15:00:55 <trown> #startmeeting RDO meeting - 2017-04-19
15:00:55 <openstack> Meeting started Wed Apr 19 15:00:55 2017 UTC and is due to finish in 60 minutes.  The chair is trown. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:57 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:00 <amoralej> o/
15:01:00 <openstack> The meeting name has been set to 'rdo_meeting___2017_04_19'
15:01:05 <dmsimard> chandankumar: we actually mute zodbot now
15:01:07 <trown> chandankumar: I think we have just been using openstack bot
15:01:08 <dmsimard> at least afaik
15:01:12 <chandankumar> ack
15:01:14 <chandankumar> \o/
15:01:23 <chandankumar> jpena: jruzicka number80 meeting time
15:01:28 <trown> #topic roll call
15:01:30 <jschlueter> o/
15:01:36 <jruzicka> o/
15:01:44 <trown> #chair jschlueter amoralej dmsimard chandankumar jruzicka
15:01:45 <openstack> Current chairs: amoralej chandankumar dmsimard jruzicka jschlueter trown
15:02:19 <myoung> o/
15:02:24 <trown> #chair myoung
15:02:24 <openstack> Current chairs: amoralej chandankumar dmsimard jruzicka jschlueter myoung trown
15:02:26 <jpena> o/
15:02:35 <trown> #chair jpena
15:02:35 <openstack> Current chairs: amoralej chandankumar dmsimard jpena jruzicka jschlueter myoung trown
15:03:23 <trown> #topic Separate module name suggestion for moving rdoinfo parser from rdoinfo/rdopkg into a separate module [ https://etherpad.openstack.org/p/rdoinfo-module ]
15:03:28 <dougbtv> o/
15:03:34 <trown> #chair dougbtv
15:03:35 <openstack> Current chairs: amoralej chandankumar dmsimard dougbtv jpena jruzicka jschlueter myoung trown
15:03:50 <trown> jruzicka: this is your topic?
15:04:03 <jruzicka> trown, this was discussed last meeting
15:04:07 <jruzicka> not sure what this means :)
15:04:12 <jruzicka> let me find the etherpad
15:05:13 <amoralej> jruzicka, iiuc, we'll have rdoinfo for pure yaml files
15:05:31 <jruzicka> rdoinfo will contain data only (yaml)
15:05:37 <amoralej> and a separated module to parse them with the functionalities currently in rdoinfo + rdopkg?
15:05:53 <jruzicka> and separate rdoinfo/distroinfo module will contain functionalities currently in rdoinfo + rdopkg
15:05:54 <chandankumar> jruzicka: etherpad link on rdo meeting etherpad
15:06:04 <jruzicka> oh right :)
15:06:13 <trown> also in the topic ^ :P
15:06:26 <jruzicka> no news to report on this so let's move along ;)
15:06:27 <chandankumar> as discussed from last night, i have summerized the stuff on an email on rdolist
15:06:41 <chandankumar> jruzicka: https://etherpad.openstack.org/p/rdoinfo-module
15:06:50 <chandankumar> s/night/meeting
15:07:06 <jruzicka> chandankumar, thanks :)
15:07:07 <amoralej> jruzicka, we can move ahead with the deps.yaml before this change, right?
15:07:22 <jruzicka> amoralej, yup, on it
15:07:26 <amoralej> ok
15:07:36 <amoralej> i need to send a new ps
15:08:20 <jruzicka> right
15:09:23 <trown> are we going to brainstorm a name?
15:09:32 <trown> that is what I understood the topic
15:09:47 <trown> or just move on?
15:09:48 <number80> brainstrowning!
15:09:48 <jruzicka> possibly, but we're missing many brains due to PTOs
15:10:04 <trown> k moving on it is :P
15:10:10 <jruzicka> yup :)
15:10:15 <trown> #topic FreeIPA Trystack migration help/news. [leanderthal] [carried from previous meeting]
15:10:20 <trown> #chair number80
15:10:20 <openstack> Current chairs: amoralej chandankumar dmsimard dougbtv jpena jruzicka jschlueter myoung number80 trown
15:10:47 <trown> anything to report here?
