15:00:02 <hogepodge> #startmeeting defcore 15:00:03 <openstack> Meeting started Wed Oct 14 15:00:02 2015 UTC and is due to finish in 60 minutes. The chair is hogepodge. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:04 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:08 <openstack> The meeting name has been set to 'defcore' 15:00:29 <hogepodge> Good morning everybody! 15:00:43 <markvoelker> o/ 15:01:11 <zehicle> o/ 15:01:13 <edmondsw> o/ 15:01:22 <hogepodge> #chair zehicle 15:01:23 <openstack> Current chairs: hogepodge zehicle 15:02:01 <hogepodge> While we wait for roll call, please take a moment to review the agenda 15:02:13 <zehicle> #link https://etherpad.openstack.org/p/DefCoreFlag.19 15:02:14 <hogepodge> #link https://etherpad.openstack.org/p/DefCoreFlag.19 15:02:46 <zehicle> eglute, will miss today 15:04:15 <hogepodge> #topic Flag Removed Tests 15:04:22 <hogepodge> #link https://review.openstack.org/#/c/229177/ 15:04:51 <catherine1> o/ 15:05:01 <hogepodge> I'm still working on the script. I'll probably have it done in Tokyo. Lots to get done before the summit. 15:05:41 <markvoelker> So on this one, I think we'd basically talked about two ways to handle it: a script and an alias field. The former is still useful even if the latter is used. 15:05:56 <markvoelker> #link https://review.openstack.org/#/c/232685/1 alias field 15:06:17 <markvoelker> The bigger question I think is just how to handle this for the current Guidelines, and for that the script is probably best. 15:06:38 <hogepodge> I think we should move quickly to add an alias field. My preference is that it be optional to make it easy to get it started. 15:06:49 <markvoelker> hogepodge: +1 15:06:52 <catherine1> can we add the alias to current guideline (2015.07)? 15:07:33 <markvoelker> catherine1: Hmm. It would require a schema change to a Board-approved doc. Not impossible I guess, but... 15:07:38 <hogepodge> catherine1: if it's optional, it should be an easy enough change. We might need board approval? We're not changing the content of the guideline, just the administration of it. 15:07:53 <catherine1> we can work on RefStack to handle it .. then we can lift the requirement of certain version of Tempest 15:08:07 <hogepodge> board approval could happen in a couple of weeks 15:08:07 <zehicle> we can ask for board votes on those 15:08:19 <zehicle> they are not substantive and easy for the board to approve 15:08:21 <zehicle> even by email 15:08:32 * markvoelker will sign up to do the legwork of reformatting 2016.next in 1.4 schema FWIW 15:09:00 <zehicle> if we think schema tweaks are needed that don't change the contents of the doc then we should move ahead and ask for them 15:09:17 <catherine1> alias block should be optional just like flagged test block 15:09:18 <zehicle> do we want 1.4 for the 2016.01 spec? 15:09:33 <hogepodge> So, get a 1.4 schema merged then send to board via e-mail for application to 2015.07? 15:09:33 <zehicle> I was expecting to cut the 2016.01 draft after this meeting 15:10:00 <hogepodge> zehicle: I think 1.4 has important improvements that we want sooner rather than later. 15:10:02 <zehicle> for 2015.07, we can create a patch and then ask for board approval before we merge 15:10:17 <zehicle> if the patch is ready for the board meeting, we'll just add it to the agenda 15:10:23 <markvoelker> zehicle: if we don't use 1.4 for the 2016.01 draft, we don't get the target_approval field, have the 2-sources-of-truth problem for status, and don't get aliases. I think we should do it. 15:10:46 <zehicle> +1 15:11:07 <catherine1> +1 for adding alias to all (2015.07 and future json ..) 15:11:08 <zehicle> to help w/ merge conflicts, I think we should resolve the new capabilities 15:11:13 <hogepodge> Ok, so what's the plan for getting it ready? 