15:00:30 <number80> #startmeeting RDO meeting - 2017-02-01
15:00:30 <zodbot> Meeting started Wed Feb  1 15:00:30 2017 UTC.  The chair is number80. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:30 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:30 <zodbot> The meeting name has been set to 'rdo_meeting_-_2017-02-01'
15:00:31 <openstack> Meeting started Wed Feb  1 15:00:29 2017 UTC and is due to finish in 60 minutes.  The chair is number80. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:33 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:35 <openstack> The meeting name has been set to 'rdo_meeting___2017_02_01'
15:00:37 <number80> #topic roll call
15:00:42 <number80> agenda is here
15:00:45 <number80> https://etherpad.openstack.org/p/RDO-Meeting
15:00:45 <chandankumar> \o/
15:00:51 <number80> please add your items
15:01:13 <number80> #chair chandankumar apevec dmsimard jpena amoralej
15:01:13 <zodbot> Current chairs: amoralej apevec chandankumar dmsimard jpena number80
15:01:14 <openstack> Current chairs: amoralej apevec chandankumar dmsimard jpena number80
15:01:24 <jpena> o/
15:01:39 * apevec only lurking from the other mtg
15:01:40 <rdogerrit> Bernard Cafarelli proposed openstack/networking-sfc-distgit: Add initial spec file  https://review.rdoproject.org/r/4656
15:01:56 <number80> shush
15:02:02 <tosky> o/
15:02:15 <trown> o/
15:02:16 <number80> #chair tosky
15:02:16 <zodbot> Current chairs: amoralej apevec chandankumar dmsimard jpena number80 tosky
15:02:16 <openstack> Current chairs: amoralej apevec chandankumar dmsimard jpena number80 tosky
15:02:20 <number80> #chair trown
15:02:20 <zodbot> Current chairs: amoralej apevec chandankumar dmsimard jpena number80 tosky trown
15:02:21 <openstack> Current chairs: amoralej apevec chandankumar dmsimard jpena number80 tosky trown
15:02:23 <jruzicka> o/
15:02:32 <number80> #chair jruzicka
15:02:32 <zodbot> Current chairs: amoralej apevec chandankumar dmsimard jpena jruzicka number80 tosky trown
15:02:34 <openstack> Current chairs: amoralej apevec chandankumar dmsimard jpena jruzicka number80 tosky trown
15:02:56 <number80> dmsimard: /mode -v doesn't work on zodbot?
15:03:03 <dmsimard> number80: he needs +q
15:03:08 <dmsimard> +q with -v mutes
15:03:12 <dmsimard> +q with +v can talk
15:03:32 <number80> good
15:03:34 <rdogerrit> Harry Rybacki proposed rdo-infra/ci-config: Pass job type to publisher handling collect-docs  https://review.rdoproject.org/r/4529
15:03:43 <number80> next topic
15:03:55 <number80> #topic ci.centos.org and CBS outage announced for March 14th
15:04:03 <number80> https://lists.centos.org/pipermail/centos-devel/2017-January/015646.html
15:04:26 <leifmadsen> o/
15:04:38 <number80> Not much to say, this is due to these services being hosted in the community cage that is being reorganized
15:04:41 <number80> #chair leifmadsen
15:04:41 <zodbot> Current chairs: amoralej apevec chandankumar dmsimard jpena jruzicka leifmadsen number80 tosky trown
15:04:43 <openstack> Current chairs: amoralej apevec chandankumar dmsimard jpena jruzicka leifmadsen number80 tosky trown
15:05:16 <trown> reasonably good timing release schedule wise
15:05:21 <rdobot> [sensu] NEW: master.monitoring.rdoproject.org - check-delorean-master-head-current @ http://tinyurl.com/hcnq3ll |#| Build failure on centos7-master-head/current: keystone, glance, gnocchi, heat, cinder, zaqar, python-gnocchiclient, mistral, murano-dashboard: http://trunk.rdoproject.org/centos7-master-head/report.html
15:05:24 <trown> early in pike release
15:06:05 <number80> Yes
15:06:26 <number80> Then let's move to the next topic
15:06:36 <number80> #topic ci.centos.org cloud and jenkins maintenance window
15:06:45 <number80> https://www.redhat.com/archives/rdo-list/2017-February/msg00000.html
15:06:47 <number80> dmsimard: ^
15:06:53 <dmsimard> o/
15:07:15 <number80> raise your points and ask for help if you need to :)
15:07:25 <dmsimard> There's a maintenance window tonight at 8PM EST/01:00 UTC for updates in ci.centos.org land
15:07:55 <dmsimard> We'll update the Jenkins master as well as the cico cloud
15:08:11 <dmsimard> I would like another window after Ocata is out to upgrade our jenkins slaves to CentOS 7.3
15:08:23 <dmsimard> That's it for me
15:11:11 <chandankumar> number80: next topic?
