15:59:48 #startmeeting defcore 15:59:49 Meeting started Wed Jun 29 15:59:48 2016 UTC and is due to finish in 60 minutes. The chair is markvoelker. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:59:50 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:59:53 The meeting name has been set to 'defcore' 16:00:07 #chair hogepodge 16:00:12 Current chairs: hogepodge markvoelker 16:00:16 o/ 16:00:24 Hi 16:00:26 #link https://etherpad.openstack.org/p/DefCoreLunar.9 Today's agenda 16:00:37 o/ 16:00:59 o/ 16:01:21 Hi gang, let's dive in... 16:01:28 #topic Board meeting recap 16:01:30 o/ 16:02:05 Yesterday was the most recent BoD meeting and we shared a few updates with them 16:02:14 #link https://docs.google.com/document/d/13yB9LbtFGh24lVo-KtwVBRXx7zYbvLDY_7EcQ1aAmos/edit?usp=sharing DefCore Board Report, June 2016 16:02:40 A few things to note: 16:03:27 Generally, the Board confirmed it's ok with nonsubstantive updates to approved Guideline docs. They'd like us to note them on future Board meeting agendas, but we likely don't need to discuss/reapprove them with the Board 16:04:14 o/ 16:04:31 On the additionalProperties issue, there was some discussion though not a lot in the way of conclusions 16:04:57 The next step for us is to review Chris's proposal and hash out what we think is a good way forward, then bring that to the Board next month 16:05:27 For those who need the link: 16:05:42 #link https://review.openstack.org/#/c/333067/ Testing waiver for vendors using Nova 2.0 API with additional properties 16:06:17 Please regard that as the official proposal, superseding anything on the mailing lists. 16:06:37 Any questions on Board stuff? 16:06:53 o/ 16:06:56 Otherwise, this is a good segue to our next topic, which is the proposal... 16:06:59 one thing i would add is there was support for a short term solution that unblocks the process while we work on a longer term solution. allison raised a concern but was ok when she understood that we were not talking about an unlimited timeframe for having an administrative solution 16:07:22 jbryce: thanks, good point 16:07:59 o/ 16:08:12 #topic Waiver for additional responses on Nova API calls proposal 16:08:18 That link again: 16:08:26 #link https://review.openstack.org/#/c/333067/ 16:08:48 A few updates on the original proposal and the changes you see there. 16:09:03 Now that we've had a lot of discussion on the ML's and in meetings and with the Board, it's time to really get the wheels turning on this. 16:09:32 I'd like to see DefCore folks prioritize getting this reviewed and weighing in on the patch this week if at all possible 16:09:44 Introducing a way to configure tempest upstream to ignore additional properties breaks encapsulation between tempest and tempest-lib 16:10:17 mtreinish wrote up the script in that patch to parse subunit output and generate a list of tests that failed because of additional properties, and report some data about that. 16:11:02 I need to update the patch to be clearer about how to use the script, but the process would be that a vendor would run the tests, then send over subunit data that could be processed with the script 16:11:31 hogepodge: I'll add this in the review, but we probably need to think really carefully about how we message that to vendors 16:11:59 E.g. I don't want somebody to just run Refstack following directions elsewhere, fail, and give up without even knowing about the program 16:12:03 hogepodge: so that script identifies the tests .. what is the action after that? 16:12:08 (assuming we implement it) 16:13:00 catherineD: I review the test results and determine if they pass with the additional properties, not too different from what happens now. It's more workload for me, and it's likely for reference I'll need to send the results to refstack since this is much more manual process 16:13:00 Anything else we need to discuss on this now? Questions folks have? 16:13:23 there’s been some good feedback on the review from mtreinish and ken’ichi but i think they are on board with overall goal behind it. and to me, effectively what it does is return the state of testing against the existing defcore specs to where they were before the tempest changes 16:14:09 hogepodge: so you will generate the JSON file to upload to RefStack? 