15:11:37 <markvoelker> I think the plan goes like this: 15:11:47 <hogepodge> zehicle: agreed, that's the next item on the agenda. Once we're good to merge I can start the process and do the reconciliation. 15:11:52 <zehicle> if the block is not required in the schema, then merges won't be a problem 15:12:03 <zehicle> kk 15:12:08 <markvoelker> First, we finalize schema 1.4 (see patch above plus https://review.openstack.org/#/c/224868/ ) 15:12:16 <zehicle> I was hoping to close 2015B first 15:12:26 <markvoelker> At the same time we can finish up the existing patches for scoring (next agenda items) 15:12:28 <zehicle> I'm OK to merge it 15:12:44 <catherine1> the block is not require in the schema .. just require in the JSON 15:12:46 <markvoelker> Then once both of those are done I'll reformat 2016.next to 1.4 and submit that 15:13:01 <catherine1> just not require in the JSON 15:13:16 <zehicle> merged 15:15:21 <catherine1> markvoelker: do you see my comment on the 1.4 patch? 15:16:01 <catherine1> about what should be put in the test vs alias areas .. 15:16:19 <hogepodge> Ok, so we've resolved the testing and alias field. 15:16:32 <markvoelker> catherine1: the one saying refstack could handle it if alias fields weren't included in all test blocks? Yes, and that sounds great. 15:16:49 <hogepodge> And we're good on the process changes? 15:17:05 <hogepodge> markvoelker: catherine1: yes, it's going to make everyone's lives easier if alias is optional. 15:17:07 <zehicle> I tried to address the two outstanding 15:17:24 <zehicle> basically, we're saying that the Foundation will collect data but leaving it open about what that is 15:17:45 <zehicle> hogepodge, that's a new topic? 15:17:58 <hogepodge> #topic process changes 15:18:13 <hogepodge> #link https://review.openstack.org/#/c/207209/7 15:18:13 <zehicle> #link https://review.openstack.org/#/c/207209/ 15:18:31 <zehicle> basically, we're saying that the Foundation will collect data but leaving it open about what that is 15:18:43 <hogepodge> I'm ok with the language on this, it's vague enough to give the Foundation some flexibility, but states that it's a requirement. 15:18:51 <zehicle> my understanding was that we could not agree on what would be required 15:19:05 <zehicle> since it's in hacking, we can tighten it as we learn more 15:19:06 <hogepodge> We'll be up front about what we want and why. 15:20:10 <hogepodge> And it will likely be reflected in the marketplace listings. It's something we need to work out with the marketing and web team, what system information we think is valable to show to consumers and users. 15:20:22 <hogepodge> markvoelker: I know you had concerns about scope and reasoning 15:20:56 <markvoelker> I kinda think we should just drop the HACKING change and leave the 2015B change. HACKING seems like the wrong place for this to begin with and if we're punting to the Foundation to decide what they want and the format they want it in, we should just let them put that on the interop page. 15:21:13 <zehicle> the idea of putting those parts in hacking was to allow the committee to adjust 15:21:31 <zehicle> the key element right now is getting it into the process for board approval 15:21:51 <markvoelker> zehicle: that's why I think we should just keep the 2015B part of the change and drop the rest 15:22:20 <hogepodge> zehicle: I tend to agree with Mark that being overly specific at the beginning might bite us in the end. I'd prefer to take the list to the Foundation and make sure it's reflected on the Interop page. 15:22:21 <zehicle> could do - the hacking file now reads as a suggestion list only 15:22:30 <markvoelker> The 2015B change says the Foundation may require vendors to submit details and specify a format to submit them in, so the HACKING stuff seems unnecessary 15:22:53 <markvoelker> (and, FWIW, controversial since it still contains a lot of the stuff that was controversial in earlier patchsets) 15:23:06 <zehicle> I'm ok w/ that. at this point the hacking stuff is noise w/o teeth 15:23:52 <zehicle> IMHO, less guidance is not more transparent. 