17:00:57 <tdawson> #startmeeting CentOS PaaS SIG
17:00:57 <centbot> Meeting started Wed Aug  9 17:00:57 2017 UTC.  The chair is tdawson. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:57 <centbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:01:03 <tdawson> #topic roll call
17:01:06 * jdetiber waves
17:01:13 <tdawson> Hi jdetiber
17:01:32 * _ari_ here
17:01:36 <_ari_> barely ;)
17:02:01 <tdawson> Hi _ari_ ... we'll take what we can get
17:02:18 <jdetiber> Are any of us ever fully here? ;)
17:02:19 <_ari_> :)
17:05:15 * tdawson wishes sdodson were here today.
17:05:28 <sdodson> i'm here
17:05:42 <sdodson> now
17:05:42 <tdawson> #chair tdawson jdetiber sdodson _ari_
17:05:42 <centbot> Current chairs: _ari_ jdetiber sdodson tdawson
17:05:48 <tdawson> hi sdodson
17:06:00 <tdawson> I've got a question for you ... when we get to it in the meeting.
17:06:08 <tdawson> #topic OpenShift on CentOS Current Status
17:06:11 <tdawson> Starting with rpms
17:06:32 <tdawson> The 3.6 origin and openshift-ansible packages are in testing
17:06:40 <tdawson> They have been there through much of the past week.
17:07:03 <_ari_> tdawson: \o/
17:07:24 <tdawson> I don't know if others have been able to do much testing, but I tested origin cluster up, and atomic-openshift-installer install
17:07:32 <tdawson> origin cluster up worked
17:07:39 <sdodson> I did an install using them, no problems.
17:07:59 <tdawson> atomic-openshift-installer install didn't ask for the origin stuff ... that I noticed.
17:08:34 <tdawson> sdodson: Did we get the patch into openshift-ansible so we don't have to patch it for CentOS?
17:08:56 <sdodson> tdawson: for the quick installer, no, I've never gotten to that.
17:09:23 <sdodson> Our intention is actually to stop shipping the quick installer for 3.7.
17:10:01 <tdawson> sdodson: Ah, ok
17:10:50 <tdawson> sdodson: Actually, after going through the steps ... I'm not seeing anywhere that it asks for Atomic Openshift vs Origin
17:11:33 <tdawson> sdodson: I guess ... if people want, we could take the " atomic-openshift-installer install " off our quick setup instructions.
17:12:23 * tdawson notes that his testing is still going on during this meeting.
17:12:28 <sdodson> tdawson: I'm fine with that.
17:13:38 <tdawson> OK ... I'll at least mark it as depricated.
17:14:23 <tdawson> Anyway, I'll send out an email after this meeting, asking for people to reply if they've tested .... if we don't have any bad tests, I'll tag things as released tomorrow.
17:15:13 <sdodson> sounds good, thanks
17:15:27 <tdawson> We haven't had quite the pounding on the door to get 3.6 released like we did for 3.5 ... but I'd still like to get it released sooner rather than later.
17:15:49 <sdodson> i've had several people ask me about it
17:16:01 <tdawson> I think that's it for rpms ...
17:16:13 <tdawson> Moving on to Automated rpm building and Automated testing
17:16:24 <tdawson> _ari_: How has that progressed this week.
17:16:28 <_ari_> So we are building the candidates fine I fixed a bug
17:16:38 <_ari_> unfortunately the master branch is failing
17:16:53 <_ari_> I forwarded an error to you guys about CBS failures
17:17:07 <tdawson> _ari_: Yep ... I was only able to look at it very briefly
17:17:10 <_ari_> now I am seeing something else but wasn't sure what was happening
17:17:49 <tdawson> What was the something else?
17:17:55 <_ari_> I don't know how to bug cbs the only difference from master to a release is the version and we don't check out that version is that the issue
17:17:56 <jdetiber> _ari_: same here, I've been battling with some family-related things that have been taking a lot of my time lately
17:18:32 <_ari_> tdawson: last one was installing tito and build-deps
17:18:39 <_ari_> https://ci.centos.org/job/paas-bfs-future-matrix/PYTHON=system-CPython-2.7,TOPOLOGY=duffy_3node_cluster,nodes=paas-sig-ci-slave01/239/consoleFull
17:19:21 <_ari_> tdawson: I would say the cbs is the main issue
17:19:21 <tdawson> Oh, I wonder if build deps changed with 3.7 ...
