15:00:57 <chandankumar> #startmeeting RDO meeting - 2018-01-17
15:00:57 <zodbot> Meeting started Wed Jan 17 15:00:57 2018 UTC.  The chair is chandankumar. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:57 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:57 <zodbot> The meeting name has been set to 'rdo_meeting_-_2018-01-17'
15:00:57 <openstack> Meeting started Wed Jan 17 15:00:57 2018 UTC and is due to finish in 60 minutes.  The chair is chandankumar. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:58 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:00 <openstack> The meeting name has been set to 'rdo_meeting___2018_01_17'
15:01:08 <chandankumar> #topic Roll Call
15:01:12 <dmsimard> \o
15:01:13 <amoralej> o/
15:01:14 <jpena> o/
15:01:25 <bcafarel> o/
15:01:29 <chandankumar> #chair dmsimard amoralej jpena bcafarel
15:01:29 <zodbot> Current chairs: amoralej bcafarel chandankumar dmsimard jpena
15:01:31 <openstack> Current chairs: amoralej bcafarel chandankumar dmsimard jpena
15:01:49 <dmsimard> I have a *lot* of topics, let's try to keep things short and prevent sidetracking so we have time to go over everything
15:02:14 <chandankumar> number80: apevec weshay sshnaidm|mtg meeting time
15:02:19 <number80> o/
15:02:29 <chandankumar> #chair number80
15:02:29 <zodbot> Current chairs: amoralej bcafarel chandankumar dmsimard jpena number80
15:02:30 <openstack> Current chairs: amoralej bcafarel chandankumar dmsimard jpena number80
15:02:42 <chandankumar> jruzicka: meeting time
15:02:57 <mary_grace> o/
15:03:05 <chandankumar> #chair mary_grace
15:03:05 <zodbot> Current chairs: amoralej bcafarel chandankumar dmsimard jpena mary_grace number80
15:03:10 <openstack> Current chairs: amoralej bcafarel chandankumar dmsimard jpena mary_grace number80
15:03:28 <chandankumar> So starting with the first topic
15:03:33 <chandankumar> #topic Nomination of new RDO infrastructure cores
15:03:37 <chandankumar> dmsimard: go ahead
15:03:43 <dmsimard> Hi
15:03:47 <dmsimard> I'm David
15:03:51 <weshay> hello
15:04:01 <chandankumar> #chair weshay
15:04:01 <zodbot> Current chairs: amoralej bcafarel chandankumar dmsimard jpena mary_grace number80 weshay
15:04:02 <openstack> Current chairs: amoralej bcafarel chandankumar dmsimard jpena mary_grace number80 weshay
15:04:06 <chandankumar> dmsimard: :-)
15:05:02 <dmsimard> There's a few new RDO infrastructure cores I'd like to do. I'll be sharing most of the background and context around these nominations in a blog post soon but...
15:06:13 <dmsimard> The general idea is that the core developers of Software Factory, nhicher, tristanC, mhu and fbo know Software Factory, Zuul, Nodepool and Gerrit better than we do.
15:06:31 <dmsimard> When I say "we", it's us, mere RDO mortals
15:07:00 <EmilienM> ouch, bad timing for this meeting
15:07:11 <chandankumar> #chair EmilienM
15:07:11 <zodbot> Current chairs: EmilienM amoralej bcafarel chandankumar dmsimard jpena mary_grace number80 weshay
15:07:12 <openstack> Current chairs: EmilienM amoralej bcafarel chandankumar dmsimard jpena mary_grace number80 weshay
15:07:14 <dmsimard> EmilienM: yeah..
