15:01:12 <trown> #startmeeting RDO meeting (2016-01-27)
15:01:12 <zodbot> Meeting started Wed Jan 27 15:01:12 2016 UTC.  The chair is trown. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:12 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:01:12 <zodbot> The meeting name has been set to 'rdo_meeting_(2016-01-27)'
15:02:04 <nmagnezi> number80, jruzicka, how do we proceed?
15:02:17 <dmsimard> o/
15:02:25 <trown> #chair dmsimard jschlueter
15:02:25 <zodbot> Current chairs: dmsimard jschlueter trown
15:02:30 <number80> nmagnezi: send me patch, I'll apply it
15:02:33 <number80> o/
15:03:01 <gkadam> #chair gkadam
15:03:06 <trown> #chair number80
15:03:06 <zodbot> Current chairs: dmsimard jschlueter number80 trown
15:03:06 <jpena> o/
15:03:15 <number80> #chair jpena gkadam
15:03:15 <zodbot> Current chairs: dmsimard gkadam jpena jschlueter number80 trown
15:03:16 <trown> #chair jpena gkadam
15:03:16 <zodbot> Current chairs: dmsimard gkadam jpena jschlueter number80 trown
15:03:21 <nmagnezi> number80, can you just create a rdo-libery branch out of rpm liberty? once this is in place i'll submit my patch to rdo-libery
15:03:26 <number80> gkadam: only a chair can chair people
15:03:35 <trown> nmagnezi: can we discuss after the meeting
15:03:39 <number80> nmagnezi: only direct push to rdo-* branches
15:03:46 <number80> (we'll see post meeting)
15:04:15 <nmagnezi> number80, not sure I followed.. sorry :)
15:04:23 <mflobo> o/
15:04:31 <trown> #chair mflobo
15:04:31 <zodbot> Current chairs: dmsimard gkadam jpena jschlueter mflobo number80 trown
15:04:38 <trown> ok lets get started
15:04:46 <trown> #topic testday progress
15:05:44 <number80> do we have an etherpad?
15:06:02 <dmsimard> https://etherpad.openstack.org/p/RDO-Meeting
15:06:10 <dmsimard> oh, for testday
15:06:16 <number80> *nods*
15:06:18 <jpena> https://etherpad.openstack.org/p/rdo-test-days-mitaka-m2
15:06:26 <number80> thanks
15:06:31 <jpena> so far, I've been the only one to write bugs :)
15:06:43 <dmsimard> psh. I write bugs every day :P
15:06:51 <number80> today I'm focusing on upstream test guide (hand-made deployment)
15:07:55 <trown> I will work on https://www.rdoproject.org/testday/mitaka/testedsetups2/ for RDO Manager today
15:08:17 <trown> #action trown make test matrix for rdo-manager at https://www.rdoproject.org/testday/mitaka/testedsetups2/
15:08:51 <trown> I think it may be good to scale down what is listed for packstack there
15:09:37 <number80> so far, no packaging issue?
15:10:16 <trown> number80: well delorean 'current' is failing for mitaka, but I have another agenda topic that is related
15:10:25 <dmsimard> EmilienM: did you figure out what was that gnocchi-upgrade issue ?
15:10:26 <number80> ok
15:10:37 <ukalifon1> Hi. I'm trying to install a virtual environment for mitaka test day, according to the quickstart guide at https://github.com/trown/tripleo-quickstart. The ansible fails with: "fatal: [host0]: FAILED! => {"changed": false, "failed": true, "msg": "The checksum for /home/stack/.quickstart/undercloud.qcow2 did not match d99ca16a7008e9434b3e870a6bbaaa15; it was 72ba88f07c25f6d37a6b447735b38a0f."}".
15:10:37 <ukalifon1> I checked and there is no undercloud.qcow2 under /home/stack/.quickstart at all ...
15:10:40 <dmsimard> delorean current installs but fails tempest with nova-api internal errors
15:10:55 <EmilienM> dmsimard: yes, tl;dr; we needed a release in gnocchi iiuc, pradk ^
15:11:04 <trown> ukalifon1: awesome! I will help you out after the meeting
15:11:08 <dmsimard> EmilienM: ok so current-passed-ci is good for you ?
15:11:38 <EmilienM> dmsimard: I'm not sure as today, I've been in meeting since 7am :)
15:11:42 <dmsimard> ok
15:11:53 <EmilienM> I'll check it asap but pradk might also know
15:12:36 <trown> ok anyone want to volunteer to clean up packstack test matrix at https://www.rdoproject.org/testday/mitaka/testedsetups2/
15:12:52 <trown> or do we think we will get coverage for all of those scenarios?