15:10:56 <chandankumar> trown: it is also taken from previous meeting, leanderthal|afk is leading the effort, she seems to be afk.
15:11:10 <chandankumar> anyone from trystack team may highlight on it.
15:11:15 <trown> k, we can come back to it if there is some news about it
15:11:31 <trown> #topic RDO CI can it be adjust to at least keep the logs from the last passing promotion job and sub-jobs? (jschlueter)
15:11:31 <chandankumar> wfoster: ^^
15:12:26 <trown> jschlueter: we can only set a number of days for jenkins to keep job history
15:12:56 <trown> so we could up the number of days, but we couldnt dynamically say keep job history until a pass
15:14:47 <trown> https://github.com/rdo-infra/ci-config/blob/3b4b72572416d744a87d50b17df23f3caac3f490/jenkins/jobs/defaults.yml#L15-L16 is where it would be changed
15:15:05 <trown> along with below that on 27-28
15:15:21 <myoung> perhaps we should investigate keeping job history externally, or internally set up something to scrub ci.centos.  I think not being able to see jenkins/job history for say, an entire release is problematic for trend/historical analysis alone
15:16:12 <trown> I do not see any reason to have only 5 days of history above... so if someone wanted to submit a patch to up that to 31 like the rest of our jobs, I think that is fine
15:16:14 <myoung> e.g. how many jobs did we run in ocata?  how many failed?  how does that compare to newton?  et
15:16:15 <myoung> etc
15:16:39 <amoralej> it seems rdo-delorean-promote-master has been removed?
15:16:42 <trown> myoung: ya that seems like something would need to be set up to poll jenkis
15:16:45 <myoung> (aside from changing defaults)
15:16:56 <myoung> trown: aye
15:18:55 <trown> amoralej: hmm ya that is my fault... I thought that was gone from jjb with move of weirdo jobs to master-current-tripleo
15:19:24 <trown> amoralej: repushing jjb will bring it back
15:19:29 <amoralej> trown, it's not merged yet, https://review.rdoproject.org/r/#/c/6130/
15:19:48 <amoralej> i have no problem, i proposed it :) but there was not agreement on it
15:20:23 <trown> amoralej: ya we were cleaning up the image server, and I for some reason thought that patch merged
15:20:45 <trown> amoralej: anyways, we can either merge the patch or repush jjb if we arent going to merge the patch
15:21:00 <amoralej> i'd repush jjb
15:21:25 <trown> k
15:21:56 <amoralej> thx
15:22:08 <trown> anything else on the logs topic? jschlueter?
15:23:00 <amoralej> iirc, jenkins maintains logs for last green job, main problem is that it doesn't check status of pipeline, only isolated job
15:23:39 <dmsimard> another thing to consider
15:23:43 <trown> hmm. maybe that is true if you dont set logrotate
15:23:51 <trown> https://github.com/rdo-infra/ci-config/blob/3b4b72572416d744a87d50b17df23f3caac3f490/jenkins/jobs/defaults.yml#L15-L16 though explicitly keeps 5 days
15:23:53 <dmsimard> is that we're not really doing a good job at purging old logs from artifacts.ci.centos
15:24:05 <dmsimard> so even though the job is no longer there, the logs probably are
15:24:34 <trown> well the job he refrenced in the etherpad actually would not push logs to artifacts
15:24:38 <jschlueter> trown: no just a though ... I thought there should be a way for a parent to keep it's children jobs around ... but haven't looked
15:24:41 <trown> it is just the get-hash script
15:25:16 <trown> jschlueter: just need to bump the number of days on https://github.com/rdo-infra/ci-config/blob/3b4b72572416d744a87d50b17df23f3caac3f490/jenkins/jobs/defaults.yml#L15-L16 to 31 like the other jobs
15:25:27 <trown> jschlueter: and then again on line 27-28
15:25:31 <trown> ok. moving on
15:25:40 <trown> #topic TestDay Pike.1 - last Master RDO Promotion CI was 16 days ago? what is plan for a hash to use for test day?
