15:00:44 <number80> #startmeeting RDO Packaging meeting (2015-01-15)
15:00:44 <zodbot> Meeting started Wed Jan 14 15:00:44 2015 UTC.  The chair is number80. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:44 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:59 <number80> #topic roll call
15:01:06 <number80> present yourself :)
15:01:15 <number80> .hellomynameis hguemar
15:01:16 <zodbot> number80: hguemar 'Haïkel Guémar' <karlthered@gmail.com>
15:01:26 * kashyap waves
15:01:27 <apevec> .hellomynameis apevec
15:01:28 <zodbot> apevec: apevec 'Alan Pevec' <apevec@gmail.com>
15:01:30 <kbsingh> hello
15:01:32 <rbowen> .hellomynameis rbowen
15:01:33 <zodbot> rbowen: Sorry, but you don't exist
15:01:38 <kbsingh> :)
15:01:39 <kashyap> :)
15:01:40 <apevec> lol
15:01:42 <rbowen> Wow. That's *harsh*
15:01:50 <kbsingh> i am sure rbowen exists, I've seem him
15:01:50 <kashyap> .hellomynameis kashyapc
15:01:51 <number80> lol
15:01:55 <kashyap> .hellomynameis kashyapc
15:01:57 <zodbot> kashyap: kashyapc 'kashyap chamarthy' <kchamart@redhat.com>
15:02:08 <kashyap> rbowen, It needs your FAS account
15:02:09 <number80> rbowen: it uses data from the Fedora Account System :)
15:03:00 <jruzicka> .hellomynameis jruzicka
15:03:01 <zodbot> jruzicka: jruzicka 'Jakub Ruzicka' <jruzicka@redhat.com>
15:03:07 <jruzicka> Oh nice ;)
15:03:22 <apevec> ok, let's wait few min
15:03:24 <number80> gchamoul, mrunge, chandankumar: ping
15:03:26 <rbowen> .hellomynameis rcb
15:03:28 <zodbot> rbowen: rcb 'Rich Bowen' <rbowen@redhat.com>
15:04:00 <gchamoul> number80: pong
15:04:39 <gchamoul> .hellomynameis gchamoul
15:04:39 <zodbot> gchamoul: gchamoul 'Gael Chamoulaud' <gchamoul@redhat.com>
15:05:36 <number80> eggmaster ?
15:05:50 <eggmaster> number80: hello
15:06:02 <number80> ok, I think we could start ?
15:06:06 <kashyap> Yeah
15:06:07 <apevec> yep
15:06:08 <eggmaster> ah ok, /me sees now
15:06:14 <eggmaster> .hellomynameis eggmaster
15:06:15 <zodbot> eggmaster: Sorry, but you don't exist
15:06:20 <eggmaster> I don't exist
15:06:21 <gchamoul> lol
15:06:22 <number80> :)
15:06:30 <apevec> eggmaster, FAS account
15:06:35 <kashyap> eggmaster, Use yous FAS account name instead of phony names :P
15:06:37 <eggmaster> .hellomynameis slinabery
15:06:38 <zodbot> eggmaster: slinabery 'None' <slinaber@redhat.com>
15:06:47 <number80> meeting agenda lives here: https://etherpad.openstack.org/p/RDO-Packaging
15:06:50 <eggmaster> heh, 'None', fair enough
15:06:52 <jruzicka> eggmaster, Hey None!
15:06:57 <apevec> eggmaster, not sure what's worse: to not exist or be 0.75
15:07:00 <number80> #topic pending RDO updates
15:07:02 <eggmaster> apevec: lol
15:07:14 <apevec> right, so re. topic
15:07:28 <number80> apevec put the latest updates on the etherpad
15:07:28 <apevec> until we have rdopkg update CI public, I'll post regular updates
15:07:31 <apevec> what's in the queue
15:07:44 <rbowen> #link https://etherpad.openstack.org/p/RDO-Packaging
15:07:45 <apevec> yep, current snapshot is in ethpad, so I don't spam irc
15:07:56 <apevec> I'll attach it to the edited minutes for the list
15:08:07 <number80> #chair apevec rbowen jruzicka gchamoul kashyap eggmaster
15:08:07 <zodbot> Current chairs: apevec eggmaster gchamoul jruzicka kashyap number80 rbowen
15:08:19 <apevec> I'll run glance CVE fixes through Phase2 today
15:08:22 <number80> https://etherpad.openstack.org/p/RDO-Packaging
15:08:34 <apevec> and sync everything live
15:08:49 <kashyap> apevec, Phase2 is a kind of CI run?