15:12:29 <number80> yes
15:12:40 <number80> #topic Updates on tempest
15:12:43 <number80> chandankumar: ^
15:13:13 <chandankumar> So we can now use tempest rpm in puppet-tempest with tempest-workspace https://review.openstack.org/#/c/425085/
15:13:42 <chandankumar> As well as we have blog up on how to use new tempest rpm with ocata https://github.com/redhat-openstack/website/pull/817
15:13:51 <chandankumar> thanks to tosky for the blog
15:14:15 <number80> kudos to both of you
15:14:21 <chandankumar> amoralej: has tested new tempest rpm and workflow with packstack, packstack support for tempest rpm is coming soon.
15:14:57 <chandankumar> amoralej: just one query related to packstack tempest support are we going to keep install from source option in packstack or to remove it?
15:15:09 <amoralej> i'd just remove it
15:15:27 <jpena> yes, I'd just switch to rpm install
15:15:33 <amoralej> for ocata and master
15:15:39 <jpena> one can always clone from source and run it manually if needed
15:15:40 <amoralej> we'll keep it as is for newton and mitaka
15:16:00 <chandankumar> amoralej: ack will get a patch up for packstack to install tempest through rpm
15:16:40 <chandankumar> that's it from myside on this
15:16:43 <amoralej> i'll prepare it as soon as i get a slot :)
15:17:18 <myoung> o/
15:17:22 <dmsimard> yeah I think packstack doesn't need to be able to install tempest from source.
15:18:02 <chandankumar> #chair myoung
15:18:02 <zodbot> Current chairs: amoralej apevec chandankumar dmsimard jpena jruzicka leifmadsen myoung number80 tosky trown
15:18:03 <openstack> Current chairs: amoralej apevec chandankumar dmsimard jpena jruzicka leifmadsen myoung number80 tosky trown
15:18:39 <chandankumar> dmsimard: patch is coming for packstack tempest
15:19:09 <number80> good
15:19:13 <tosky> agreed, if all bits are installed from RPMs, tempest should not be different
15:20:49 <chandankumar> number80: i have a followup topic related to tempest in p-o-i
15:20:51 <number80> I guess this is the tricky part: ensuring that everything is installed
15:21:02 <number80> ok next
15:21:08 <number80> #topicViews on adding support of tempest rpm in p-o-i
15:21:11 <number80> #topic Views on adding support of tempest rpm in p-o-i
15:21:18 <number80> continue then :)
15:21:38 <chandankumar> So today i was working on installing tempest from rpm in p-o-i https://review.openstack.org/#/c/427578/
15:22:01 <chandankumar> but as discussed with EmilienM and dmsimard i will be problematic as
15:22:11 <eggmaster> (tempest from rpm \o/)
15:23:20 <chandankumar> we need to pin some commits in puppet-ci and adds depends on patch for testing tempest patches upstream
15:24:07 <chandankumar> I need just views on that whether to add this support or not if adding suport it will affect weirdo jobs also in RDO CI
15:24:35 <chandankumar> dmsimard: jpena amoralej apevec ^^
15:24:38 <dmsimard> adding support for installing from RPMs while still defaulting to source will have no impact
15:24:43 <dmsimard> because the default doesn't change
15:24:56 <amoralej> +1 on that
15:24:59 <dmsimard> the question is how do we handle the different pins that we sometimes need
15:25:14 <dmsimard> i.e, pinning to a commit
15:25:21 <number80> Yep, but it would be nice to have it work from packages and try to change default for Pike
15:25:28 <amoralej> so, when using rpms, pinning should be done at packaging
15:25:48 <dmsimard> number80: default in p-o-i will never change, it's an upstream project and also has jobs on tempest gate
15:25:57 <amoralej> so when deploying from rpm we should never need to do the pin in p-o-i
15:26:19 <number80> I meant for our jobs
15:26:20 <amoralej> dmsimard, but we can deploy from rpm in promotion pipeline, right?