15:13:21 <trown> I would rather test fewer things with a high priority than lots of things with random priority
15:13:27 <dmsimard> imcsk8: ^ ?
15:14:05 <EmilienM> dmsimard: gnocchi still broken in current-passed-ci, it's running 1.3.1
15:14:12 <EmilienM> IIRC we needed a new version upstream
15:14:16 <EmilienM> pradk: any progress on that?
15:14:22 <trown> EmilienM: it is not broken in current-passed-ci
15:14:36 <EmilienM> dmsimard: it was the only blocker in puppet openstack CI
15:14:37 <trown> EmilienM: you are not using yum-plugin-priorities so you get the deps version
15:14:49 <EmilienM> trown: that's a workaround
15:14:54 <EmilienM> we never had to do that before
15:14:56 <dmsimard> pretty sure yum-plugin-priorities is installed, no ?
15:15:01 <EmilienM> yep
15:15:11 <trown> EmilienM: it is a workaround that has always been there for delorean
15:15:25 <EmilienM> trown: https://github.com/openstack/puppet-openstack-integration/blob/master/manifests/repos.pp#L16-L26
15:15:32 <EmilienM> what should we do?
15:15:37 <number80> EmilienM, pradk if you have review, send link
15:15:39 <trown> https://review.gerrithub.io/254505 is meant to address it
15:15:59 <dmsimard> EmilienM: you should definitely get https://github.com/openstack/puppet-openstack_extras/blob/master/manifests/repo/redhat/redhat.pp to install yum-plugin-priorities when setting up the repos
15:16:01 <trown> EmilienM: I am not arguing that it should be fixed on RDO side, just that the package in delorean is not the issue
15:16:07 <EmilienM> it seems yum-plugin-priorities is not installed
15:16:10 <EmilienM> http://logs.openstack.org/90/272890/2/check/gate-puppet-openstack-integration-scenario001-tempest-dsvm-centos7/ccf6da2/logs/rpm-qa.txt.gz
15:16:12 <trown> the content of the package anyways
15:16:18 <trown> the metadata of the package is the issue
15:16:57 <trown> anyways, how is this related to testday?
15:17:41 <pradk_> EmilienM, number80, this should address the tag issue upstream https://review.openstack.org/#/c/272680/
15:17:41 <dmsimard> I opened a parenthesis when number80 asked if there were any packaging issues and talked about gnocchi
15:17:43 <dmsimard> we can move on :)
15:17:56 <number80> pradk_: thank you, sir!
15:18:24 <trown> ok, I still have "crickets" for the question regarding packstack test matrix
15:18:34 * trown does not touch packstack
15:19:10 <EmilienM> dmsimard, trown : correct ? https://review.openstack.org/273099
15:19:15 <dmsimard> imcsk8: are you around ? Is there anything specific you would like us to test ?
15:19:31 <dmsimard> paramite, jpena, social ^
15:19:57 <sshnaidm> trown, maybe it's better to define also requirements for host? I mean VIRTHOST. You wrote only "ssh root@$VIRTHOST -C 'virsh --version'", maybe to add all packages installation into playbooks so it could run on "blank" host?
15:20:18 <trown> sshnaidm: after the meeting we can chat
15:20:29 <sshnaidm> trown, ok
15:20:58 <trown> ok, we should move on
15:20:59 <jpena> dmsimard: wow, that's a lot of test setups
15:21:21 <trown> jpena: ya, that is my thought... there is no way we cover all of that
15:21:29 <social> dmsimard: multinode packstack install with heat, sahara, neutron(with lbaas fwaas and vpnaas) enabled while having standalone neutron and storage nodes with SSL on IPv6 and as an bonus finished by tempest run
15:22:06 <social> dmsimard: also you could enable gluster iscsi and nfs backends for cinder >.>
15:22:09 <dmsimard> social: so you're volounteering to test that? :)
15:22:13 <social> I think that should break
15:22:36 <social> If I get time for that from EmilienM ^^ ?:)
15:22:53 <trown> #topic rdopkg tags support
15:23:22 <trown> jruzicka: is this you?
15:23:27 <jruzicka> yes
15:23:51 <jruzicka> #link https://trello.com/c/KxKtVkTz/62-restructure-rdoinfo-for-new-needs
15:24:40 <number80> now, we have to prepare the migration of the rdoinfo database + delorean instances
15:24:53 <jruzicka> trown, can haz chair? Or do the meetbot commands work for everyone?
