13:02:46 #startmeeting XCI-biweekly 13:02:46 Meeting started Wed Aug 28 13:02:46 2019 UTC. The chair is mbuil. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:02:46 Useful Commands: #action #agreed #help #info #idea #link #topic. 13:02:47 The meeting name has been set to 'xci_biweekly' 13:02:56 #link https://etherpad.opnfv.org/p/xci-meetings 13:03:01 #topic rollcall 13:03:07 #info Manuel Buil 13:03:41 #info Fatih Degirmenci 13:05:48 #info Nikos Mimigiannis 13:05:53 some people might have joined to zoom and found themselves in dovetail meeting 13:06:02 need to send update 13:06:06 * mbeierl raises hand 13:06:10 I did that :) 13:06:49 mbeierl: Sorry Mark! We had a zoom collision two weeks ago 13:07:05 and we realized that we always liked IRC meetings, so why not continuing doing that 13:07:12 Ya. I joined last week and it was Dovetail... 13:07:37 but it's all good! 13:07:47 I got to hear some Dovetail updates! 13:08:02 Didn't the Zoom meeting numbers change as well 13:08:26 yes, but we put the zoom number which has been used by Dovetail for a long tim 13:08:27 e 13:08:41 because i didn't see Dovetail meeting in Wiki meeting calendar and assumed it is free 13:09:15 That is one of the things that needs attention overall: Wiki cleanup 13:09:45 #topic Meeting day/time/occurrence 13:09:45 yes, there are probably lots of outdated info there 13:10:00 #link https://doodle.com/poll/6eswe8pbgdifmp7i 13:10:32 #info Results show a preference for 12h UTC. That means one hour earlier... is that fine for everyone? 13:10:49 not for US & Canada I suppose... 13:12:11 fdegir: yes, that's a good point 13:12:42 keeping the time as it is has 6 supporters against 8 who would like to have it one hour earlier 13:13:08 But for example I know Panos is moving to another company next week, so I am not sure whether he will participate 13:13:36 i can talk to Burak and Sarp 13:13:37 fdegir: have you talked to Sarp and Burak? 13:13:45 yes, please do that 13:13:52 but we need to have progress with the project proposal 13:14:02 otherwise i'm not sure how interested they would be 13:14:39 fdegir: they are only interested in participating if XCI is an official OPNFV project? 13:15:51 they will probably be interested in XCI no matter what but they don't know the history comparing to the others who signed up as committers 13:16:00 #info Panagiotis Karalis 13:16:13 as these guys are totally new to OPNFV and XCI 13:16:41 ok 13:16:59 so it will require another round of meetings and presentations to them 13:17:12 and explain why it would be valuable to have them in OPNFV 13:17:26 as one of them is from Turkcell who at the same time a member of CNTT 13:17:35 pkaralis: your vote in doodle is still the same even if you move between companies? 13:17:39 and the other is a totally new company entering the open source business 13:17:54 fdegir: ok, I understand. Please talk to them, let's see what they think 13:17:58 so the way things are with XCI is not good for OPNFV either since they sent 4 developers 13:18:05 they all signed the LF CLA 13:18:09 i mean the company 13:18:27 one of them is listed as committer but there are more people who are blocked due to the situation we are in 13:18:39 this is especially important given the situation with OPNFV's lack of developers 13:18:54 we do our best to promote OPNFV but get blocked due to unrelated stuff 13:19:19 and I think this is a good point to switch to the/new topic like Status of XCI Project Proposal or something 13:19:39 we have to bring this to closure soon - kill it or make it work 13:19:48 there is no point keeping things standing still 13:20:11 ok, let's move the time plan discussion then for next time, when things are clearer 13:20:23 no time plan, time slot sorry 13:20:47 mbuil, no, it seems that it needs an update from my side 13:20:58 #info we move the time slot discussion for next meeting, when things are clearer (situation of some people changed in the last weeks) 13:21:18 and also TF is waiting to see what's going to happen 13:21:21 #topic Status of XCI Project Proposal 13:21:21 this is such a mess 13:21:32 fdegir: you probably know more than me about this topic 13:21:56 well, this is open discussion so everyone can share their thoughts 13:22:11 let me info in the latest status and we chat 13:22:45 you have two choices: 1) bring it to TSC for project vote or 2) come back again to tech discuss meeting to get more input from community 13:23:41 #info The project proposal was presented to the community during weekly tech discuss meeting on 2019-06-10 and received comments 13:24:07 jento: i'm not sure what difference it would make if we bring it to tech discuss meeting 13:24:23 as mbuil pointed out, people do not share their comments during the 2 week review period 13:24:42 we get tons of comments during the meeting and the meeting ends without us having chance to respond to them 13:24:59 those comments could have been provided during the review period and perhaps addressed 13:25:04 or not 13:25:35 as i personally do not agree to majority of those comments but responding to them is the responsibility of the entire project team 13:25:49 From my perspective (as I was not participating in OPNFV as much as I should have been). I saw the proposal go out. I read the wiki for