15:07:14 <apevec> there's never good timing
15:07:20 * EmilienM leaves the chair
15:07:20 <chandankumar> #chair apevec
15:07:20 <zodbot> Current chairs: EmilienM amoralej apevec bcafarel chandankumar dmsimard jpena mary_grace number80 weshay
15:07:20 <openstack> Current chairs: EmilienM amoralej apevec bcafarel chandankumar dmsimard jpena mary_grace number80 weshay
15:07:39 <apevec> #unchair EmilienM
15:07:39 <zodbot> Current chairs: amoralej apevec bcafarel chandankumar dmsimard jpena mary_grace number80 weshay
15:07:40 <openstack> Current chairs: amoralej apevec bcafarel chandankumar dmsimard jpena mary_grace number80 weshay
15:08:01 <number80> +2 for them
15:08:06 <apevec> +2
15:08:08 <dmsimard> All four of them have shown great skill in developing Software Factory but also operating and maintaining softwarefactory-project.io (where DLRN and rdopkg are hosted, amongst other things)
15:08:20 <jpena> +2
15:08:20 <weshay> +1
15:08:24 <chandankumar> +1
15:08:34 <number80> (except for mhu, I want him to wear a funny hat everytime he touches the infra)
15:08:52 <dmsimard> Over the past few weeks, they have also been getting involved in different RDO projects such as DLRN -- but they're also consistent contributors to upstream projects like Storyboard, Nodepool and Zuul.
15:08:53 <apevec> mhu has a collection
15:08:57 <amoralej> +1
15:09:05 <number80> yes, that's not a too harsh condition for him :)
15:09:09 <dmsimard> You guys are voting before I ask the actual question :P
15:09:16 * dmsimard gives background
15:09:22 <rdogerrit> Merged rdoinfo master: Promote CBS tags update for ocata-testing  https://review.rdoproject.org/r/11365
15:09:22 <mhu> number80, apevec I knew my dapper sense of fashion would be my ultimate downfall
15:09:24 <number80> dmsimard: well go ahead :)
15:10:20 <dmsimard> So I'd like to formally nominate tristanC, nhicher, mhu and fbo as RDO infrastructure cores -- this is not limited to the "config" repo (https://www.rdoproject.org/infra/review-rdo-infra-core/ has slightly confusing vocabulary) but for the larger part of the RDO servers as well.
15:11:27 <dmsimard> As cores, they will gain operational knowledge on how Software Factory runs at scale (RDO runs 20 000 jobs a month) and they are already familiar with the remainder of the components we are using, from the OpenShift registry to DLRN and ci.centos.org inbetween.
15:11:49 <dmsimard> Cast your votes and I'll summarize this in an email to the mailing list :)
15:13:09 <dmsimard> I don't see any objections, I'll take it up to the ML
15:13:16 <dmsimard> This ties in to our next topic
15:13:25 <chandankumar> #topic  Nominations for configuration cores
15:13:40 <chandankumar> weshay: go ahead
15:13:52 <chandankumar> #link https://www.rdoproject.org/infra/review-rdo-infra-core/
15:13:55 <dmsimard> ah weshay sneaked a topic and messed up my order, okay go ahead :P
15:14:19 <weshay> howdy.. last night EmilienM was looking for more folks w/ the ability to change the jobs in rdo software factory
15:14:30 <weshay> proposed rlandy as an additional
15:14:50 <weshay> specifically for https://github.com/rdo-infra/review.rdoproject.org-config
15:15:05 <weshay> so proposing rlandy as core on https://github.com/rdo-infra/review.rdoproject.org-config
15:15:23 <weshay> I can follow up w/ an email
15:15:24 <weshay> that is all
15:16:06 <chandankumar> weshay: thanks :-)
15:16:11 <chandankumar> moving next
15:16:15 <chandankumar> #topic Let's chat about review.rdoproject.org and softwarefactory-project.io
15:16:16 <dmsimard> I'm +1 for rlandy becoming core on the config repo, she's learned a great deal over the past few weeks and is familiar with the upstream jobs as well as the ones in review.rdo
15:16:37 <number80> +2 yeah, pretty active in reviewing changes and good understanding of the infra
15:16:51 <dmsimard> ok, my topics were out of order -- sorry if this is feeling a bit rushed
15:17:38 <dmsimard> So, tying in to the topic of making the software factory core developers members of the RDO infrastructure core group -- let's chat about review.rdoproject.org and softwarefactory-project.io.
15:18:34 <dmsimard> Each has different tenants: RDO essentially has "rdo", "rdo-infra" and "opstools" while SF.io has themselves (inception!), and other projects such as DLRN, skydive, rdopkg, DCI, etc.