15:08:51 <mrunge> number80, pong. Sorry, I can't attend.
15:09:00 <number80> mrunge: no problem
15:09:06 <kashyap> Judging from Phase1 (smoke test)
15:09:13 <eggmaster> apevec: I'm not overoptimistic about phase2 working, ping me when you start it pls.
15:09:14 <apevec> kashyap, yes, there's link in ethpad explaining CI steps
15:09:25 <number80> #info apevec will post update on what's in the queue until RDO CI goes public
15:09:28 <eggmaster> (I can help)
15:09:31 <apevec> eggmaster, it worked yesterday
15:09:40 <eggmaster> apevec: oh. good.
15:09:43 <kashyap> apevec, Yes, thank you (line #30)
15:09:44 * eggmaster makes sure it actually did something
15:09:56 <apevec> but yes, I'll ping you if needed
15:10:43 <apevec> has anyone planned to push any RDO updates today?
15:10:43 <eggmaster> apevec: ok, good. it does seem to have worked. snapshot provision failed a few days ago and I'd assumed the prob there would bite phase2 as well, but appears not.
15:11:14 <apevec> eggmaster, I'll keep my fingers crossed
15:11:19 <kashyap> eggmaster, Is there a bug for that?
15:11:26 <kashyap> that == snapshot provision
15:11:27 <weshay> is phase1/phase2 posting to gerrithub now?
15:11:40 <apevec> kashyap, it's internal infra
15:11:41 <jruzicka> kashyap, RDO CI: phase1: smoke test, phase2: full tempest run, and then there is one more CI on the stage repo before pushing AFAIK
15:11:46 <apevec> and/or CI job
15:12:02 <apevec> jruzicka, thanks for the concise summary!
15:12:05 <eggmaster> kashyap: there is only https://trello.com/c/ISpc8D31
15:12:20 <eggmaster> kashyap: no bug, but I'm tracking it.
15:12:21 <apevec> #info RDO CI: phase1: smoke test, phase2: full tempest run, and then there is one more CI on the stage repo before pushing
15:12:23 * DV watches from a distance :-)
15:12:41 <kashyap> eggmaster, apevec, Noted, thx.
15:12:42 <apevec> right, we have public trello board for tracking tasks
15:12:53 <apevec> bz is filed if there's actual packaging bug
15:13:00 <number80> #info public trello board for tracking task
15:13:14 <apevec> #info https://trello.com/b/HhXlqdiu/rdo
15:13:47 <apevec> ok, any question re. updates?
15:14:10 <apevec> weshay, it's 3rd topic :)
15:14:13 <number80> looks good, some people will be happy to get these updates :)
15:14:38 <number80> then, let's move to the next topic ?
15:14:56 <apevec> 3 2 1 move
15:15:03 <number80> #topic EL6 Juno
15:15:20 <apevec> we had initial thread on the rdo-list
15:15:26 <apevec> I'd like to summarize:
15:15:40 <apevec> for Juno we use Fedora dist-git, master branch
15:15:47 <apevec> (will be f22 in Feb)
15:16:20 <apevec> and I'd like to hear if there are concerns about accepting fedora contributor agreement
15:16:26 <apevec> (or whatever is that called)
15:16:31 <weshay> eggmaster, have you had a chat w/ jruzicka regarding rdo-update gerrit topics?
15:16:37 <eggmaster> One further point re rdopkg. I note that rdopkg CI never does 'full tempest run'. I think what apevec meant there was that for phase2, we do a full packstack aio provision on a base image (phase1 uses prebuilt snapshotted image and installs updated RPMs).
15:16:39 <apevec> it's technically not CLA
15:16:43 <eggmaster> Just want to avoid misunderstanding.
15:16:49 <eggmaster> weshay: I have not.
15:16:50 <rbowen> I'm afraid that notice of this meeting may have been late enough that some of the people with concerns around that might not be here.