16:14:38 catherineD: yes, which is what I've done for some clouds anyway (that have historically send me data as subunit files) 16:14:47 s/send/sent 16:15:58 o/ 16:15:59 hogepodge: we can look at the possibility for refstack-client to execute the script to identify those tests ... let's discuss more later 16:16:16 ++ 16:16:22 ++ 16:16:45 catherineD: thanks for that 16:16:53 catherineD: sure, that would be useful. I'm looking to unblock things for vendors, so I don't mind the extra labor while these things are possibly implemented 16:17:17 catherineD: thanks 16:17:32 Ok, our next topic is one more bit that's related to the additionalProperties issue, so let's cover that too 16:17:36 #topic Update next to explicitly allow additional responses on Nova 2.0 API 16:17:52 #link https://review.openstack.org/#/c/335247/ Update next to explicitly allow additional responses on Nova 2.0 API 16:18:02 Title pretty much says it all here. =) 16:18:37 So again, we need to make some traction soon on the waiver program proposals, and this is related to that. 16:18:47 Let's see if we can't get this one reviewed this week. 16:19:08 It officially updates the guideline to recognize that Nova 2.0 API has extensions and possibly additional properties, and encodes it in the designated sections. 16:19:29 #action everybody Please review https://review.openstack.org/#/c/333067/ and https://review.openstack.org/#/c/335247/ ASAP (at least this week) 16:19:51 Anything else to discuss on this one now? 16:19:53 do we need to update 2016.01 with https://review.openstack.org/#/c/335247/ ? 16:20:04 since that is what being tested today 16:20:21 knowing that 2016.01 is an approved Guideline 16:20:44 catherineD: I had the same thought and figured I'd add that comment to the patch once we make some headway with the waiver program. 16:20:59 markvoelker: ++ 16:20:59 catherineD: no, those are approved guidelines, and I think it would be inappropriate to change it. The waiver program is meant to hit 2016.01 and 2015.07. But if there's consensus and board approval I could see the benefit of it 16:21:09 But IMHO that can be a separate patch...getting it into next.json won't require Board approval, so I'm fine with doing them separately 16:21:49 I think the underlying issue is what we do about older versions that are still supported by vendors. The code hasn't changed, but the vendors can't control customers not upgrading. 16:21:52 I am just thinking that we allow those test to pass .. we need to allow those code too 16:23:01 hogepodge: I guess it might be a fair question to ask if we should ammend the older Guidelines with a note about the waiver program as one means of communicating it. 16:23:13 Particularly if we're going to ask the BoD to approve the waiver program anyway. 16:23:19 markvoelker: sure, that makes sense 16:23:19 But we can discuss that in the patch. 16:23:40 Ok, anything further here or shall we move on? 16:23:42 catherineD: yeah 16:23:46 we can move on 16:23:59 #topic DefCore Mid-cycle 16:24:08 VanL: around? 16:24:31 * markvoelker hears crickets 16:24:40 I would need to know sooner rather than later the actual location 16:24:44 since the trip is longish 16:24:53 the location is known! 16:24:58 eglute: it's not confirmed! 16:25:11 #chair eglute 16:25:12 Current chairs: eglute hogepodge markvoelker 16:25:16 eglute: VanL told us to hold until he could confirm :D 16:25:16 I thought it was confirmed that it will be at Rackspace in San Antonio? 16:25:28 Ok, VanL had an action item to confirm the facilities for the midcycle but seemed confident he could find somewhere to put us up. 16:25:37 I'll drop him a line this afternoon and see if we can get confirmation 16:25:56 markvoelker: and if he could recommend some hotel also that'd be rad 16:26:01 or eglute :D 16:26:07 Once I do, I'll send out a note the ML confirming so folks can make travel plans. I'll also post a first draft of the agenda. 