15:20:08 <dmsimard> What I'd like to propose is that we merge the two instances of software factory into softwarefactory-project.io. This will allow projects to collaborate easily with each other if need be (CI jobs, depends-on, etc.) but it also means the operational and maintenance strain is split amongst more people on less infrastructure.
15:20:59 <dmsimard> This also optimizes our resource utilization where we now have one (larger) cluster instead of two "medium" ones and we have a larger, common nodepool capacity
15:21:12 <apevec> fbo, could SF support multiple DNS domains within the same SF instance?
15:21:19 <number80> Well, difficulty is that we also want to keep the branding (e.g: logo and also url but that can be rewrite rules though)
15:21:33 <rdogerrit> Ricardo Noriega proposed openstack/tap-as-a-service-distgit rpm-master: Initial spec file for Tap as a Service project  https://review.rdoproject.org/r/11351
15:21:50 <dmsimard> number80: They recently drafted a spec upstream to do exactly that: https://github.com/openstack-infra/infra-specs/commit/b3df3dc134e8a9d66c2088f64ee09fc60107b997
15:22:21 <dmsimard> number80: tl;dr, as the OpenStack foundation wants to host additional "communities" (such as Kata containers), they need to be able to share infrastructure -- ideally white labelled
15:22:37 <number80> dmsimard: this is fine, that is the only objection I would have against it but if it is adressed, it's fine
15:23:04 <dmsimard> Zuul, for example, already implemented it's own mailing lists within the OpenStack infrastructure: http://lists.zuul-ci.org/cgi-bin/mailman/listinfo/zuul-announce
15:23:07 <apevec> good link, I'd actually like that end result would be that we move all our resources upstream
15:23:28 <dmsimard> apevec: baby steps, but yes, this opens up different opportunities in the future
15:23:39 <apevec> ack
15:24:46 <dmsimard> I mean, at what point is it problematic for us to work from inside review.openstack.org (or a sub-branded review.rdoproject.org within the same hosting) if we have control over the CI and jobs
15:25:04 <number80> dmsimard: so do the infra group have a tentative schedule for that migration?
15:25:12 <dmsimard> We could run *everything* inside the OpenStack gate and the capacity that we have, share it with the upstream OpenStack nodepool instead
15:25:15 <number80> or are you just checking if you can attempt it?
15:25:21 <dmsimard> But that's just dreaming
15:25:46 <apevec> dmsimard, it is not
15:25:48 <dmsimard> number80: There's no formal plans yet, the technical discussions have not happened yet
15:25:49 <apevec> it's good goal!
15:26:06 <dmsimard> number80: I'm bringing up the idea just now
15:26:07 <number80> dmsimard: dreamland is never as far as we think it is (quotes shamelessly stolen to amonville)
15:26:46 <dmsimard> I'm going to miss Alexis so much :(
15:26:53 <pabelanger> I suspect PTG is when we decide new name for openstack-infra (eg: openci) and split from openstack gate
15:27:31 <apevec> pabelanger, somebody already reserved it :) github.com/openci
15:27:43 <pabelanger> yup, that is us
15:27:45 <dmsimard> Okay, so -- in general, is there any strong objections to a plan where we would merge review.rdoproject.org and softwarefactory-project.io ? I don't suspect this will happen overnight, it will require some work to migrate things over -- but what's interesting is that it also provides us with a "bridge" to migrate to Zuul v3. We migrate things progressively from review.rdo (zuulv2) to sf.io (zuulv3).
15:28:04 <fbo> apevec: this is not tested but it should possible. We need to check at the authentication level if it cause trouble as the token verification is based on the domain.
15:28:22 <amoralej> could storyboard be disabled for rdo repos?
15:28:23 <apevec> fbo, thanks, sounds promising
15:28:26 <number80> dmsimard: Alexis is still here :)
15:28:33 <dmsimard> number80: you know what I mean :p
15:28:43 <apevec> number80, yeah, we can't get rid of him
15:28:47 <dmsimard> amoralej: I don't think it hurts if no one uses it
15:28:58 <pabelanger> The idea of openci is very appealing, because you don't need to manage zuul / nodepool, you just show up with your cloud resources and everything starts working. BYOC
15:29:11 <amoralej> only problem is if users starts creating tickets that nobody is looking at :)
15:29:22 <amoralej> but yeah
15:29:22 <apevec> amoralej, where is it enabled?
