16:00:45 #startmeeting defcore 16:00:46 Meeting started Wed Oct 5 16:00:45 2016 UTC and is due to finish in 60 minutes. The chair is markvoelker. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:48 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:49 o/ 16:00:50 The meeting name has been set to 'defcore' 16:00:54 o/ 16:00:57 #link https://etherpad.openstack.org/p/DefCoreLunar.21 Today's Agenda 16:01:01 Hi everyone 16:01:06 #chair eglute_s hogepodge 16:01:07 Current chairs: eglute_s hogepodge markvoelker 16:01:20 o/ 16:01:23 o/ 16:01:24 o/ 16:01:40 o/ 16:02:04 Hi everyone. Just a few weeks until Barcelona left! Let's get to it... 16:02:08 should we be in 20 not 21? 16:02:27 o/ 16:02:35 eglute: 20 was last week, no? 16:02:45 i thought it was 19 last week 16:02:59 Heh...both pads show the same date 16:03:18 i created 20 yesterday but didnt change the date 16:03:21 my bad 16:03:27 Ah, ok then. 16:03:44 #link https://etherpad.openstack.org/p/DefCoreLunar.20 Today's Agenda (for real this time) 16:04:14 #topic 2017.01 Guideline 16:04:29 If folks could take a quick moment to update their status on the etherpad there please... 16:04:58 done... 16:05:12 I would like to talk about Ceilometer! 16:05:26 eglute_s: the floor is yours 16:05:43 I talked to the PTL of telemetry, and as my notes indicate, there are a couple issues 16:05:56 the major one is that they are deprecating Ceilometer API 16:06:09 #link http://lists.openstack.org/pipermail/openstack-dev/2016-October/105042.html 16:06:21 yup. That's why it's interesting 16:06:30 eglute_s: has that been decided? I lost the thread yesterday in a barage of meetings and haven't caught up yet. 16:06:31 if there is no API for it, I don't think we can include it in defcore 16:06:49 the PTL sounded pretty sure that it has been decided 16:07:26 but, reading the email, seems like it might be still up for discussion 16:08:16 Ok. The basic idea there was that Ceilometer had basically been decomopsed into Aodh/Panko/Gnocchi. Those are currently not tc-approved-release. 16:08:28 they've been talking abour it for a while 16:08:54 Was the PTL aware that we can't consider projects that aren't tc-approved-release? Are they interested in getting the decomposed projects tagged? 16:09:14 i told him we cant consider it if there are no APIs 16:09:55 he didnt seem to be interested in defcore, though he was very helpful 16:10:11 Even with the new APIs ... we still need to wait for adoption and tests to mature? 16:10:11 My take was that it is still up for discussion after heat auto-scaling was brought up 16:10:14 additionally, ceilometer's tests are in tempest plugin 16:10:46 shamail i do hope it will be discussed more 16:10:50 I think they are more focused on getting everything lined up and in good shape without ceilometer 16:11:07 though at this point, since it's future is uncertain, i do not think we should push it forward with scoring 16:11:35 I'm not so worried about plugins, as we can work on moving tests over if they're good candidates. 16:11:36 don't think it is a good candidate for interop if the project itself does not seem to be stable 16:11:41 good to wait until at least next cycle 16:11:52 catherine_d|1, Rockyg i agree 16:12:19 they are doing good and importannt work, but it's still midflight 16:12:20 eglute_s: Agreed, but I think we should start evaluating a future path. E.g. does it make sense to ask for tc-approved-release on some or all of the decomp projects, can tests for them be moved to tempest, etc. 16:12:26 hogepodge PTL was willing to work on moving the tests, so not too much of an issue 16:12:42 The answer to all of those might be "no", but we should start that discussion sooner rather than later. =) 16:13:29 markvoelker i think we should wait for the projects to settle. right now there is too much change going on there in my opinion... 16:13:32 * markvoelker is also curious if heat autoscaling is going to eventually move to senlin, but that's another topic 16:14:17 for the decomp projects, i think we need to wait to see how the adoption goes 16:14:54 Are Aodh/Gnocchi/Panko listed in the user survey this time around? 16:14:55 and what the deprecated API will be replaced with, if it is in fact, deprecated. 16:15:12 At last gnocchi and aodh are getting good adoption with other opnstack projcts 16:16:59 So, I think there's a couple of things here then. 16:17:26 yup. 16:17:30 First, let's see if we can talk to the user survey folks about making sure Aodh/Gnocchi/Panko get listed (so we can guage adoption down the line) 16:17:49 i am looking at the survey now and only see Ceilometer 16:17:53 Second, obviously let's follow the decision around the old Ceilometer API 16:18:46 Third, once that decision has been reached we should think about future potential paths for Ceilometer. 16:19:02 anything on telemetry? 16:19:14 Rockyg no telemetry either. 16:19:24 E.g. it'd be a shame if they decide next week that Ceilometer goes away and Gnocchi/Aodh/Panko are the future, but nobody even starts work on porting tests or whatnot for six more months 16:19:59 Make sense? 