15:24:58 <trown> whoops
15:25:03 <trown> #chair jruzicka
15:25:03 <zodbot> Current chairs: dmsimard gkadam jpena jruzicka jschlueter mflobo number80 trown
15:25:12 <jruzicka> #link https://trello.com/c/KxKtVkTz/62-restructure-rdoinfo-for-new-needs
15:25:30 <jruzicka> #info tags/overrides are supported in current rdoinfo and rdopkg >= 0.34
15:26:06 <jruzicka> #info apevec is going to integrate tags into delorean
15:26:09 <jruzicka> right?
15:28:08 <trown> hmm I think apevec is not around, but that sounds right to me :)
15:28:21 <jruzicka> that's all from me then.
15:28:37 <jruzicka> (he said it yesterday :-p)
15:29:07 <trown> :)
15:29:28 <trown> #topic Create M2Tesday symlinks
15:29:57 <trown> I would like to create symlinks to the following known good delorean repos for testday
15:30:18 <trown> #link http://trunk.rdoproject.org/centos7-liberty/61/71/61712212b9ed9a1d76233c874875400fccd8cae8_dd86149b/
15:30:29 <trown> #link http://trunk.rdoproject.org/centos7/55/17/5517b8e9aea3ded1052209384b4194d2caa97541_673a78a2
15:30:53 <number80> ack
15:31:18 <trown> the idea being that we would have done this last week, and use this virtual 'code freeze' to stabilize our docs/UX for testday
15:31:48 <dmsimard> +1
15:32:01 <jruzicka> oh, nice touch
15:32:06 <jruzicka> +1
15:32:40 <number80> trown: there isn't a newer snapshot known as good?
15:32:59 <number80> this snapshot was made when we had many broken packages
15:33:00 <trown> number80: nope those are current-passed-ci as of right before this meeting
15:33:05 <number80> ok
15:33:29 <trown> ya that is actually another proposal I have that is not on the agenda
15:33:32 <dmsimard> number80: broken packages shouldn't make it to current-passed-ci ?
15:33:48 <dmsimard> did we miss something ?
15:33:51 <number80> dmsimard: it means that we ship very old builds
15:33:52 <trown> dmsimard: well, it is a bit of semantics
15:33:54 * apevec belated o/
15:33:57 <number80> some of them are one month old
15:34:01 <trown> #chair apevec
15:34:01 <zodbot> Current chairs: apevec dmsimard gkadam jpena jruzicka jschlueter mflobo number80 trown
15:34:20 <trown> number80: ya for this testday, that is the best we have
15:34:25 <apevec> (just testing if everything works in case me-hit-by-bus :)
15:34:27 <number80> trown: sure
15:34:29 <dmsimard> number80: because there hasn't been any commits to them so delorean hasn't built something new ?
15:34:46 <apevec> ack to what jruzicka  said
15:34:46 * dmsimard trying to understand
15:34:49 <trown> dmsimard: no because FTBFS means that we have not successfully built them in a while
15:34:53 <number80> dmsimard: no build failure (missing deps mostly)
15:35:04 <trown> so we have a package, but it is not actually latest
15:35:08 <dmsimard> number80: okay, liberty ? We have been doing okay on mitaka.