15:26:20 <dmsimard> amoralej: so pin through rdoinfo source-branch ?
15:26:24 <amoralej> yes
15:26:30 <chandankumar> so basically in rdoinfo we pinn at particular version if rdoinfo providea a feature to pin at particular commit that will be great
15:26:35 <jpena> yes, that's what I would do if we had to pin
15:26:38 <dmsimard> amoralej: no, I would not use rpm in the puppet-specific promotion pipeline
15:26:40 <amoralej> and if we need to backport patches we'll carry the patches in distgit
15:26:50 <dmsimard> amoralej: because we could end up promoting something that'd break p-o-i upstream
15:27:23 <amoralej> and in the current-passed-ci pipeline?
15:27:30 <amoralej> because now we have the oposite
15:27:35 <amoralej> we can promote broken tempest
15:27:42 <amoralej> which is not good
15:27:48 <dmsimard> yeah.. it's hard to tell
15:28:09 <dmsimard> trying to picture it in my head but there's quite a few subtle things in there
15:28:17 <amoralej> i think, that could be good, deploying from source in puppet-promotion and from rpm in current-passed-ci, wdyt?
15:28:51 <chandankumar> amoralej: so option is making configurable in run_test.sh to install from rpm or source to satisfy both ci?
15:29:12 <dmsimard> well, we can try and see what happens.. I guess, worst case, we revert back to using from source in current-passed-ci or something
15:29:15 <amoralej> yes chandankumar, who doable it is?
15:29:22 <dmsimard> and then fix whatever is breaking
15:29:40 <dmsimard> amoralej: not much different than any of the toggles that's already embedded in run_tests.sh
15:29:40 <chandankumar> amoralej: as discussing today then that option comes
15:29:54 <dmsimard> amoralej: like tempest_version and manage_repos and others
15:30:26 <dmsimard> I guess I will need to parameterize that option in weirdo
15:31:22 <chandankumar> dmsimard: amoralej i will work on parameterize part let see how it goes.
15:31:28 <amoralej> ok
15:31:46 <chandankumar> that's it from myside on this.
15:32:01 <number80> ok
15:32:16 <number80> #topic New way of tracking issues affecting jobs in RDO CI
15:32:18 <number80> amoralej: ^
15:32:48 <amoralej> yes, that's for awareness, we have changed we we track issues in RDO CI jobs
15:33:01 <amoralej> we have been using etherpad
15:33:06 <amoralej> in the past
15:33:20 <amoralej> now, we are moving to a public trello board
15:33:40 <amoralej> #link https://trello.com/b/WXJTwsuU/tripleo-and-rdo-ci-status
15:34:11 <amoralej> we still have to improve categorization and labels, but it's available already
15:34:12 <chandankumar> dmsimard: what about using this script https://gist.github.com/dims/54b391bd5964d3d208113b16766ea85e from  gate jobs - papercuts for tracking ci failures.