17:19:39 <sdodson> It requires golang-1.8 I'm told
17:19:47 <jdetiber> tdawson: likely, golang-1.8 is the big change
17:19:50 <_ari_> tdawson: should I checkout 3.7 when doing the master branch I skip that step
17:20:01 <jdetiber> but there is also a new requirement for building the windows and darwin clients
17:20:09 <sdodson> the error in that build is that epel is screwed up
17:20:14 <_ari_> jdetiber: that should be handled by skuznets container
17:20:46 <jdetiber> _ari_: I believe that is the case now, but it wasn't a couple of weeks ago :)
17:20:49 <_ari_> sdodson: ah ok then that is a non issue the CBS one though not sure should I checkout the 3.7 branch
17:21:08 <jdetiber> _ari_: also, CBS wouldn't use the build container
17:21:53 <_ari_> jdetiber: well we build the source rpm from the build container
17:21:58 <_ari_> so doesn't it use that
17:22:39 <jdetiber> _ari_: not quite, the RPMs build natively within the mock environments provided by koji
17:23:11 <_ari_> jdetiber: ok then I don't know what is happening can anyone validate that it works manually from their environment
17:23:54 <tdawson> I can do that this week, this afternoon hopefully, unless jdetiber want's to.
17:24:22 <jdetiber> tdawson: If you have the time, I'll defer to you, I'm still trying to get caught up with dayjob :)
17:24:46 <tdawson> OK, I'll give it a try.
17:24:48 <_ari_> tdawson: that would be great thanks just ping me
17:25:11 <tdawson> I'm still rather concerned about the error from before (the one on cbs - https://cbs.centos.org/koji/getfile?taskID=203979&name=build.log&offset=-4000
17:25:53 <tdawson> Hopefully that was just a transitory bug in their script.
17:26:14 <jdetiber> tdawson: goversioninfo is the new binary needed for win/darwin clients
17:26:26 <tdawson> jdetiber: What provides it?
17:26:34 <jdetiber> tdawson: there is an internal build that is floating around now
17:26:42 <tdawson> jdetiber: Oh ... nifty
17:28:45 <tdawson> OK ... well ... this is where my not being part of anything official gives us the hurts ... I'll ping around so we can get this in cbs.
17:28:45 <_ari_> tdawson: on another front there is a general centos CI pipeline library we are working on https://github.com/CentOS/cico-pipeline-library/pull/6
17:29:04 <tdawson> _ari_: Oh, I heard of this last week ... how is that going?
17:29:34 <_ari_> tdawson: great other project can leverage it and this will be the basis for migration for the sig jobs
17:30:02 <tdawson> Cool
17:30:37 <_ari_> tdawson: that is it for automation
17:30:39 <tdawson> jdetiber: I found the version of goversioninfo, I'll get that into cbs so we can build.
17:30:49 <tdawson> _ari_: OK, thank you
17:31:15 <jdetiber> tdawson: awesome, thanks. I just found out about the build today, otherwise I would have built it when I was messing around with that test failure previously
17:31:18 <tdawson> _ari_: I'm sorry we haven't been as responsive the past week to help get these bugs worked out ... hopefully this next week we'll be better at that.
17:31:36 <tdawson> _ari_: But thank you for all the work you do on our automation.
17:31:48 <_ari_> tdawson: no worries we all get busy
17:32:00 <_ari_> tdawson: np cool stuff
17:32:05 <tdawson> So, now on to multi-arch
17:32:21 <tdawson> jdetiber: Any progress, or anything to report on multi-arch ?
17:32:26 <bstinson> i have an update on machinery
17:32:38 * jdetiber defers to bstinson for his update :)
17:33:07 <bstinson> i taught duffy how to do multiarch things this week, we're deploying that code tomorrow along with documentation and 36 ppc64 VMs ready for use
17:33:35 <tdawson> Very nice
17:34:16 <tdawson> bstinson: As much as I like ppc64, I know I'm going to get asked about this, is aarch64 in the pipeline too?
17:34:16 <jdetiber> bstinson: yay, I'm looking forward to playing around with it :)
17:34:36 <bstinson> tdawson: yes indeed, we have a ton of hardware for it that is waiting to be wired up
17:34:45 <tdawson> Cool
17:34:48 <jdetiber> heh, I wasn't going to say anything ;)
17:34:50 <bstinson> the hard part (teaching duffy new tricks) is done though so it will be a matter of just adding them
17:35:09 <jdetiber> bstinson: now we just have to teach CentOS how to SystemZ :)
17:35:12 <tdawson> That's great
17:35:39 * tdawson wonders if jdetiber is going to mention s390
17:35:40 <bstinson> jdetiber: i need to play with emulation myself soon
17:36:04 <jdetiber> tdawson: heh
17:36:38 <jdetiber> bstinson: I'll try to do some playing around there as well. I know user-mode emulation works for upstream k8s binary and container builds
17:37:09 <bstinson> jdetiber: once i get off this duffy stuff, i can work on finding a home in CI for a bit of the bootstrap work
17:37:28 <jdetiber> Other multi-arch updates. I've gotten further with internal pipeline testing, but have been battling through some unit test failures that didn't exist prior to the 1.7 rebase
17:37:50 <jdetiber> bstinson: thanks, let me know how I can help
17:39:29 <tdawson> Any other multi-arch stuff?