15:16:55 <apevec> but legal depts might see otherwise...
15:17:13 <weshay> eggmaster, please do
15:17:15 <apevec> rbowen, good point, but we can continue discussion on the rdo-list
15:17:31 <number80> rbowen: I'll post the minutes on the rdo list
15:17:32 <apevec> just wanted to get attention and get it resolved asap
15:17:35 <jruzicka> eggmaster, no tempest run for phase2?
15:17:52 <apevec> jruzicka, eggmaster - let's keep this for the next topic :)
15:17:54 <eggmaster> jruzicka: there is tempest run in phase1 and phase2
15:18:07 <jruzicka> ok, ok
15:18:08 <eggmaster> jruzicka: but it is not a 'full tempest run', i.e. we just do network_basic_ops
15:18:26 <rbowen> In particular, Tim (noggin143) isn't here.
15:18:49 <apevec> ok, so re. EL6 Juno - in the meantime I'm happy to review patches posted on rdo-list
15:19:04 <number80> #action apevec moving forward the EL6/Juno discussion to the list
15:19:07 <apevec> and we can push on contributor's behalf
15:19:31 <apevec> I guess it would be could if they could at least provide s-o-b line in the commit message?
15:19:40 <apevec> s/be could/be good/ !
15:19:53 <apevec> but i'm not a lawyer...
15:20:25 <number80> apevec: we should ask RH Legal for that (any volunteer ?)
15:20:34 <apevec> I'll take it
15:20:54 <apevec> also need to check about centos dist-git once that's open for self-service
15:20:55 <number80> #action apevec contact RH Legal about patch sent through mailing-list
15:20:55 <kashyap> S-o-B is sensible regardless if you're a lawyer or not (of course I'm not one too) -- https://www.kernel.org/doc/Documentation/SubmittingPatches
15:21:14 <apevec> kbsingh said there is none CLA for centos  :)
15:21:24 <apevec> but I'd like to double-check w/ legal
15:22:09 * number80 suggests reusing the Fedora Contributor agreement
15:22:19 <number80> minimal and already approved by Legal
15:22:43 <apevec> number80, please put URL as info for the minutes
15:22:52 <kashyap> Yeah, but IIRC, some folks wanted a little clarification on that at Paris summit?
15:22:55 <rbowen> As I understand it, the objection isn't even the agreement itself, just the pain of clearing it through legal at various organization.
15:23:02 <kbsingh> we dont have a cla for centos, anyone can come contribute
15:23:29 <kbsingh> for patches sent to existing code - we already assume that its the sme license as the code that is being patched.
15:23:43 <kashyap> apevec, Previous discussion on rdo-list about Fedora Project Contributor Agreement -- https://www.redhat.com/archives/rdo-list/2014-November/msg00015.html
15:24:03 <number80> https://fedoraproject.org/wiki/Legal:Fedora_Project_Contributor_Agreement
15:24:23 <rbowen> How about using the upstream (OpenStack) CLA, since this is downstream of that, and most of the people in question will have already signed that one?
15:24:45 <number80> (btw, #link is optional if the link comes first)
15:24:51 <kashyap> Fedora doesn't use "CLA "either -- the above page explicitly avoids that terminology
15:24:57 <kbsingh> why is there a need for a cla ?
15:25:28 <apevec> kbsingh, not cla, but something to ensure we're not getting legally problematic code
15:25:36 <gchamoul> rbowen: +1
15:25:56 <kashyap> kbsingh, Please read the above mailing list thread that's noted, which describes what Fedora means by FPCA
15:25:59 <number80> apevec: +1
15:26:36 <number80> rbowen: since specs will end up in Fedora git, it will have to comply with Fedora guidelines
15:26:50 <kbsingh> regardless of what Fedora means by it - it just seems drama for the sake of drama
15:27:03 <chandankumar> Hello All
15:27:09 <kbsingh> apevec: what would this problematic code mean ( ie, can you give your top 3 worst case scenarios )
15:27:10 <number80> kbsingh: that's required by RH Legal
15:27:12 <apevec> kbsingh, better safe than sorry :)
15:27:25 <kbsingh> number80: its not, we dont have one - and its fine with RH Legal
15:28:10 <rbowen> We have an action item for apevec to ask RH Legal that, so perhaps we table this until we have an answer to that.