15:29:28 <dmsimard> amoralej: but this is a problem we have to some extent today with bugzilla
15:29:37 <amoralej> if softwarefactory-project.io
15:29:37 <dmsimard> and bugzilla isn't linked in review.rdo
15:29:50 <jpena> storyboard is not enabled on softwarefactory-project.io, anyway
15:30:08 <amoralej> oh, i thought it was, sorry then
15:30:15 <fbo> also is same instance mean same gerrit ?
15:30:24 <dmsimard> fbo: yes
15:30:34 <dmsimard> fbo: same nodepool, same gerrit, same everything
15:30:45 <dmsimard> it'll be fun :)
15:30:51 <apevec> we'd use namespaces in gerrit
15:31:05 <apevec> review.rdo -> openstack/* puppet/*
15:31:09 <dmsimard> OH, it would be the opportunity to fix the openstack/ namespace mess in review.rdo
15:31:14 <fbo> I fear about the mess to find stuff on gerrit
15:31:25 <apevec> yeah lots of details to think through
15:31:35 <fbo> Then we need to renew the effort on project dashboard for gerrit
15:31:50 <dmsimard> fbo: https://github.com/openstack/gerrit-dash-creator is a thing
15:31:51 <number80> is developping custom gerrit plugins out of picture? (just asking)
15:31:53 <fbo> one for SF, DCI, RDO ..
15:32:02 <pabelanger> gerrit support different dashboard
15:32:05 <pabelanger> what dmsimard just linked
15:32:07 <fbo> dmsimard: yes that's partially implemented with it
15:32:25 <dmsimard> fbo: if anything, this is an opportunity to improve the user experience because OpenStack's gerrit has 1500+ projects :)
15:32:30 <pabelanger> openstack projects us gerrit-dash-creator upstream, to help with the noise
15:32:30 <fbo> https://softwarefactory-project.io/dashboard/
15:32:45 <dmsimard> fbo: if you're scared about a few hundred projects, it's concerning to think what users believe about the upstream gerrit
15:32:46 <fbo> this is the dash-creator as background
15:33:47 <fbo> dmsimard: we need to evolve to keep the platform easy to work with that all :)
15:33:53 <dmsimard> and that's great
15:34:00 <dmsimard> evolving, scaling, improving
15:34:05 <dmsimard> it's awesome
15:34:21 <dmsimard> So, anyway, I don't see strong objections so I'll take this one to the mailing list as well -- we have other things to talk about. Thanks !
15:34:45 <number80> ack
15:35:02 <chandankumar> dmsimard: moving to next topic
15:35:23 <chandankumar> #topic New grafana for review.rdoproject.org
15:35:27 <chandankumar> #link https://review.rdoproject.org/grafana/
15:35:57 <dmsimard> This new thing also weirdly ties in with this whole idea of adding cores to RDO's infrastructure and merging review.rdo and sf.io
15:36:38 <dmsimard> We've added influxdb, telegraf (for metrics) and grafana for review.rdo and we realized that, hey, we could just deploy that stack across all of RDO's servers for free -- let's do it
15:37:26 <dmsimard> So we did it, and all of our servers are on there right now -- it took maybe an hour at most and most of it had to do with short-circuiting how sf-config worked, nhicher and I will iron out the details to make that work seamlessly in the future
15:38:12 <dmsimard> Originally, the feature was only "supported" for servers part of the Software Factory cluster but there was really nothing preventing us from using it as it was generic enough
15:38:51 <dmsimard> The TripleO and TripleO CI teams will also be sending job metrics there shortly so it should be quite useful. I'm happy about that !
15:39:03 <dmsimard> I don't have anything else, was just an announcement :)
15:39:28 <amoralej> nice work dmsimard!
15:40:00 <dmsimard> I can't take all the credit, it's the SF implementation that did it all
15:40:31 <amoralej> nice work SF! :)
15:40:46 <amoralej> i see some jobs metrics also
15:40:54 <amoralej> that's WIP?