the proposal and thought it was good. I thought it was then going to go to TSC vote and was not aware that the tech discuss meeting was a gate 13:26:20 and tbh, what is in the proposal is not very different from what XCI has been doing as a subproject since its approval 13:26:30 the tech discuss meeting isn't a gate 13:26:38 the only difference is CNTT pieces and I'm not sure if we have to get approval from CNTT to implement their specs 13:26:59 as this is an open source project and we are doing community work 13:27:06 Do we have a summary of what the feedback was? 13:27:15 agreed 13:27:21 http://meetbot.opnfv.org/meetings/opnfv-meeting/2019/opnfv-meeting.2019-06-10-12.58.html 13:28:09 let's spend a lot of time discussing feedback, let's look at what to do moving forward 13:28:11 so suggestions 1, 2, and 3. That is all? 13:28:15 let's not spend 13:28:28 we are aware of Functest issues and working on fixing them but it has never been asked to any project's that were reviewed 13:28:41 what are the issues with Functest? 13:28:43 as everyone is aware of that passing functest is a release criteria 13:28:51 testing on baremetal 13:29:21 and we already stated that in project proposal that XCI will conform the release criteria which includes functest and baremetal 13:29:37 oh, functest requires testing on baremetal, or functest doesn't pass XCI's baremetal installation? 13:29:54 the stack installed by XCI toolchain doesn't pass Functest Smoke 13:30:12 ya, an installer cannot be expected to pass a release gate *before* becoming a project 13:31:22 i believe we are pretty close to make that work as people, especially mbuild has been putting tremendous effort to get things to work again 13:31:29 mbuil* 13:31:30 To clarify my understanding: functest uses its own OpenStack for its own sanity/gate checks, which gives confidence in the Functest results. Next installers can use this known good Functest to test their installation. 13:31:44 yes 13:31:49 gotcha 13:32:03 and functest itself is fine but the installer used by XCI fails to pass that on baremetal 13:32:18 re: 2nd point raised is about separating CICD pieces 13:32:24 ok, that is not an issue for project proposal. 13:32:40 well, XCI is about cross CI/CD and it uses a toolchain to achieve that since none of the installers supported XCI 13:32:42 sorry - baremetal being actively worked on does not stop the proposal 13:33:23 and XCI is perhaps one of the very early initiatives that raised the visibility in cross-CICD or e2e integration issue across open source 13:33:56 I must admit, I do not understand point 2 13:34:10 we are actually not after having an installer but we want to continue the effort to work with the communtiyies we have been working with 13:34:19 and this hasn't been an issue during the entire lifetime of XCI 13:34:24 until we proposed the project 13:34:59 finally, XCI has no vendor specific thing in it as it uses totally community developed toolchain 13:35:18 re 3rd point: XCI does not talk about CICD toolchain at all 13:35:30 can you clarify in the proposal that XCI is primarily about the CICD part and second abnout the installer part? 13:35:30 perhaps I need some education here. 13:35:40 although fdegir you might have just answered it 13:35:50 it is mainly about methodology, principles, and practices and it is totally agnostic to the CICD toolchain like Jenkins or Travis or whatever 13:35:53 clarifying that the installer part is open to any tool 13:36:00 I was going to ask: how does XCI provide CD by itself? 13:36:27 mbeierl: XCI does not provide CD but tries to enable CD way of working 13:36:48 and so, there is nothing to separate? 13:36:54 CD=Continuous Delivery (just to make sure it is not mixed with Continuous Deployment) 13:37:05 and in CD, people try to keep master branch working 13:37:18 so you can issue a release at any point in time 13:37:25 correct 13:37:43 rather than spending time to develop things, do big bang integrations late in release cycle and try to get things working again, you do all these things all the time 13:37:51 the part that I did not understand about XCI and the comment is - what part of XCI is "CD" specific such that it can be separated? 13:38:11 mbeierl: no part as we don't mention any specific tool or technology there 13:38:49 this sentence is what summarizes the CI/CD parts of the project: XCI project follows DevOps, Continous Integration (CI) and Continuous Delivery (CD) principles and best practices by applying gating, promotion, and confidence level mechanisms to ensure what is developed and tested always satisfies the criterias set forth by OPNFV Release Process in 13:38:49 collaboration with OPNFV Test Working Group. 13:39:00 so (mark says, somewhat tongue in cheek) XCI is already compliant with comment #2 13:39:30 yeah, you could say that 13:39:39 as the CI portion is already separated from the non-existent CD portion? 13:40:13 did you mean installer part is separated from CI/CD part? 13:40:50 let me re-read #2 then 13:41:34 hmm. nope. I still do not understand comment #2 after this discussion 13:42:13 georgk: so, XCI is not about the tools and technologies as it is agnostic to all those but more about the methodologu 13:42:43 I think that's exactly what I wanted to say 13:42:54 georgk: what matters is to deploy and test the stack we compose according to principles and practices that's set forth by the industry 13:43:25 so the current tolls XCI utilizes are bifrost and openstack-ansible 13:43:34 but the work is goi ng to introduce openstack-helm 13:43:36 deploy and test? Does that mean automatically running Functest? 