload APIs and no way to discover which a deployment has enabled | 16:23 |
| eglute | johnthetubaguy hogepodge VanL definitely an e-mailing thread, or a gerrit issue, or both. we need to track this discussion | 16:24 |
| johnthetubaguy | so what I am saying is, Nova is not the blocker here, a lack of a stable API is a blocker | 16:24 |
| *** VanL has joined #openstack-defcore | 16:24 | |
| johnthetubaguy | eglute: +1 to the email thread | 16:24 |
| johnthetubaguy | just wanted to share context before hogepodge creates that thread | 16:24 |
| *** zehicle_irl has quit IRC | 16:24 | |
| eglute | so stable api- is that nova api or glance api | 16:24 |
| eglute | johnthetubaguy really appreciate you bringing this up | 16:25 |
| johnthetubaguy | the nova one is stable, but we don't really want people to use it | 16:25 |
| johnthetubaguy | its a proxy API, and can time out in nasty ways, best to talk to glance instead, it was an old stop gap thing that we can't ever delete | 16:25 |
| johnthetubaguy | although some deployers don't like you use the image upload, because it has some security issues (depending on who you ask) | 16:26 |
| johnthetubaguy | then you go to glance | 16:26 |
| *** VanL has quit IRC | 16:26 | |
| eglute | so... what would you suggest? | 16:27 |
| johnthetubaguy | well glance is currently buidling their v3 API, v1 should never have been exposed publicly, v1 is almost deprecated, but has been exposed by major public clouds, v2 isn't deployed many places and was only finished in kilo (or juno, its not clear), v2 doesn't have enough features to implement a v1 proxy due to some things that are not complete, its messy | 16:27 |
| johnthetubaguy | eglute: so for listing images, I think we can add Nova's image API, because there is no other option | 16:28 |
| johnthetubaguy | eglute: I think we remove uploading images from defcore until we know how to make that work in an interoperable way | 16:28 |
| johnthetubaguy | the first part really pains me to say that | 16:28 |
| * markvoelker reads scrollback | 16:28 | |
| eglute | johnthetubaguy can you start the email with all that info? | 16:28 |
| johnthetubaguy | I dearly want to delete that code, but it appears to be the only widely deployed thing | 16:29 |
| *** VanL has joined #openstack-defcore | 16:29 | |
| johnthetubaguy | eglute: so on the -dev ML with [defcore][glance][nova] tags? | 16:29 |
| eglute | something about being first, not best? | 16:29 |
| eglute | no | 16:29 |
| eglute | defcore-committee | 16:29 |
| johnthetubaguy | hmm, OK | 16:29 |
| eglute | and -dev as well i suppose | 16:29 |
| eglute | so, both? :) | 16:29 |
| * markvoelker sighs | 16:29 | |
| johnthetubaguy | well, thats usually not recommended... | 16:30 |
| johnthetubaguy | best to pick one place | 16:30 |
| markvoelker | johnthetubaguy: yes, this is a mess. It gets messier b/c several tempest tests use the Glance v1 API to upload images as part of setting up a test to test something else. | 16:30 |
| johnthetubaguy | thats the cool thing about using tags on the dev list, you can do a cross topic post more easily | 16:30 |
| johnthetubaguy | markvoelker: yeah, its total PITA | 16:30 |
| eglute | defcore is pretty small list i think, so should be ok i think... | 16:31 |
| markvoelker | ...and QA has said they really really don't want to refactor those tests.... | 16:31 |
| johnthetubaguy | markvoelker: I think configuring tempest with "pre loaded" image is a better way forward, in the short term | 16:31 |
| johnthetubaguy | markvoelker: so I am seeing defcore and tempest with different goals here, but thats another (quite painful) ML thread | 16:31 |
| markvoelker | johnthetubaguy: yup. | 16:32 |
| * markvoelker thinks this wouldn't be any fun if it was easy, right? | 16:32 | |
| johnthetubaguy | markvoelker: +1 | 16:32 |
| *** zehicle_irl has joined #openstack-defcore | 16:32 | |
| johnthetubaguy | so my take is, we need interop, we don't have it, so we should focus on identifying whats missing, while also keeping what we have working from getting any worse | 16:33 |
| johnthetubaguy | I suspect you could make that sound more positive, but its getting late in the day to attempt that... | 16:33 |
| *** fallenpegasus has joined #openstack-defcore | 16:33 | |
| *** openstackgerrit has quit IRC | 16:33 | |
| *** openstackgerrit has joined #openstack-defcore | 16:34 | |
| markvoelker | johnthetubaguy: +1. I've been thinking of starting up a scratchpad of "stuff that we really need to fix that we're not going to be able to fix very quickly" to present at Summits, TC meetings, whatever. | 16:34 |
| markvoelker | Just to get people aware of what we're finding. | 16:34 |
| johnthetubaguy | markvoelker: so I would love that to be flagged tests, or flags that don't point to a test | 16:34 |