15:41:00 <chandankumar> moving to next topic
15:41:15 <dmsimard> amoralej: yes, that's what sshnaidm|mtg, trown and others are working on
15:41:41 <trown> yep, thanks for the really quick work there dmsimard
15:41:51 <chandankumar> #topic Cleaning up github.com/[redhat-openstack,openstack-packages,rdo-packages,rdo-common,rdo-infra]
15:42:24 <dmsimard> tl;dr, I was working on a solution to search code across all our different git repositories and I realized we had a lot of old stuff lying around
15:42:28 <apevec> let's separate legacy redhat-openstack and openstack-packages from rdo-*
15:42:47 <apevec> I (re)started redhat-openstack cleanup
15:42:55 <apevec> thanks for bringing that up
15:43:09 <dmsimard> apevec: can we just delete some of those ? or maybe rename them and prefix them with retired- or something (github handles the redirections automatically so it should not be an impact if someone somewhere actually still uses them)
15:43:11 <apevec> for openstack-packages - I'll drop all and leave it empty
15:43:27 <apevec> just to reserve namespace for the future use
15:43:33 <apevec> dmsimard, yes, massive dropping it will be
15:43:38 <dmsimard> I was thinking we could have a github organization called "rdo-archive" or something to that effect and just move old stuff there
15:43:48 <apevec> redhat-openstack is more complicated, there are some active projects there
15:44:02 <apevec> I'm pinging folks around
15:44:21 <apevec> dmsimard, no, let's not create new org :)
15:44:27 <dmsimard> okay, if there are legit projects that would be more appropriate in another namespace, let's write those down too
15:44:45 <apevec> yes, I'll come up with the list and propose migration
15:44:46 <dmsimard> I was thinking about redhat-openstack/website to rdo-infra but I don't want to make things complicated for would-be contributors
15:44:55 <dmsimard> apevec: thanks !
15:44:57 <apevec> e.g. whitebox-tempest-plugin is now in review.rdo
15:45:42 <dmsimard> apevec: what do you mean by separating redhat-openstack and openstack-packages from rdo- ?
15:45:44 <jtomasek> apevec: since ui-deps 8-4 is built now, will the upstream jobs automatically use that or do I need to change something somewhere?
15:45:54 <dmsimard> apevec: aren't they already separated ?
15:46:28 <chandankumar> jtomasek: meeting is in progress,:-)
15:46:44 <number80> apevec: maybe sending a list of project on the dev list and ask people to act on it
15:48:23 <jtomasek> chandankumar: oops, sorry
15:48:53 <dmsimard> ok let's follow up with this topic later
15:48:55 <dmsimard> we have more things
15:49:05 <dmsimard> chandankumar: we can do codesearch (sorry again for out of order agenda)
15:49:28 <chandankumar> dmsimard: yup
15:49:38 <chandankumar> #topic New tool to search code
15:49:50 <chandankumar> #link https://codesearch.rdoproject.org/
15:49:58 <dmsimard> OpenStack has this thing: http://codesearch.openstack.org/
15:50:07 <chandankumar> dmsimard: thanks for working on this, it is much needed, for me sometime
15:50:10 <dmsimard> It allows you to quickly search patterns across all of the OpenStack projects
15:50:34 <dmsimard> I've been wanting to implement it for RDO for a *very* long time but there was a blocker that prevented us from integrating it
15:50:45 <apevec> dmsimard, I mean to separate cleanup discussion
15:50:47 <fbo> dmsimard: is it a packaged version you installed ?
15:50:57 <apevec> rdo-* should be fully just mirrors from review.rdo
15:51:02 <dmsimard> apevec: ah ok
15:51:34 <dmsimard> The software that runs codesearch, hound ( https://github.com/etsy/hound ) assumed that the git repositories had a master branch and it actually turns out that most RDO repositories don't have a master branch -- we use things like rpm-master, pike-rdo, etc.
15:52:00 <chandankumar> dmsimard: can we also include fedora dist-git also under rdo codesearch?