16:20:04 ++ 16:20:04 markvoelker yes 16:20:09 * gema sneaks out for dinner, will read logs later o/ 16:20:35 What would be nice is a telemetry api and all the projects could plug in to that. Would make our jobs easier. 16:20:47 hogepodge ++ 16:20:48 agree 16:20:50 Ok. So eglute_s would you be able to handle those AI's for the moment? 16:21:23 markvoelker yes, i don't think anything is immediate though 16:21:46 eglute_s: sure, mostly keeping abreast of things at the moment 16:22:01 i guess user survey is more immediate. i will send some emails :) 16:22:22 Sounds good. Thanks! 16:22:37 #action eglute_s to follow up with user survey folks about telemetry projects 16:22:47 (speaking of user survey, newest data was just released) 16:23:07 #action eglute_s will keep abreast of the Ceilometer API deprecation discussion 16:23:29 Ok, further discussion on Ceilometer? Or shall we move on? 16:23:32 hogepodge do you have a link? 16:23:38 move on! 16:24:05 shamail: around? 16:24:20 Here! 16:24:32 Anything to discuss on Cinder? 16:25:25 Not much yet, I got guidance from yourself and hodgepodge on how to start with scoring... I will be reviewing third party sources tonight. I have a meeting scheduled for Friday with the PTL 16:25:34 Ok, thanks 16:25:48 I think I bit off more than I can chew for the first time helping but I'll get through it :) 16:26:02 For those not aware, Cinder may be a little interesting too...they revved the CURRENT API from v2 to v3 in Mitaka. 16:26:45 (functionally the same, but with microversions...but v3 was completely undocumented until a few weeks ago, so adoption possibly not great right now...) 16:27:06 Anything else on Cinder? 16:27:22 Not at the moment 16:27:40 Ok, next up: Glance 16:28:05 markvoelker: the current Cinder tests in DefCore guideline is based on v2 right? 16:28:12 catherine_d|1: right 16:28:14 I've spoken with the PTL and should have a patch up today-ish. Got delayed last week by a ridiculous pile of pre-summit stuff that's piling up. 16:28:34 if it's just a microversion with guaranteed backwards compatibility, it may not be a problem 16:28:47 Barring objections I'll probably also be submitting a governance patch to add glance_store as tc-approved-release 16:28:54 I would imagine there will be a conf parameter to set v2 vs v3 just like Keystone in tempest.conf 16:29:00 And then consider adding the non-driver bits of it as designated sections 16:29:06 same for cindr except for possible additions 16:29:37 * markvoelker switches back to Cinder for a moment since it seems I moved on too quickly 16:30:17 So, one thing to bear in mind with Cinder is that 2017.01 will include Liberty (in whcih v3 didn't exist) 16:30:25 it tool us a cycle or two to drop v2 Keystone tests from the must-pass list .. I guess we will do the same for Cinder? 16:30:41 tool --> took 16:31:08 So basically we'll need to consider how that transition will work and when/if we switch guidelines to v3 16:31:57 * catherine_d|1 API version change really does not help Interop 16:31:57 * markvoelker sees he and catherine_d|1 are thinking along similar lines here 16:32:37 yup. api revs are bears 16:32:42 Also worth noting: there's nothing that says both API's can't be required in the same Guideline, so it's not like there has to be an either/or. v2 is not DEPRECATED, it's SUPPORTED 16:33:06 exactly 16:33:14 But I'll leave it to our Cinder point folks (hi shamail_ !) to come up with proposals to discuss. 16:33:37 I'm not sur we've even managd the nova tests for that all that wll. 16:34:03 OK, anything else on Cinder? 16:34:05 We've just let th nova team gnerate the tests as they see fit 16:34:08 * markvoelker pauses a little longer this time 16:34:24 it appears that v3 will be backwards compatible with v2 because microversions 16:34:35 (confirmed with cinder devs) 16:34:48 I'll confirm when I meet 16:34:53 Ah ok 16:34:54 so this should not be a major issue for us. v3 should just work with interop tests 16:34:58 hogepodge: that is good news .. 16:35:29 (that's assuming the host api supports an early enough microversion, but that's the case with all microversion apis) 16:36:36 Move on? 16:36:45 yes 16:36:47 yeah 16:37:00 Ok, back to Glance: see above. Any questions there? 16:38:03 looks good to me 16:38:07 Hearing none, let's move on to Heat. 16:38:16 catherine_d|1: what's the word? =) 16:38:50 I added a summit session request to the Heat team after contact the PTL 16:38:50 Sounds like we're mostly in the same boat as last time where we don't have good tests in Tempest? 16:39:18 yes 16:39:47 basically need the Heat team to put priority to move the tests to Tempest 16:41:05 that all for Heat .. 16:41:18 Ok. We may also want to discuss/think through whether the API alone is all we want to define capabilities for down the road. 16:41:23 Heat summit eitherpad https://etherpad.openstack.org/p/ocata-heat-sessions 16:41:46 Heat's a little different than some projects in that we might want to define capabilities that test support for various resources in templates, for example. 