17:40:43 <tdawson> I'll take the silence for a no ... or else we're having network issues.
17:40:57 <jdetiber> tdawson: nothing else from me, I'm hoping now that duffy is up and running I can start replicating the internal pipeline work in centos ci :)
17:41:09 <jdetiber> nah, just multi-tasking issues :)
17:41:40 <tdawson> OK, moving on to Documentation
17:42:13 <tdawson> As said earlier, I'm going to change the quickstart to say that the openshift-ansible quick installer is going away on 3.8
17:42:29 <tdawson> err ... that's 3.7
17:42:30 <jdetiber> tdawson: 3.8 or 3.7?
17:42:30 <sdodson> 3.7
17:42:32 <sdodson> yeah
17:42:56 <tdawson> Once 3.6 is marked as release, I'll do my usually search through the docs to make sure nothing needs to be changed.
17:43:13 <tdawson> Last time we didn't have to do anything, so hopefully this time will be the same, which makes it quicker.
17:43:25 <tdawson> Anything else that needs to be documented?
17:43:45 <tdawson> Do we need to add a link to the mulit-arch documentation that bstinson was talking about earlier?
17:43:58 <jdetiber> sdodson: does anything need to be documented around the service broker?
17:44:00 <tdawson> That seems more generic, and not necessarily Paas specific.
17:44:07 <bstinson> tdawson: that's not published yet, but it will be in the release announcement
17:44:20 <bstinson> stay tuned to ci-users@centos.org and centos-devel@centos.org
17:44:23 <tdawson> bstinson: OK, I'll hold off
17:44:28 <jdetiber> tdawson: I don't see a need to add anything to the PaaS SIG until we start having multi-arch jobs :)
17:44:48 <tdawson> OK
17:44:54 <sdodson> jdetiber: it's not enabled by default, it's tech preview. There should be docs for it on docs.openshift.org though.
17:45:24 <tdawson> Moving on to images and image building
17:45:25 <jdetiber> sdodson: ack, was just trying to think of the biggest changes I noticed
17:46:08 <tdawson> sdodson: jdetiber: Do you want me to add some multi-arch tech preview links?
17:46:23 <tdawson> Is anyone here for images and image building?
17:46:59 <jdetiber> tdawson: I'd rather wait until we have some CI in place first
17:47:29 <tdawson> I know that mz1000 and team have been doing some image work, but I think all of that is on centos-devel
17:47:49 <tdawson> Looking at the time, I'm going to move on to minishft and kompose
17:48:30 <tdawson> And ... I'm not finding any of those guys online to even ping
17:48:41 <tdawson> OK, moving on to Open Floor
17:48:49 <tdawson> #topic Open Floor
17:49:16 <tdawson> Does anyone have anything that wasn't already brought up?
17:49:21 <tdawson> Or need to discuss something further?
17:50:27 <tdawson> My manual 3.7.0 test build is here - https://cbs.centos.org/koji/taskinfo?taskID=207126
17:50:56 <tdawson> I expect it to fail because I haven't got goversioninfo into cbs yet
17:51:27 <tdawson> But this way I can verify that it needs it, and that providing goversioninfo fixes it.
17:51:51 <_ari_> tdawson: is this the issue affecting my build?
17:52:11 <tdawson> _ari_: Well, the one you emails us about last week, yes.
17:52:18 <_ari_> tdawson: cool
17:52:31 <_ari_> that would make sense because this worked at one point
17:52:37 <tdawson> _ari_: As for the latest ones, that don't make it to cbs, I'm not totally sure what's going on there.
17:52:40 <_ari_> tdawson: just let me know
17:52:51 <tdawson> _ari_: Will do
17:53:00 <_ari_> :)
17:54:20 <tdawson> Anything else before I close the meeting?
17:55:15 <_ari_> not from me
17:55:24 <jdetiber> nor I
17:55:42 <tdawson> Then everyone have a great week ... we'll talk here next week.
17:55:52 <_ari_> tdawson: thanks you as well
17:56:05 <tdawson> #endmeeting