15:28:15 <apevec> ack
15:28:20 <kbsingh> cool
15:28:47 <number80> kbsingh: I'd ask Tom Callaway (Spot) about the Fedora CLA, he knows the whole thing but that wasn't optional :/
15:28:49 <kashyap> I wouldn't call it "drama". Anyway, mooving on
15:28:58 <number80> *npds*
15:29:03 <apevec> yep, next topic
15:29:21 <number80> #topic rdopkg update CI public
15:29:26 <number80> apevec, eggmaster ?
15:29:34 <apevec> eggmaster, jruzicka - stage is yours now :)
15:30:11 <apevec> there's bigger CI job refactoring planned
15:30:28 <apevec> but in the meantime I'd like to move internal rdo-update.git public
15:30:51 <apevec> i.e. make internal CI job reporting there while keeping Jenkins and cloud infra internal
15:30:59 <apevec> eventually, everything moves public
15:31:07 <apevec> weshay, eggmaster - right?
15:31:33 <eggmaster> apevec I assume that also means that updates submitted via 'rdopkg update' would go only to gerrithub rdo-update project, i.e. we'd make the cut in rdopkg util to report there by defaukt
15:31:35 <jruzicka> are we no longer ashamed of the results? :)
15:31:37 <eggmaster> heh, default
15:31:43 <apevec> eggmaster, yes
15:31:59 <jruzicka> eggmaster, that's already implemented in the open sourced rdopkg
15:32:03 <apevec> jruzicka, I'll take blame for casual false positives
15:32:18 <apevec> jruzicka, right, we just need to switch CI jobs
15:32:24 <eggmaster> jruzicka: I think we'd have to live with the 'shame' (i.e. the, ahem, occasional 'false' result due to flaky CI infra) for now
15:32:49 <eggmaster> yeah so there are two things needed to be changed in jobs:
15:33:02 <eggmaster> 1) scrub the URL back to the internal jenkins job from the result report
15:33:15 <eggmaster> 2) make the jobs trigger off gerrithub rdo-update instead of internal
15:33:20 <eggmaster> very little work.
15:33:53 <eggmaster> really, imho, we should just decide on a time when to make the cut and I'll do 1&2 before anyone submits next update.
15:34:33 <kbsingh> can i help provide semi reliable hardware for the ci ?
15:34:34 <apevec> eggmaster, let's try early next week and I'll flush current updates
15:34:38 <weshay> the right thing to do is just to move all rdopkg ci public
15:35:00 <apevec> weshay, eventually yes
15:35:26 <weshay> meh
15:36:05 <apevec> weshay, this half-assed step is only to buy us time to do this refactoring properly
15:36:43 <weshay> apevec, we need to figure out from jruzicka and eggmaster is a refactor is possible
15:37:21 <weshay> if we can set the gerrit topic for rdo-update to the rdo release and platform dist
15:37:24 <apevec> right, so we don't have eta and I need publicly submittable updates asap
15:37:34 <jruzicka> fair enough
15:37:36 <eggmaster> weshay: if you're talking about keying off the topic, yes that's something I discussed with jruzicka before holidays
15:37:48 <jruzicka> better refactor it properly
15:38:06 <weshay> so is it possible?
15:38:48 <jruzicka> I seriously wanted to help eggmaster with it but I can't stand Jenkins.
15:38:50 <jruzicka> what is?
15:39:34 <weshay> setting rdo-update review topic to include all the keys that would define which rdo dist and platform it addresses
15:39:39 <kbsingh> jruzicka: if not jenkins, then what ?