16:41:56 heat is kind of a funny case because the api doesn't do a whole lot, most of the action is in the templates with is heavily deployment dependent. 16:42:08 hogepodge: exactly. 16:43:08 I'll defer to catherine_d|1 on figuring out a proposal around how to deal with that though, since she's playing point and I have way too much on my hands as it is. =) 16:43:18 resources may not be capabilities that should be scope for interop ... but those ae discussion we can bring up 16:43:56 Ok, anything else on Heat? 16:43:59 it's the case for scenario testing 16:44:05 nothing more from me 16:44:13 that is it from me 16:44:23 Ok, moving on then 16:44:27 Next up is Keystone 16:44:51 Last week hogepodge asked for some help here, and I've sent an email to the PTL and started (barely) digging in to look for capabilities 16:45:08 Not a lot to report here this week I guess, unless hogepodge has any updates 16:45:48 Any questions/comments on Keystone? 16:46:01 I don't, I can begin to look again next week. My backlog and lingering illness has buried me for the short term 16:46:22 * markvoelker hopes hogepodge feels better soon and will continue to work on this too 16:46:52 Ok, time's short...let's move on to nova 16:46:59 shamail_: that's you again 16:47:16 or shamail: whichever of you is actually Shamail today. =) 16:47:44 Looks like you're meeting with the PTL tomorrow? 16:48:01 Hi, Nova is same status as Cinder. I have a meeting scheduled with the PTL tomorrow. 16:48:20 Ok, thanks. Anything else? 16:48:52 I am going to be looking at sources tonight 16:48:52 Yes, that's the plan (and hopefully go in prepared) 16:49:36 Ok. If nothing further, let's move on to Neutron 16:49:46 which is me again 16:50:10 I spoke with the PTL and did some looking myself too. The TC approved adding neutron-lib to tc-approved-release 16:50:15 I have a scoring patch up here: 16:50:23 #link https://review.openstack.org/#/c/381393/ Neutron scoring patch 16:50:46 There's one additional capability I'm still looking at that's a bit more recent 16:50:57 But the existing patch there is pretty simple. 16:51:06 Have a look and feel free to start discussing anytime 16:51:22 Questions/comments on Neutron right now? 16:51:30 Not at the moment 16:51:35 not right now... 16:51:49 Ok then, on to Swift. eglute_s: that's you! 16:51:55 markvoelker should we wait to review your patch on neutron? 16:51:56 oh hai 16:52:08 eglute_s: nah, go ahead and start commenting. 16:52:10 notmyname did a lot of work on swift! 16:52:20 #link https://gist.github.com/notmyname/102e4aba7084598638f47cee47f62bb1 16:52:25 i have not had a chance to review it. 16:52:35 once I do, i will start scoring 16:52:55 i am sure i will need help with scoring 16:53:09 please work with me as you go through that. I'm happy to help clarify my intent, etc 16:53:27 there's some new things, yeah, but also som renames and better classifications for existing stuff 16:53:32 notmyname i will! and really appreciate your help 16:53:35 Cool. Sounds like you and notmyname are making good progress. 16:53:43 mostly notmyname yes, :) 16:53:51 When do you think we can expect to see a patch up? 16:54:19 hopefully by friday. I am swamped by some other things, so not sure how much time i will have today 16:54:42 so, as an asside, something that the defcore team could produce that would be *super* helpful would be a script that lists current capabilites and tests (i wrote my own) and then correlates that with what's available in tempest (I did that part by hand) 16:54:44 Ok, I'll keep an eye out. Remember, first rev doesn't need to be perfect, just needs to be enough to kickstart discussion in gerrit 16:55:08 notmyname can you share the scripts with us? 16:55:18 I think catherined|2 did something like that a while back maybe? 16:56:13 yea I did that a few cycle ago ... strictly by name ... not very accurate ... 16:56:15 this is all I have https://gist.github.com/notmyname/586fab93aab607d36b2bc6861b638afb 16:56:38 thank you notmyname 16:56:43 the .sh is run from inside a tempest directory 16:57:17 Ok gang, down to the last couple of minutes unfortunately 16:57:18 gives every test name found, what file it's in, and its idempotent id 16:57:32 notmyname that looks good! 16:57:44 then I used some nice features of sublime text to clean it up 16:58:07 thus resulting in the original gist i sent you (or that's what i used to do the capabilities correlation) 16:58:10 Certainly worth looking over for folks who are still test hunting 16:58:25 Two quick notes before we have to close for today 16:58:31 notmyname: I write that script all the time :-D 16:58:38 #topic Summit Planing 16:58:43 #link https://etherpad.openstack.org/p/DefCoreBarcelona Summit Planning pad 16:59:06 If you have any other topics, get them on that pad today. We'll start hammering out the agenda shortly. 16:59:24 And lastly 16:59:30 #topic RefStack Updates 16:59:45 #link https://etherpad.openstack.org/p/refstack-ocata-summit RefStack Planning pad 17:00:08 Please take note of the RefStack pad and add topics if necessary 17:00:13 And with that, we're all out of time 17:00:22 Over to #openstack-defcore 17:00:25 Thanks! 17:00:28 #endmeeting