| eglute | agree on the interop part.... | 16:35 |
| markvoelker | johnthetubaguy: I think it will inherently be in many cases | 16:35 |
| markvoelker | The image upload thing being one such case. | 16:35 |
| johnthetubaguy | true | 16:35 |
| markvoelker | But I think I'm looking for a way to call out the bigger-picture stuff more specifically. | 16:36 |
| markvoelker | The flagged tests lists will also contain stuff that's just short-term bugs, so gets a little TL;DR-ish | 16:36 |
| johnthetubaguy | yeah, I think focusing on some use cases, and what bits can't be tested yet, might be an interesting way of looking at these things | 16:37 |
| markvoelker | +1 | 16:37 |
| * markvoelker is being recalled to his delayed lunch meeting | 16:37 | |
| markvoelker | Thanks for the dicsussion johnthetubaguy | 16:38 |
| *** fallenpegasus has quit IRC | 16:38 | |
| johnthetubaguy | markvoelker: happy to get stuck in with some of this | 16:38 |
| johnthetubaguy | interop is super important | 16:38 |
| * eglute happy markvoelker and johnthetubaguy are part of defcore | 16:38 | |
| johnthetubaguy | as a project (Nova) we do try to care about it, but I like how defcore is helping keep that thinking honest | 16:39 |
| eglute | johnthetubaguy do you think you can come to the defcore mid-cycle? | 16:39 |
| johnthetubaguy | eglute: honestly, I don't think its likely, mostly due to how much transatlantic flights hurt me, I am thinking I could be at the end of a vidyo link though | 16:40 |
| johnthetubaguy | eglute: although there is a direct flight from London to austin now, and I have been looking for an excuse to try that | 16:40 |
| *** VanL has quit IRC | 16:40 | |
| eglute | ok, we will try to accommodate remote participation anyways, | 16:40 |
| eglute | i understand about flying! | 16:41 |
| johnthetubaguy | btw, did you folks see this blog about Nova's API | 16:47 |
| johnthetubaguy | https://dague.net/2015/06/05/the-nova-api-in-kilo-and-beyond-2/ | 16:47 |
| johnthetubaguy | its some good context on our focus on iterop | 16:47 |
| eglute | thanks johnthetubaguy reading it now | 16:49 |
| zehicle_irl | thanks - reviewing | 16:53 |
| johnthetubaguy | turns out I fail at email, just finishing off my half written email... | 16:58 |
| markvoelker | eglute: answering your earlier question, looks like I've just successfully shuffled the things that needed shuffling, so I should make it to Austin. | 17:01 |
| eglute | excellent, glad to hear it | 17:02 |
| *** alex_klimov has quit IRC | 17:07 | |
| *** fallenpegasus has joined #openstack-defcore | 17:15 | |
| markvoelker | eglute: looks like we have three +2's on https://review.openstack.org/#/c/190751/ | 17:19 |
| markvoelker | (thanks zehicle) | 17:19 |
| markvoelker | Could we merge that and I'll drop the -1's on Chris's patches that ran afoul of the old language? | 17:20 |
| *** zehicle_irl has quit IRC | 17:39 | |
| *** VanL has joined #openstack-defcore | 17:41 | |
| *** VanL has quit IRC | 17:46 | |
| hogepodge | johnthetubaguy: nice post, thanks | 17:50 |
| *** zehicle_irl has joined #openstack-defcore | 18:09 | |
| *** markvoelker has quit IRC | 18:25 | |
| *** markvoelker has joined #openstack-defcore | 18:25 | |
| *** markvoelker_ has joined #openstack-defcore | 18:47 | |
| *** markvoelker has quit IRC | 18:47 | |
| *** VanL has joined #openstack-defcore | 18:55 | |
| *** fallenpegasus has quit IRC | 18:55 | |
| *** zehicle_irl has quit IRC | 18:57 | |
| *** openstackgerrit has quit IRC | 19:16 | |
| *** openstackgerrit has joined #openstack-defcore | 19:16 | |
| openstackgerrit | Merged openstack/defcore: Re-introduce test removal clause https://review.openstack.org/190751 | 19:23 |
| *** frayedknot has quit IRC | 19:57 | |
| *** frayedknot has joined #openstack-defcore | 19:59 | |
| *** galstrom is now known as galstrom_zzz | 20:30 | |
| *** frayedknot has quit IRC | 20:51 | |
| *** frayedknot has joined #openstack-defcore | 20:53 | |
| *** VanL_ has joined #openstack-defcore | 20:57 | |
| *** VanL has quit IRC | 21:00 | |
| *** openstackgerrit has quit IRC | 21:24 | |
| *** openstackgerrit has joined #openstack-defcore | 21:24 | |
| *** alex_klimov has joined #openstack-defcore | 21:43 | |
| *** VanL has joined #openstack-defcore | 21:47 | |
| *** VanL_ has quit IRC | 21:51 | |
| *** VanL has quit IRC | 21:52 | |
| *** fallenpegasus has joined #openstack-defcore | 22:02 | |
| *** alex_klimov has quit IRC | 22:12 | |
| *** dwalleck has joined #openstack-defcore | 22:28 | |
| *** dwalleck has quit IRC | 22:30 | |
| *** dwalleck has joined #openstack-defcore | 22:30 | |
| *** fallenpegasus has quit IRC | 22:48 | |
| *** fallenpegasus has joined #openstack-defcore | 22:57 | |
| *** fallenpegasus has quit IRC | 23:23 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!