15:24:05 <zehicle> but we can leave it to the foundation to work 15:24:16 <hogepodge> #action zehicle to update "hardware reporting" patch to drop HACKING requirements 15:24:18 * zehicle updates the patch 15:24:24 <markvoelker> Ok, so drop the HACKING change? I'd also suggest a slight rewording of the 2015B change since the two additions say similar things 15:24:30 * markvoelker adds that to gerrit 15:24:35 <hogepodge> #action hogepodge to take hacking list to Foundation for review and update Interop page 15:26:09 <hogepodge> #link next process review https://review.openstack.org/#/c/207205/ 15:26:28 <SamD> FWIW dwalleck and I should be able to happily submit configuration info on our next Rax certification run as long as we keep it general (I.E. ranges of nodes, etc...)... 15:27:04 <hogepodge> SamD: how about hypervisor? I think it's public information anyway 15:27:10 <markvoelker> hogepodge: on this one, can someone clarify what A5(3) means? It's a bit confusing. 15:27:38 <SamD> Hypervisor should be fine as well but I'm not 100% certain. :-) 15:28:43 <SamD> Things like Hypervisor, node range, cinder backend etc, should be fine from our point of view... 15:28:45 <markvoelker> E.g. is it saying that if the Foundation recommends a component be included in Platform, we apply criteria based on the audience for the component? 15:29:16 * zehicle requests moving on 15:29:37 <markvoelker> I'm also not quite sure what the last sentence of A4(7) means..."components will be evaluated based on their use in the Platform". 15:30:10 <zehicle> sorry - I see we're on the second change 15:30:42 <hogepodge> SamD: we can meet at the Summit and talk over some of those issues. We have no interest in revealing too much special infra sauce, just helping figure out capabilities. 15:31:00 <hogepodge> zehicle: yeah, working on language clarification 15:31:40 <zehicle> markvoelker, this reflects a discussion from the midcycle where we wanted to make sure Components were scored based on a smaller "widely deployed" benchmark 15:32:18 <SamD> hogepodge: Yes. Sounds like a plan. 15:32:41 <markvoelker> zehicle: sure, but what does it mean that "Components will be evaluated based on their use in the PLatform"? 15:34:52 <zehicle> When we look at pulling a Component into the Platform, then we need to see how the COMPONENT is used 15:35:01 <zehicle> not just the capabilities 15:35:32 <scotticus> zehicle what does component mean? example wise? 15:35:52 <zehicle> Component has a specific meeting for DefCore. 15:36:00 <zehicle> It's a collection of Capabilities 15:36:11 <hogepodge> zehicle: so maybe change to "deployed" or "in-field" usage? 15:36:28 <markvoelker> scotticus: http://git.openstack.org/cgit/openstack/defcore/tree/doc/source/process/Lexicon.rst#n32 15:36:46 <scotticus> sweet, a glossary 15:37:03 <zehicle> deployed matches the vocabulary that markvoelker asked us to use 15:37:12 <hogepodge> markvoelker: what could we do to clarify that language? 15:37:13 <markvoelker> scotticus: see also http://www.openstack.org/brand/interop/....basically, the components we have today are Compute and Object Storage 15:37:48 <edmondsw> are other components being evaluated? 