15:26:42 <trown> the current-tripleo hash is working for everything except ci.centos jobs
15:27:14 <trown> it works on my local libvirt setup, the weirdo and packstack jobs, and in all of the upstream tripleo jobs
15:27:18 <jschlueter> so what is plan for PIKE.1 Testday ?
15:27:22 <trown> I think we should just use that for testday
15:27:48 <trown> I am trying to figure out why it is not working in ci.centos, but it seems resource related
15:27:52 <trown> ie random timeouts
15:27:55 <amoralej> pipeline is running now
15:28:00 <amoralej> any hope it may pass?
15:28:11 <trown> amoralej: probably depends on if we get the fast nodes
15:29:02 <amoralej> so, if https://ci.centos.org/view/rdo/view/promotion-pipeline/job/rdo_trunk-promote-master-current-tripleo/40/ passes we could promote the has for current-passed-ci also
15:29:11 <amoralej> s/has/hash/
15:30:23 <trown> is anyone here?
15:30:32 <myoung> yes
15:30:51 <dmsimard> I am, trying to think of something
15:30:52 <amoralej> \o/
15:31:17 <dmsimard> trown: so earlier when you said that you can't reproduce locally, it means that for a given hash that you tried, it also passed ?
15:31:42 <trown> dmsimard: ya current-tripleo works for me locally
15:32:02 <amoralej> that's the one being tested right now
15:32:04 <dmsimard> all weirdo jobs are currently passing, if we're confident that we're hitting "infrastructure" issues we can promote manually
15:32:18 * trown votes for ^
15:32:19 <dmsimard> it sort of sucks but what do you want
15:32:35 <dmsimard> trown: was the plan to move oooq jobs to rdo-cloud ?
15:32:40 <trown> the only possible impact is quickstart gates failing
15:33:04 <myoung> if we a.) update promotion log, b.) ensure oooq image is existing on images.rdoproject, rdophase2 can give additional data points on faster hw (skipping ahead to next topic)
15:33:09 <trown> but we have lots of gates that are not on ci.centos now, so that is less of an impact
15:33:19 <amoralej> +1 to manual promotion if only oooq fails with infra issue
15:33:31 <dmsimard> amoralej: look at the past few runs, only the oooq jobs have failed
15:33:52 <trown> dmsimard: eventually we will run some ovb jobs there, but we will always need some libvirt jobs
15:34:14 <amoralej> in fact, last run of rdo-delorean-promote-master passed one of the oooq jobs, only one failed
15:35:00 <dmsimard> trown: ok
15:35:34 <trown> ok agreed to manually promote current-tripleo hash?
15:36:09 <dmsimard> It looks like we've agreed to manually promote https://trunk.rdoproject.org/centos7-master/current-tripleo/ ( /b4/97/b49764cdc7eb0057677efea224603ddf6d4b42c0_ad1f2ce1 )
15:36:22 <dmsimard> trown: can you take care of the image upload and I'll do the dlrn hash symlinks ?
15:36:36 <trown> dmsimard: ya
15:36:59 <dmsimard> #action dmsimard to promote current-passed-ci and current-tripleo-rdo to /b4/97/b49764cdc7eb0057677efea224603ddf6d4b42c0_ad1f2ce1 manually
15:37:04 <trown> #agreed manually promote  https://trunk.rdoproject.org/centos7-master/current-tripleo/ ( /b4/97/b49764cdc7eb0057677efea224603ddf6d4b42c0_ad1f2ce1
15:37:22 <trown> #action trown update image symlinks to point to  /b4/97/b49764cdc7eb0057677efea224603ddf6d4b42c0_ad1f2ce1
15:37:56 <trown> I am skipping the other topic that is a duplicate of this
15:37:56 <trown> #topic DLRN API: call for users/testers
15:38:21 <jpena> that's mine
15:38:31 <myoung> moment...are we promoting both community and tripleo jobs?
15:38:36 <myoung> or just community?