16:26:19 markvoelker: +1 16:26:20 gema: Sure, I can probably help a bit with that too 16:26:28 sounds good, thanks 16:26:59 have we also decided on the duration? 2 or 3 days? 16:27:14 VanL says it is confirmed to be at the Castle via text message 16:27:25 For those flying, airport is SAT 16:27:26 eglute: cool, thanks 16:27:48 he has a room reserved for 30, for 3 days 16:27:57 eglute: thanks! 16:28:04 cathrineD: from a look at the topics on the agenda I'm planning for 3 days 16:28:16 Tues/Wed/Thur 16:28:31 the only request is that everyone take care of their own lunch 16:28:43 eglute: is there a canteen at the castle? 16:29:02 Something I'd like to consider is getting foundation legal involved with what is needed for data security compliance on refstack.openstack.org. What are our contractual obligations, and how does that translate into policies. 16:29:03 there are food trucks and or vendors selling food 16:29:08 sounds great 16:30:58 Rockyg: did you get that on the etherpad? I don't recall it being there when I started drafting the agenda, but if there's time I can see about working it in. 16:31:35 Didn't get it on the etherpad. Might just be to discuss what we think we might need from legal 16:31:48 ok 16:31:57 Anything else midcycle-related to discuss today? 16:32:13 Rockyg: Let's add that to the agenda for mid-cycle 16:32:38 #topic Status of Network Admin Test Refactoring 16:32:43 Location: 1 Fanatical Pl, San Antonio, TX 78218 16:32:45 ok. maybe an hour just to kick off how we want to move forward on this. 16:33:11 Cool name for the address and building. Hope they live up to the names 16:33:24 A quick refresher: for the next Guideline we wanted to include a bunch of networking capabilities, but it turns out many of those tests use admin creds 16:33:38 (mostly unnecessarily...the capabilities themselves are typically exposed to end users) 16:33:50 So we've been working to see if we can eliminate unnecessary use of admin creds 16:34:31 We also found a few that actually do really require admin rights that should be removed 16:34:43 ONe set of those are the extened attributes tests: 16:34:54 Do we have a list of the tests that need refactorring? 16:35:02 #link https://review.openstack.org/#/c/300608/ Flag network port tests that require admin credentials in 2016.01 16:35:10 Or we could add new tests that don't use admin creds? 16:35:46 I meant to do work on the refactoring, but was sidetracked. I'd like to return to the work and make the network test list as robust as possible 16:35:58 but we're running out of time, so the list helps 16:36:21 Rockyg: In most cases they're tests that use a base class that requires admin privs. So, no need for new tests, we just need to base them on a non-admin subclass. 16:36:39 I just did that for a bunch of the router tests: 16:36:51 #link https://review.openstack.org/#/c/335558/ Remove unnecessary admin credential usage 16:37:06 * markvoelker glances at zuul and sees passing results from CI so fra 16:37:59 So hogepodge and I still have a way to go, but looks like progress. I'm going to work some more on this Thursday/Friday. 16:38:04 We need to publicize which tests need work. There are resources available and we can coordinate with QA to ensure they get reviewed. 16:38:34 markvoelker: and anyone else who wants to work on this, can we create a work list so we don't duplicate effort? 16:38:49 markvoelker: are you target to add these tests to 2016.08? 16:39:07 catherineD: that's the goal in refactoring them 16:39:12 Sure. Please file bugs. =) 16:39:59 There's a complete list of tests affected in one of the patches we have outstanding, but I neglected to put that in the etherpad for today since it was covered in earlier meetings...I'll add it after we're done 16:41:32 On a semi-related note, during the last round of scoring we also noted that we'd like to start including GET / for services moving to microservices, but a few projects have no tests for that yet 16:41:59 I submitted one for Neutron, which has now landed. I'll do glance next, but the networking refactor has higher priority 16:42:20 markvoelker: can you update next to propose that capability? 