15:37:49 <markvoelker> hogepodge: I can suggest some wording in gerrit, but I wanted to understand what the intent was in case my perception was wrong 15:38:10 <zehicle> markvoelker, is the intent clear then? 15:38:26 <hogepodge> Components roughly correspond with marks. So storage with "OpenStack Powered Storage", compute with "OpenStack Powered Compute", and compute and storage with "OpenStack Powered Platform" 15:38:55 <hogepodge> one could image a database component that corresponded to "OpenStack Powered Database" 15:38:57 <markvoelker> zehicle: I think so...how about I propose a change in gerrit after the meeting and you can tell me if I understood it correctly? =) 15:39:01 <zehicle> they directly correspond 15:39:08 <zehicle> +1 15:39:18 <hogepodge> zehicle: it's not one to one is what I was hitting at. :-D 15:39:18 <markvoelker> Ok, 20m to go...suggest we move on 15:39:22 <zehicle> markvoelker, my question was if more discussion was needed 15:39:39 <hogepodge> #action markvoelker to suggest wording for policy change 15:39:54 <markvoelker> zehicle: I think it's sufficient to carry further conversation in gerrit at this point. Thanks! 15:40:16 <hogepodge> What's our timeline for getting a final thumbs up/down? 15:40:28 <hogepodge> zehicle: when do you need it for the board? 15:40:37 <markvoelker> hogepodge: I can put my suggestion in today FWIW. 15:40:51 <zehicle> Friday 15:40:59 <zehicle> needs to have time for board review 15:41:22 <hogepodge> Ok, so let's plan to work the next two days to have all of the new reviews ready to go or drop. 15:41:23 <zehicle> We can still make patches to the 2015B draft after the merges 15:41:30 <hogepodge> Next topic 15:41:38 <hogepodge> #topic Finalize Advisory for Tokyo Board Meeting 15:41:53 <zehicle> you skipped 2016.01 create 15:42:07 <zehicle> related topic anyway 15:42:09 <hogepodge> #topic Plan for 2016.01 Creationroll 15:42:32 <hogepodge> zehicle: you have the stage 15:43:08 <zehicle> I'd like to create the 2016.01 Guideline ASAP so that we can collect comments on it 15:43:21 <zehicle> plan would be to use the .next file as is 15:43:34 <zehicle> then submit patches to remove non-consensus items 15:44:19 <markvoelker> zehicle: so to do that we need to finish the next three reviews on the agenda, yes? 15:44:19 <zehicle> ideally, that means we'd need to merge the items in process AND 1.4 schema changes 15:44:28 <hogepodge> I'd like to get the items from the three patches in the next topic in place. 15:44:50 <zehicle> markvoelker, yes HOWEVER, there's room to pass some items that are under discussion with the plan to remove them from 2016.01 15:45:14 <hogepodge> It looks like identity is stable. Some small changes to block storage, and name changes to images 15:45:38 <zehicle> so, I'd recommend that we can reduce some of the angst about hot items 15:45:51 <markvoelker> zehicle: sure. Could you clarify one thing for me? What are we presenting the Board in Tokyo exactly? B/c the next Guideline isn't due for approval until 2016.01, right? 15:45:57 <zehicle> if we are taking them out of 2016.01 and allowing disucssion to continue on .NET 15:46:01 <hogepodge> I removed one cinder item because of the proposed one item rule, but think we should pull it back. I'll send it up directly after the meeting. 15:46:08 <zehicle> NEXT 15:46:20 <markvoelker> E.g. w're not asking them to approve it until January, just advising on what we're planning to ask for approval on? 15:46:44 <zehicle> yes 15:47:01 <zehicle> we can remove from now till then but not add 15:47:16 <markvoelker> Ok, so essentially it's advisory until January. That's fine. 15:47:18 <markvoelker> Thanks 15:47:31 <hogepodge> #link cinder/identity https://review.openstack.org/#/c/221631/ 15:47:42 <hogepodge> #link glance/images https://review.openstack.org/#/c/213353/ 15:47:47 <zehicle> the status would flip from "draft" to "preliminary" I believe 15:48:01 <hogepodge> #link keystone/identity https://review.openstack.org/#/c/213330/ 15:48:18 * hogepodge notes error on first link, should read cinder/block storage 15:48:29 <zehicle> hogepodge, I'm done w/ that topic. I just wanted everyone to know that we were not doing final versions 15:48:35 <zehicle> and state timeline 15:48:38 <hogepodge> zehicle: ok 15:48:51 <hogepodge> #topic Final advisory for .next 15:49:40 <hogepodge> Are we good to start merging those new advisory capabilities. 