15:34:24 <number80> amoralej: do we have updated the website about it?
15:34:38 <trown> ya, I think this should maintain all the benefits we had with the etherpad with increased readability
15:34:38 <amoralej> nop
15:34:52 <dmsimard> chandankumar: I saw that thread, it's weird
15:34:57 <amoralej> i'll add something in the website number80
15:34:59 <dmsimard> I don't know why they're not doing it through logstash.
15:35:08 <number80> amoralej: thank you :)
15:35:08 <chandankumar> hmm
15:35:12 <dmsimard> I think it's mostly to get data out of zuul at a point in time and "live"
15:35:22 <amoralej> and we used info in the etherpad in the dashboard too
15:35:41 <trown> amoralej: oh true, we need some trello query now
15:35:46 <amoralej> yes
15:36:40 <trown> maybe we can just replace with a static link to the board in case of failure in the short term
15:37:00 <trown> and update to some count of cards in a particular column or particular label later
15:37:05 <amoralej> yes, trown, if we use good lables
15:37:25 <amoralej> probably just link to a filtered url or something like that
15:37:38 <trown> oh, that would be even nicere
15:37:45 <trown> *nicer
15:38:03 <number80> anything else?
15:38:43 <amoralej> not from my side
15:39:20 <trown> same
15:40:35 <number80> ok
15:40:45 <number80> #topic Status of preparation for Ocata release
15:40:54 <number80> two parts
15:41:03 <number80> CI
15:41:05 <number80> https://trello.com/c/Y1nOASYi/554-ci-ocata-release
15:41:17 <number80> amoralej, dmsimard can you fill us about it?
15:41:54 <amoralej> probably dmsimard has more accurate info, but i think currently we are waiting to get new role to build images, right?
15:42:03 <dmsimard> I believe trown will have the reqs in place soon in order to get the ocata pipeline up
15:42:05 <amoralej> to implement promotion pipeline for ocata
15:42:42 <trown> yes, patches are all up just waiting on reviews
15:42:54 <rdogerrit> Harry Rybacki proposed rdo-infra/ci-config: Pass job type to publisher handling collect-docs  https://review.rdoproject.org/r/4529
15:42:56 * trown misses the days of just self-merging all of his own work
15:43:02 <amoralej> :)
15:43:58 <trown> https://trello.com/c/l4IhlYgd/504-q-to-u-create-new-image-building-role-in-tripleo-quickstart-extras-repo is the relevant card
15:44:27 <amoralej> #info information about new image building role in https://trello.com/c/l4IhlYgd/504-q-to-u-create-new-image-building-role-in-tripleo-quickstart-extras-repo
15:44:43 <amoralej> once that is merged, we can merge the new jobs for ocata?
15:44:57 <trown> we could update commit on https://review.rdoproject.org/r/4525 and just use the unmerged quickstart patches for a short time to get ocata pipeline stood up
15:45:21 <amoralej> btw, we are changing job names from ...-delorean-... to ...-rdo-trunk-...
15:45:25 <amoralej> will that be ok?
15:45:33 <trown> the approach of the quickstart patches has been agreed upon, so not a high risk there
15:46:01 <trown> there is nothing in quickstart that relies on job name
15:46:08 <dmsimard> trown: there is iirc
15:46:11 <dmsimard> build_system ?