13:43:55 so we are not bound to specific tools but rather the system under test 13:44:00 mbeierl: yes 13:44:10 and the selection of the toolchain is left to the community 13:44:22 is it possible to use XCI for just the 'installation' part and not run any post-install tools? 13:44:24 as XCI is a community effort 13:44:36 * mbeierl is trying to see if that can satisfy comment #2? 13:44:59 mbeierl: sorry but i don't understand what post-install tools are 13:45:21 post-installation Functest, Yardstick, Tempest, Rally, etc 13:45:41 use XCI as just a method to deploy latest OpenStack from source and then do nothing more 13:45:48 there is a sandbox which you can use to deploy OpenStack 13:46:11 i mean, the installation part is totally separate from how you run tests 13:46:17 tests are run by separate scripts/jobs 13:46:29 that use functest and yardstick in the end 13:46:31 ok. So... is that a possible answer to the comment? 13:46:48 i don't think it is 13:47:05 in my opinion, project should all put emphasis on CI/CD no matter what domain they work in 13:47:40 if it is not done that way, the result might not be as valuable as we can easily say "it works on my machine" 13:47:48 which is the exact reason why we should have CI/CD 13:47:54 ah - did not really read enough of comment #2. "make a separate proposal" 13:48:16 so, no that does not make sense to have 2 proposals 13:48:39 I think overall, that was the part that I was not understanding in this conversation. 13:48:52 so, what we propose is not about bifrost/osa/kubespray but actually have a project which works with the community to serve the OPNFV community 13:48:59 yes 13:49:20 whatever project wants to do with regards to toolchain must to be endorsed by the community and approved by TSC 13:49:35 so this is not a typical installer project as installers are free to do changes to the toolchain 13:49:47 so back to getting xci approved as a project... what are the next steps? 13:50:04 jento: i'm not so sure honestly 13:50:39 imo, you have two choices as I shared 13:50:46 as i summarized above, i am unable toaf 13:50:50 agree to comments 13:51:04 solve community misunderstanding or present xci as is to tsc 13:51:16 so going to tech meeting with same proposal might result in same comments and waste of everyone's time 13:52:44 i think i talked so much 13:52:51 yup, that could happen. but now we don't have a "no installer" problem as we did before 13:53:08 I asked too many questions? 13:53:11 :) 13:53:23 mbeierl: nope - questions are good as long as we're given chance to respond to them 13:53:32 keep'm coming :) 13:53:51 but what others think? 13:53:53 jento: Before moving this forward, we should also make sure we have enough developers. It wouldn't make sense to approve it and then not being able to progress on the things we are promising 13:53:54 what should we do? 13:54:21 mbuil: we have 14 people from 11 companies signed up as committers 13:54:26 mbuil: no different then any other project proposal in terms of develoeprs 13:54:44 if we assume half of those people will do something, it is still something 13:55:22 So, here is a possibility: ask to go back to tech meeting and we prepare a "here are the reasons we do not agree with the comments". If that does not pass, it should not matter. It is known that tech meeting is not a gate, and we can then go to TSC 13:55:50 i suggest you just go to TSC for vote 13:55:58 fdegir, jento: My experience is that most of the people are users, although some of them have not really do anything, so I am not sure. We could move this forward with this unkown of course 13:56:15 if that doesn't work (might be first time in OPNFV history), then come back to weekly tech discuss meeting 13:56:38 +1 from me re: straight to TSC 13:57:29 and the other concern that I have is that CNTT is classifying all projects in feature projects or installers. XCI is not exactly an installer, so that will create confusion too which should be addressed 13:57:36 your argument at TSC is XCI as a sub-project is not really different as a regular XCI project 13:58:37 mbuil: trying to solve CNTT problem before TSC approval doesn't make sense - we don't know what will come of CNTT 13:58:38 mbuil: if that happens - no development - than the law of open source takes care of it for us and it dies 13:59:05 another argument is that XCI is already active. Continuing to hide it as a sub-project does not serve the community in that it suppresses awareness 13:59:59 thanks a lot for sharing your thoughts 14:00:05 let's face it. XCI exists with or without the vote. 14:00:19 yes 14:00:21 have a good day all! 14:00:23 mbeierl: it aslways has ;) 14:00:27 always 14:00:49 we will go back to committers listed in the project and ask their opinion 14:00:58 fdegir: good idea 14:01:13 sounds good. thanks for discussion. 14:01:16 have a nice day all! 14:01:19 let's close the meeting then 14:01:23 thanks for coming 14:01:35 #endmeeting