15:49:58 <hogepodge> I want to send one more patchset up for Cinder, pulling one capability back in. 15:50:20 <markvoelker> hogepodge: there have been a couple iterations recently on those, but they're mostly in good shape. I'll peruse them one more time today, but I don't see why we couldn't land them before Friday. 15:50:25 <hogepodge> Once we're at consensus and zehicle and/or egle approves, I'd like to start merging and reconciling. 15:51:00 <hogepodge> With approval from chairs, I'd like authority to resolve conflicts once merging starts without +2 from chairs. 15:51:15 <zehicle> since we can continue to tweak after merge, I'd rather see if we can get agreement during the meeting 15:51:35 <zehicle> hogepodge, works for me. 15:52:22 * zehicle grants hogepodge approval 15:53:20 <hogepodge> The last change I want to make on cinder is to pull back copy-volume-to-image #link https://review.openstack.org/#/c/221631/11..12/2016.next.json,cm 15:53:54 <hogepodge> Aside from that I'm good to go. Anyone else have final comments on them? 15:53:59 <scotticus> you want it back in? 15:54:48 <scotticus> hogepodge ^? 15:55:03 <hogepodge> for now yes, it's not permanent, just advisory. If it goes out it's gone, but there is still open discussion on if it should be there or not. The other was dropped because it's an extension and not guaranteed to be there 15:55:04 <zehicle> we need to discuss apparently, but I would rather not see single test capabilities. Can't we put it into another Cap? 15:55:26 <hogepodge> zehicle: Yes, I'll merge image-volume interactions into one capability 15:55:33 <zehicle> hogepodge, thank you 15:55:44 <hogepodge> Can we continue on gerrit? With five minutes left I want to be sure to hit the next item. 15:55:49 <zehicle> yy 15:55:55 <markvoelker> hogepodge: ++ 15:55:55 <hogepodge> #topic Recurring Test Recommendation 15:55:58 <scotticus> i'm going to push back on the volume->image 15:57:03 <hogepodge> scotticus: understood. fwiw, I'm advocating for it now to keep discussion open, not to make it permanent. I also have reservations about the image<->volume caps 15:57:13 <hogepodge> markvoelker: you have the stage 15:57:13 <markvoelker> #link https://review.openstack.org/#/c/232128/ Recurring testing recommendations 15:57:38 <zehicle> this may be a topic for the summit. I think the intent make sense 15:57:41 <markvoelker> So a couple weeks ago we realized there were some major inconsistencies between DefCore's language and what's actually in the Foundation legal contracts 15:57:57 <markvoelker> Chris is working on changes, one of which dealt with recurring testing 15:58:11 <markvoelker> He asked us to put together some recommendations to the Foundation about how to handle that. 15:58:27 <markvoelker> I volunteered to put together a strawman to start that discussion, which is the link above. 15:58:49 <hogepodge> I'm not a fan of the four week window 15:59:01 <dwalleck> markvoelker: I'm really interested in this one. I'll have a look and add some feedback 15:59:25 <hogepodge> I prefer allowing vendors to test early if they want, otherwise must pass current tests within a year. 15:59:28 <markvoelker> hogepodge: totally fine and not unexpected. It's a conversation starter. So, definitely make suggestions in gerrit. =) 15:59:39 <hogepodge> yup, just send. :-D 15:59:41 <markvoelker> (seeing as how we have 1 minute) 15:59:48 <markvoelker> hogepodge: ty 15:59:54 <zehicle> I 16:00:09 <dwalleck> I'm already working on continuous testingDefCore for our cloud, but I'd be interesting in seeing what the concrete ask is 16:00:13 <zehicle> I'd like to get this on the summit discussion agenda. I had a suggestion too (not enough time to discuss here) 16:00:22 <hogepodge> Ok, we're at time. 16:00:30 <hogepodge> Let's move to #defcore and free the room. 16:00:33 <hogepodge> Thanks everyone! 16:00:40 <hogepodge> #endmeeting