15:52:48 <dmsimard> I tried to chase the maintainers down because the project is mostly unmaintained but to no avail -- someone eventually sent a pull request to do exactly what we needed and that person is my hero to get a feature in that'd
15:52:58 <dmsimard> err, weird copy/paste, ignore that last line
15:53:10 <dmsimard> I tried to chase the maintainers down because the project is mostly unmaintained but to no avail -- someone eventually sent a pull request to do exactly what we needed and that person is my hero: https://github.com/etsy/hound/pull/275
15:53:49 <dmsimard> The implementation is a PoC right now and it's "incubated" here, it's actually quite interesting because it would allow anyone to use hound very easily for whatever repositories they want: https://github.com/dmsimard/hound-localrepos
15:53:52 <number80> Maybe we should move to the list and move to the next topic ;)
15:54:16 <dmsimard> I'll move it up to rdo-infra soon, in the meantime the PoC is available at https://codesearch.rdoproject.org/
15:54:41 <dmsimard> jruzicka and rbowen have topics for the remaining 5 minutes
15:54:49 <chandankumar> #topic Summary of rdopkg development in 2017
15:55:01 <jruzicka> #link https://www.rdoproject.org/blog/2018/01/summary-of-rdopkg-development-in-2017/
15:55:58 <jruzicka> jschlueter(27), ktdreyer(11), amoralej(7), ssbarnea(5), number80(5), mhu(4), larsks(3), jpena(3), jjoyce(1), dmsimard(1) thanks for contributing
15:56:09 <dmsimard> hey I contributed something
15:56:12 <mhu> \o/
15:56:22 <dmsimard> great post jruzicka
15:56:34 <number80> Good job!
15:56:39 <dmsimard> it's always awesome to see a retrospect of all the improvements over time
15:57:05 <jruzicka> dmsimard, thanks and yes, it's good to sit and look what was done instead of just rushing forward ;)
15:57:35 <jruzicka> that's it, let's go to next topic
15:57:41 <chandankumar> #topic RDO team interview at the PTG?
15:57:49 <chandankumar> #link https://docs.google.com/spreadsheets/d/1MK7rCgYXCQZP1AgQ0RUiuc-cEXIzW5RuRzz5BWhV4nQ/edit#gid=0
15:57:49 <rdogerrit> Merged rdoinfo master: Bump python-botocore to 1.6.0 for Pike/Queens  https://review.rdoproject.org/r/10753
15:57:49 <rbowen> We mentioned last week that we might want to do a RDO team interview at the PTG.
15:58:03 <rbowen> The schedule is now posted, if folks who will be there want to pick a time.
15:58:09 <rbowen> That's the entire topic.
15:58:12 <dmsimard> Who is going ? I unfortunately won't be going
15:58:21 <rbowen> So those of you who will be there, discuss amongst yourselves.
15:58:32 <number80> amoralej, apevec, jpena, chandankumar are you fine with a group interview?
15:58:43 <chandankumar> number80: yup
15:58:47 <amoralej> yes
15:58:48 <jpena> number80: that sounds like a good idea
15:58:55 <amoralej> yeah, i like it
15:58:59 <chandankumar> number80: doing next try for visa doc submission tomorrow
15:59:05 <rbowen> It's good to spend some time planning what you might want to say about RDO
15:59:13 <rbowen> You have up to 30 minutes, so it's not a huge amount of itme.
15:59:29 <rbowen> Anyways, get with me at some point before then, and we can figure out what leading questions you want me to ask. :-)
15:59:31 <dmsimard> RDO will be well represented at the PTG, nice
15:59:31 <number80> rbowen: good point, I'll start a thread to decide what we want to emphasize
15:59:37 <dmsimard> pabelanger will be there too!
15:59:45 <chandankumar> moving to last topic
15:59:53 <chandankumar> #topic chair for next meeting
16:00:03 <dmsimard> I can chair the next one
16:00:07 <chandankumar> Any volunteers for chairing next meeting
16:00:14 <chandankumar> #action dmsimard to chair for next meeting
16:00:25 <chandankumar> (1 min left)
16:00:32 <chandankumar> dmsimard: thanks :-)
16:01:11 <chandankumar> skipping open discussion as times up!
16:01:15 <chandankumar> #endmeeting