15:46:28 <trown> oh in the jjb... in the new jjb there is no build_system :)
15:46:33 <dmsimard> I think it's for delorean versus cbs
15:46:35 <dmsimard> or something
15:47:36 <number80> oh
15:47:48 <trown> I got rid of build_system for new jobs... https://review.rdoproject.org/r/4525 ... not sure we can take on changing that for all jobs right now ... maybe we could handle that in https://trello.com/c/Y9GW09Zu/383-move-rdo-ci-specific-code-from-tripleo-quickstart-repo-to-rdo-infra-ci-config
15:49:45 <dmsimard> yeah maybe we can rename jobs at a later time
15:49:49 <trown> so the proposal would be to not create any *new* jobs with delorean in the name, but not to change the names of any old jobs
15:49:58 <dmsimard> there's more than meets the eye on changing job names
15:50:07 <dmsimard> like, jjb won't really rename the jobs, it'll create new ones
15:50:27 <trown> ya, the job names are really vars passed to the ci-scripts... I kind of forgot that part
15:51:07 <number80> next
15:51:09 <trown> the existing ci-scripts in tripleo-quickstart expect build_system, but going forward we will just incorporate build_system into release
15:51:30 <number80> oh you were not done sorry
15:52:32 <trown> I can be :)
15:53:01 <number80> ok
15:53:06 <number80> https://trello.com/c/bPAnPjGn/571-rdo-ocata-branching
15:53:16 <number80> we'll be branching Ocata soon
15:53:45 <number80> I submitted split changes for build system script to build ocata and mitaka (which is problematic for now)
15:53:56 <dtantsur> apevec, we merged the Sem-Ver-tagged commit, Ironic version should be bumped soon
15:54:03 <number80> amoralej and apevec will be taking care of actual branching this time
15:54:10 <number80> amoralej: anything else?
15:54:29 <amoralej> we'll be pinning the non-openstack puppet modules just today
15:54:45 <amoralej> according to last current-passed-ci
15:55:12 <dmsimard> amoralej: huh, why ?
15:55:19 <amoralej> any required update to those modules will need to be managed as a review to rdoinfo
15:55:25 <amoralej> we need to pin for stable
15:55:40 <dmsimard> amoralej: right, but p-o-i hasn't branched yet (and it's puppetfile might still be updated)
15:55:49 <number80> dmsimard: we can't branch a constantly moving target especially when they have no release schedule
15:56:11 <amoralej> dmsimard, reality is that we don't follow puppetfile in p-o-i
15:56:35 <dmsimard> amoralej: reality is that that is what we're actually testing with :/
15:57:04 <dmsimard> So we have potentially outdated RPMs then ?
15:57:34 <dmsimard> wait, didn't EmilienM work on a puppet update constraints to rdoinfo script ?
15:57:45 <amoralej> we test puppet rpms in tripleo jobs
15:57:47 <EmilienM> I did in the past
15:58:33 <amoralej> currently, puppet modules are only pinned for newton
15:59:14 <amoralej> in rdoinfo
16:00:05 <apevec> dmsimard, poi Puppetfile pins to upstream tags, which are sometimes very old
16:00:15 <apevec> NB this is about non-openstack puppet-*
16:00:28 <dmsimard> apevec: that sounds odd to me, they have a script that automatically updates to newer tags
16:00:38 <apevec> yes tags
16:00:50 <apevec> and that's very old in some puppet modules
16:00:51 <dmsimard> ex: https://review.openstack.org/#/c/421723/
16:00:54 <amoralej> many puppet modules don't push tags
16:01:24 <apevec> we should probably revisit that, but right now we are ahead of puppet release tags and need to freeze
16:01:37 <apevec> we did it for newton post GA and it was too late
16:01:42 <dmsimard> hmk, not sure I follow but I trust and have faith in you :)
16:01:52 <apevec> no, that's not good enough :)
16:02:01 <apevec> amoralej, ^ make dmsimard understand :)
16:02:08 <dmsimard> I guess I need a proper example of something that's older in p-o-i
16:02:13 <dmsimard> than what we have in RDO
16:02:30 <number80> please keep it short :)
16:02:39 <number80> we're past the hour
16:02:50 * amoralej looking for examples
16:02:51 <apevec> dmsimard, e.g. puppetlabs-mysql
16:02:54 <dmsimard> It can be off-meeting
16:02:54 <apevec> or apache
16:02:58 <apevec> ack
16:03:08 <number80> ok, let's close the meeting before
16:03:13 <number80> #endmeeting