15:39:44 <jruzicka> I'm ready to (re)write whatever tools/scripts are needed, but I'd like to not touch Jenkins itself
15:39:54 <weshay> jruzicka, you don't need to touch jenkins
15:39:58 <eggmaster> jruzicka: you should not have to dirty your hands
15:40:09 <jruzicka> yeah, poor eggmaster is doing that for me ;)
15:40:18 <weshay> if we can get the info in the topic... all we'll need is a jjb update, push the jobs to the public jenkins and go from there
15:40:29 <eggmaster> The idea was to avoid using jenkins buildflow plugin and reduce complexity of jobs
15:40:47 <weshay> right, and just use gerrit triggers
15:41:04 <weshay> it also eliminates the need for phase1/2
15:41:09 <weshay> and just have one set of tests
15:41:16 <eggmaster> by including info about what dist/release are included in a particular update in the gerrit topic
15:41:33 <eggmaster> in some way that can be parsed as regex
15:41:43 * jruzicka afk few mins
15:41:53 <eggmaster> This is what weshay and I have been casually referring to as 'rdopkg CI 2.0'
15:42:15 <eggmaster> (although I wouldn't necessarily grace the existing rig with a '1.0' :P)
15:42:19 * apevec hopes it's not Jenkins making jruzicka sick
15:42:20 <jruzicka> eliminates need for phase[12]? how?
15:42:36 <eggmaster> weshay: yeah I need that ^ spelled out for me too :)
15:42:44 <eggmaster> but possibly we are getting into the weeds here?
15:43:19 <apevec> yeah, let's open a thread on rdo-list for details
15:43:22 <weshay> jruzicka, eggmaster it's a tech discussion we can have together.. eggmaster can you set something up please
15:43:31 <weshay> or email is fine too
15:43:32 <apevec> but could we agree to just redirect gerrit next week as a 1. step?
15:44:26 <rbowen> +1
15:44:30 <weshay> sure
15:44:39 <eggmaster> yes
15:45:01 <apevec> #action apevec eggmaster jruzicka to switch rdo-update.git to public gerrit next week
15:45:13 <number80> may I suggest that the first mail exposes the current architecture of the CI ? so new folks could jump into the discussion ?
15:45:50 <apevec> eggmaster, ^ could you write that quick email to kick discussion on rdo-list?
15:46:11 <weshay> as always I'm not happy w/ the status and level of support we're getting on that jenkins server..
15:46:30 <apevec> weshay, we got offer from kbsingh to use ci.centos
15:46:53 <eggmaster> apevec: yes, although it will be very broad-strokes type.
15:46:55 <weshay> we have a public jenkins server..
15:47:00 <kashyap> Support from who? If it's hardware issues, there's no defend-able reason by anyone to fix that
15:47:01 <apevec> eggmaster, my first take was http://apevec.github.io/rdo-intro.html#8
15:47:08 <kashyap> s/to fix that/to not fix that
15:47:33 <apevec> eggmaster, broad-strokes is good
15:47:51 <weshay> crud, which needs an update
15:48:09 <kbsingh> the thing with ci.centos is that its a series of phycical machines - its not just VM's in machines
15:48:21 <kbsingh> so you can run a 128 physical node cloud if you want
15:48:21 <eggmaster> kashyap: there are multiple points of failure. First, the jenkins we're using is somewhat resource constrained, and also has some kind of misconfiguration/glitch that causes dns lookups to fail randomly.
15:48:30 * number80 suggests that we move the discussion to the list
15:48:35 <eggmaster> kashyap: also internal cloud we've been using is fraught with errors
15:48:44 <eggmaster> kashyap: but afaik none of this is 'hardware' probs
15:48:51 <weshay> kbsingh, where can we go read about ci.centos?
15:49:14 <kbsingh> weshay: http://wiki.centos.org/QaWiki/PubHardware is the hardware that runs it
15:49:27 <weshay> thank you
15:49:28 <number80> http://wiki.centos.org/QaWiki/PubHardware
15:49:43 <number80> may we move to the next topic
15:49:45 <number80> ?
15:49:46 <kashyap> eggmaster, /me listening.
15:49:52 <kbsingh> we're closing a few things off w.r.t network access with network-ops people
15:49:56 <eggmaster> kashyap: sry, full stop.
15:50:10 <apevec> number80, NB I just switched topic order
15:50:14 <number80> ok
15:50:26 <number80> #topic Kilo milestone packages
15:50:37 <number80> apevec !
15:51:10 <apevec> plan is to get working installer against Delorean repo
15:51:13 <kashyap> Why Milestone-2 packages? I mean, is there value in making them? (Instead of Milestone-3)
15:51:14 <apevec> then CI passing
15:51:33 <apevec> kashyap, we'll just take Delorean snapshot
15:51:33 <number80> kashyap: they will be provided through delorean snapshots
15:51:46 <kashyap> (M3 == FeatureFreeze, StringFreeze, DepFreeze)
15:52:00 <apevec> plan is to import Kilo to Fedora master after f22 branches
15:52:07 <apevec> i.e. Kilo RC
15:52:10 <kashyap> number80, apevec Noted. I only heard that Delorean could do packages from git, right?