15:35:18 <number80> nope mitaka
15:35:30 <number80> the oldest one was failing for 38 days
15:35:55 <dmsimard> wow, that definitely shouldn't happen
15:36:00 <number80> yup
15:36:27 <number80> but if no newer snapshot passed CI => stick to the known good one
15:36:55 <trown> #topic Proposal: Only promote delorean if there are no FTBFS
15:37:11 <trown> I meant to add this to the agenda anyways, so lets discuss :)
15:37:35 <trown> flepied has a patch to create a 'coherent' symlink https://review.gerrithub.io/#/c/259767/
15:38:00 <trown> I think going forward, we should be only trying to promote the latest 'coherent' symlink
15:38:25 <number80> french people
15:38:30 <number80> coherent => consistent
15:38:43 <jruzicka> :D
15:38:50 <number80> (I hope this will be more explicit)
15:38:52 <trown> this does not help us for this testday, but it would mean that for M3 testday we have something releasable from delorean
15:39:09 <number80> trown: for M3, we should start forking stable packages I think
15:39:24 <trown> number80: good point, for N1 then :p
15:39:49 <number80> but yes, it will require more discipline from us to catch up with all failures
15:39:50 <trown> the idea is more that whatever is in current-passed-ci is the latest code that all passed together
15:40:13 <dmsimard> I sort of disagree but I have no good arguments. Probably pessimistic but I think there will mostly always be a ftbfs package in tree and if a set of packages passes the (now) extensive CI coverage that we have it should be tagged as such
15:40:15 <trown> not mostly the latest, with some old stuff that has packaging issues with the latest
15:40:51 <number80> dmsimard: we have both symlinks, I'm not against trying
15:40:59 <trown> dmsimard: I would argue that if we have packages that are not actively maintained in delorean, they should move to deps
15:41:35 <trown> with actively maintained being defined as resolving FTBFS as a priority task
15:41:52 <dmsimard> trown: we probably need a better classifier than "not actively maintained in delorean"
15:42:06 <jpena> I assume this consistent link is per-worker, right? We have several packages where they have issues in f24 but not in centos7
15:42:22 <dmsimard> Some projects such as gnocchi, rally, etc. have stopped (or never?) had stable/<release> branches
15:42:26 <apevec> jpena, yes, per delorean instance
15:42:35 <dmsimard> these should be in deps, they're not bound to an openstack release
15:42:54 <number80> dmsimard: rally never had, and gnocchi has different stable releases policy
15:43:01 <apevec> so "coherent" name is indeed misleading, let's rename it to "consistent" as number80 proposed
15:43:35 <apevec> we can't throw them out
15:43:42 <apevec> they're really openstack projects
15:43:58 <number80> you gotta love all your kids, even the weird ones
15:44:12 <apevec> different release schedule is making issues also in upstream CI,
15:44:27 <apevec> so nobody has figured out yet how to handle them
15:44:33 <EmilienM> w00t, gnocchi is working
15:44:41 <EmilienM> https://jenkins04.openstack.org/job/gate-puppet-gnocchi-puppet-beaker-rspec-dsvm-centos7/33/consoleFull
15:44:48 <apevec> for gnocchi will be able to switch to stable/1.3 using rdoinfo tags
15:44:58 <apevec> in Delorean Liberty
15:45:03 <EmilienM> pradk_, dmsimard, trown: ^
15:45:30 <EmilienM> degorenko: fyi your patch on gnocchi-upgrade is ready I think we can merge it today
15:45:31 <apevec> but gnocchi reqs are not synced w/ upstream global reqs
15:45:42 <jpena> could we do something similar for tempest? After a conversation with dmellado yesterday, he said it uses a similar tagging scheme
15:45:44 <trown> ya I dont think we should throw out any projects fwiw, I just think we need to change maintainers in rdoinfo if the maintainer listed there is not on board with prioritizing FTBFS
15:45:46 <apevec> so this will be always an issue w/ gnocchi and such
15:45:52 <degorenko> EmilienM, nice :)
15:46:01 <pradk_> EmilienM, cool
15:46:18 <EmilienM> yeah cool
15:46:22 <dmellado> jpena: checking context
15:46:31 <dmsimard> apevec: but there may be multiple dot releases of gnocchi during a cycle
15:46:36 <dmsimard> so which do we stick to ?
15:46:39 <dmellado> could you rephrase?
15:46:53 <apevec> trown, yeah, I've old action to query all current maintainers and ask them to confirm they're still in
15:47:18 <dmellado> btw eggmaster ping, are you around? ;)
15:47:21 <apevec> dmsimard, yeah, but we have the same issue upstream
15:47:34 <apevec> so gnocchi upstream should tell us which branch works for Liberty
15:47:43 <dmellado> dmsimard: just read that, +1 for tempest, please
15:48:07 <apevec> dmellado, jpena - what tagging scheme is used for tempest?
15:48:08 <dmsimard> apevec: okay, so this isn't yet a solved issue then - thought it was. Shifts my perspective :)
15:48:27 <dmellado> apevec: as of now I'm keeping what it was there.
15:48:40 <trown> we have strayed from the proposal I think
15:48:50 <number80> *nods*
15:48:51 <dmellado> from upstream they stopped having branches since Icehouse, and then started using tags
15:49:18 <dmellado> every git tag supports a number of releases, i.e. mitaka,liberty,kilo
15:49:20 <apevec> ah right, so we have per-release branches in RDO tempest fork
15:49:26 <trown> modulo naming, and actually landing that patch, can we vote on whether to use that symlink vs 'current' in the promote CI?