16:42:45 hogepodge: I wasn't planning to this time around since we didn't propose it in our "firm" draft period. 16:42:56 markvoelker: ok 16:43:04 However I'll definitely add it for 2017 16:43:43 markvoelker: not even as advisory? Does it make sense to formally wait a year? Especially since the capability has existed for a while 16:44:25 I thought tests need to exist for 2 releases for DefCore candicates 16:44:53 catherineD: The capability needs to exist for 2 releases if I understand correctly. Tests can come and go. 16:44:54 hogepodge: we can discuss adding it, but we're a couple months away from the "solid draft published at summit 16:45:07 I don't think the tests need to exist. Just the capability 16:45:17 period now...which may mean less window for folks to raise objections/offer feedback 16:45:24 And the capability needs to have been adopted. 16:45:48 Rockyg: we also can't add capabilities that have no tests. Which is why this one wasn't in the draft. 16:46:05 No tests existed for that capability at all until last week. =) 16:46:46 * markvoelker glances at the clock 16:47:10 Ok, anything further here? We have a few more things to cover from the backlog today and 13 minutes in which to do it 16:47:47 #topic Active guideline updates to 1.5 schema 16:48:01 #link https://review.openstack.org/#/c/335251/ Active guideline updates to 1.5 schema 16:48:26 This is a formality, mostly. Just updates the schema entry to gate against the active guidelines (we're not gating yet) 16:48:49 One question, should we refer to the .rst schema file (which points to the .json) or the .json schema file? 16:48:53 So now that we've affirmed guidance from the Board about nonsubstantive changes to the approved Guidelines, we have a patch to move 215.07 and 2016.01 to schema 1.5 16:49:04 which allows us to have gate jobs checking them 16:49:38 hogepodge: the JSON is the canonical source of truth, the RST is just for human readability 16:49:43 (I think we can still gate validate even if we don't bump the version, but it's weird using a 1.5 schema to check a 1.4) 16:50:28 ++ 16:50:35 markvoelker: +1 schema file should be JSON 16:50:55 markvoelker: yeah, I was thinking that the rst is kinder to the reader, but if we want to be as formal as possible point to the json 16:51:04 I'll make the change to the patch 16:51:13 thanks 16:51:32 These updates are fairly trivial, so I think they'll be easy to review. If folks have a few minutes later today, please give them a look please. 16:52:02 Moving on... 16:52:15 #topic OS Network Extension Flags 16:52:27 #link https://review.openstack.org/#/c/329220/ OS Network Extension Flags 16:52:56 dwalleck proposed this a few weeks ago 16:53:12 Did we ever get a corresponding tempest bug filed? 16:53:47 I was hoping to get an update from them on this 16:54:06 Table for later since he doesn't seem to be around? I can follow up with him. 16:54:24 #action hogepodge to follow up with dwalleck on https://review.openstack.org/#/c/329220/ 16:54:30 Works for me. I was +1 on the patch awaiting a bug report being filed to go +2. 16:54:49 #topic Test Spec Proposal 16:54:59 #link https://review.openstack.org/#/c/317531/ Test spec proposal 16:55:12 need to do an update this week 16:55:18 otherwise comments still welcome 16:55:24 (may make my update better) 16:55:33 gema: thanks 16:55:35 yw 16:56:21 As a reminder, we'll be talking about this at the midcycle and I'd really like to see if we can finalize it, so it would be good to read this over now to prepare for discussion 16:56:25 (if you haven't already) 16:57:01 Ok, we're down to a couple minutes left.... 16:57:09 markvoelker: absolutely 16:57:13 let's finish it :D 16:57:34 A quick reminder: the CFP for Barcelona is open now for another week or two. Get those talks in if you haven't yet. =) 16:58:20 Another: please review the gerrit backlog and see what you can knock out this week 16:58:27 Any final comments before we close up? 16:58:42 Just about the backlog. Please review. 16:59:15 Ok folks, see you next week! Back to #openstack-defcore 16:59:19 #endmeeting