15:52:42 <apevec> kashyap, right and there are milestone-proposed branches so we can redirect it if needed
15:52:47 <apevec> in rdoinfo
15:52:52 <apevec> but those are details
15:52:53 <weshay> kbsingh, can rdo-ci get access to one node to try this out?
15:53:34 <kashyap> I assumed they need to be generated manually. If there's a tool for that great.
15:53:41 <apevec> kashyap, important is that we get Packstack working against delorean repo
15:53:43 <kbsingh> weshay: totally, i will be in touch and work this with you
15:53:48 <apevec> gchamoul, ^ how far is that?
15:53:54 <weshay> nice.. again thank you!
15:54:28 <rbowen> #action kbsingh to work with weshay to get rdo-ci access to a node on ci.centos.org
15:55:10 <gchamoul> apevec: I was on pto and sick leave during 4 weeks ! I will check it
15:55:30 <apevec> gchamoul, thanks!
15:55:59 <apevec> #action gchamoul to look at making Packstack working against Delorean repo for Kilo snapshots
15:56:21 <apevec> weshay, ^ once we have Packstack, Delorean CI job is already there, right?
15:57:04 <weshay> apevec, we have a job for it, it's been stale for a while, but yes
15:57:13 <apevec> running out of time, let's move to the last topic
15:57:14 <number80> \o/
15:57:26 <number80> #topic public RDO call
15:57:30 <number80> apevec, rbowen :
15:57:49 <apevec> I propose bi-weekly on the phone, weekly quick status sync on #rdo with minutes posted on rdo-list
15:58:03 <number80> +1
15:58:04 <apevec> on the phone or Bluejeans, you tell me
15:58:07 <rbowen> Did we get a phone bridge for that?
15:58:10 <apevec> and what about time slot
15:58:23 <rbowen> We talked about doing a Hangout, but that's limited to 10 speakers, so doesn't really work out.
15:58:25 <apevec> rbowen, I can generate temp bridge# in Bluejeans
15:58:43 <apevec> we can test this next
15:58:47 <rbowen> For those not familiar, Bluejeans is audio/video, but folks can join phone-only if they wish.
15:59:04 <rbowen> I don't know what the limit is on video participants, but it's more than Hangouts anyways.
15:59:23 <apevec> yeah, I also need to to get the list of access numbers around the world...
15:59:24 <gchamoul> rbowen: 100 !
15:59:32 <number80> rbowen: as long as it's usable through a mere phone, it should be fine (we need to ensure that access infos are accessible)
15:59:37 <rbowen> Sticking with this same timeslot seems like a good idea unless there are objections.
15:59:43 <rbowen> Yes, there are global access numbers.
15:59:57 <number80> great
15:59:59 <eggmaster> #action slinaber to start email thread describing current rdopkg CI setup
16:00:00 <apevec> ok, let's keep this slot for now
16:00:29 <number80> #info RDO irc weekly status will be every wednesday at 15:00 UTC on #rdo@freenode
16:00:34 * eggmaster drops for next mtg
16:00:50 <apevec> #action apevec to post conf# details to try Bluejeans next week
16:01:23 <number80> good
16:01:37 <rbowen> I believe that's all for that item.
16:01:38 <apevec> number80, we'll try BJ in parallel with irc, irc is good for taking notes
16:01:45 <number80> apevec: ack
16:01:51 <number80> #topic open discussions
16:02:12 <number80> anyone wants to add a new topic ? (will be dealt next week since we're running out of time)
16:02:19 <number80> or on the list ;)
16:02:57 <apevec> going once
16:03:26 <number80> well, let's close the meeting, I'll sent minutes to the list
16:03:31 <gchamoul> apevec: I think we can record a meeting in bj ?
16:03:32 <apevec> thanks everyone!
16:03:34 <rbowen> Thanks, number80
16:03:35 <number80> see you next week !
16:03:40 <number80> thanks for attending !
16:03:43 <number80> #endmeeting