15:49:36 <dmellado> apevec: in our fork we do have branches, as we introduce some patches
15:49:52 <dmellado> so those git-tags were converted into brances for dowstream
15:49:56 <apevec> trown, yes, lets' just rename it
15:50:14 <dmellado> but I'm thinking about changing the model, as it might happen that we'd have two identical branches
15:50:38 <trown> apevec: rename which? in the patch? or what gets produced by successful CI run?
15:50:40 <dmellado> i.e. liberty/mitaka would be the same now, besides some extra configuration
15:50:41 <apevec> number80, that was your suggestion: please propose "coherent"  -> "consistent" or something better?
15:50:49 <number80> yup
15:51:01 <number80> that's what meant flepied
15:51:17 <number80> and coherent makes no sense for native speakers
15:51:20 <trown> apevec: I would prefer to keep 'current-passed-ci' since it is used alot of places
15:51:20 <apevec> trown, https://review.gerrithub.io/259767 is merged
15:51:21 <number80> *here
15:51:40 <trown> apevec: ah so it is
15:51:45 <dmsimard> current-passed-ci is used in upstream projects
15:51:52 <dmsimard> have to be careful if we want to change that
15:51:56 <apevec> trown, "current-passed-ci" is fine but  based on "consistent" not "current"
15:52:30 <trown> right, to me the implication there is that current would not pass CI if it is not also consistent
15:52:48 <trown> ie if we checked for consistancy as a CI step
15:53:02 <trown> we get that for free with using the symlink though
15:53:17 <apevec> yep
15:54:20 <trown> ok +1 to change the name of the 'coherent' symlink to 'consistent' and start using the 'consistent' symlink for the basis of promotion to 'current-passed-ci'
15:54:29 <trown> anyone opposed to that?
15:54:32 <dmsimard> So we won't be running CI on current anymore ? Only consistent ?
15:55:10 <apevec> yes, that's the proposal
15:55:14 <trown> dmsimard: nope, it requires prioitizing FTBFS
15:55:25 <dmsimard> It feels like consistent "promotions" won't occur often and we will be losing out on a lot of value because a lot of things can change across consistent promotions
15:56:02 <jschlueter> what has been frequency of consistent == current recently?
15:56:02 <apevec> it should be often, FTBFS is now made clearly visible
15:56:07 <number80> dmsimard: goal is to be better at fixing FTBFS
15:56:19 <apevec> we have bot harassing us here
15:56:37 <dmsimard> Sorry for being pessimistic T_T
15:56:37 <trown> dmsimard: we have to prioritize FTBFS the same/more as if the actual code is broken
15:56:46 <trown> that is the value add of RDO after all
15:56:53 <trown> packaging
15:57:04 <dmsimard> I'm okay with the general idea if consistent doesn't get largely out of date
15:57:04 <apevec> also one day we'll have status.rdoproject.org where this will turn Delorean status into RED :)
15:57:05 <trown> or one of the major ones anyways
15:57:14 <number80> dmsimard: prepare for the worst, aim for the best :)
15:58:02 <trown> #agreed  change the name of the 'coherent' symlink to 'consistent' and start using the 'consistent' symlink for the basis of promotion to 'current-passed-ci'
15:58:38 <apevec> number80, ^ action yourself for rename review
15:58:48 <apevec> so we can merge and deploy asap
15:59:01 <apevec> BTW is trunk.rdo down? http://trunk.rdoproject.org/centos7/status_report.html is not loading for me...
15:59:18 <apevec> or is it just under testday load?
15:59:21 <number80> #action number90 rename symlink from coherent to consistent
15:59:24 <trown> #action trown put up review for rdo-infra to change base repo to 'consistent' instead of 'current'
15:59:26 <slagle> it's down for me too
15:59:29 <dmsimard> apevec: it was up like 5 minutes ago, checking
15:59:35 <trown> number90? upgrade?
15:59:36 <number80> apevec: slow but only two packages broken
15:59:41 <number80> gnocchi + magnum
16:00:26 <number80> f22 has few more
16:00:27 <ukalifon1> I can't get the repos needed for the mitaka test day: wget http://trunk.rdoproject.org/centos7/delorean-deps.repo  .... times out
16:00:29 <trown> #topic open discussion
16:00:49 <trown> maybe we should just end the meeting and deal with this latest fire?
16:00:57 <trown> chair for next meeting?
16:01:12 <ibravo> Does test day includes a new set of RDO Manager instructions?
16:01:31 <jpena> I can chair the next meeting, I haven't done it for a while
16:01:34 <trown> ibravo: I have updated https://www.rdoproject.org/rdo-manager/
16:01:37 <trown> thanks jpena
16:02:00 <trown> #endmeeting