15:38:41 <amoralej> both
15:38:55 <myoung> ok...so https://trunk.rdoproject.org/centos7-master/promote-current-tripleo-rdo.log + images then
15:39:09 <amoralej> + symlink
15:39:20 <myoung> (not just symlinks) - so rdophase2 should pick it up, and we'll look to that hash for next osp12 imports
15:39:29 <apuimedo> jpena: which repo should I install in centos 7 to be able to install the stuff from the delorean repos? for stuff like eventlet
15:39:31 <amoralej> yes, we'll update the log file also
15:39:37 <myoung> thanks
15:40:37 <amoralej> jpena, so what's the plan for dlrn api?, what's the best way to integrate it in jobs?
15:40:51 <jpena> we have the DLRN API running on the internal build server
15:40:52 <amoralej> there is a client library, right?
15:41:13 <jpena> and now we have a client library and ansible module, both available from https://github.com/javierpena/dlrnapi_client
15:41:22 <jpena> (soon to be migrated into gerrit)
15:41:27 <dmsimard> jpena: let's move that to rdo-infra ?
15:41:30 <amoralej> ok
15:41:58 <jpena> the next step would be to get it tested in actual CI jobs
15:42:14 <jpena> and that's why I'm calling out for testers :)
15:42:19 <dmsimard> jpena: to test itself ? or to actually use
15:42:43 <jpena> dmsimard: the client/module are tested and they work, but we should start using it in some jobs
15:43:39 <jpena> actually, while creating the ansible module I discovered a bug in the DLRN API, looking for reviews at https://review.rdoproject.org/r/6147
15:44:00 <myoung> is it a goal that rdophase2 jobs should be able to use / benefit from dlrn api, or is it a requirement to be in the ci.centos sandbox to play/use for now?
15:44:24 <jpena> for now, the API can only be enabled in the ci.centos network
15:44:28 <dmsimard> I'm not familiar with phase names, rdo phase 2 is rdo on rhel ?
15:44:36 <jpena> so you need to be inside to use it
15:44:46 <myoung> rdophase2 is centos images running internally in bare metal and faster hw
15:44:54 <myoung> in addition rdophase2 has RDO on RHEL
15:45:00 <jpena> as soon as we can use the RDO Cloud as productive, the API will be exposed publicly
15:45:12 <myoung> i download images when they are promoted and run them, *in addition to* building RDO on RHEL
15:45:49 <myoung> centos variants include a virt-min, bare metal undercloud, and 2 full BM configs (rlandy's bond with vlan + single nic vlan)
15:46:14 <dmsimard> right, okay
15:46:22 <myoung> we were running centos virt-ha (192 gb ram, 40 core) but disabled it, we can turn it back on if it's useful.
15:46:33 <rdogerrit> Ihar Hrachyshka proposed openstack/neutron-lbaas-distgit: Synced most minimal dependencies with upstream requirements.txt  https://review.rdoproject.org/r/6291
15:46:34 <rdogerrit> Ihar Hrachyshka proposed openstack/neutron-lbaas-distgit: Require same major version of neutron  https://review.rdoproject.org/r/6292
15:47:02 <myoung> the idea is that we can provide additional coverate *in the promotion pipeline* and report back.  RDO on RHEL is a canary in coal mine for imports
15:47:24 <dmsimard> I think the plan is to re-evaluate promotion criterias based on requirements and then put that into practice with dlrn api, we won't need single pipelines anymore and any job can in theory "participate" in the vetting process of a "symlink"
15:47:37 <myoung> +2
15:47:51 <dmsimard> and since jobs will be running against the same hashes, in theory one job result could be used for more than one "symlink"
15:48:10 <myoung> we have jobs running now, they just happen after "promotion" in RDO (rdophase1).  all logs from rdophase2 already go to thirdpartylogs
15:49:13 <trown> anything else on this topic?
15:49:19 <dmsimard> not for me
15:49:29 <dmsimard> have to step away to pick up kid at school for lunch, brb
15:49:45 <trown> #topic Anouncements
15:49:58 <trown> #info Pike Milestone 1 RDO test day is on April 20, 21 , 2017
15:50:04 <trown> that is tomorrow and friday
15:50:16 <trown> #info Fill the RDO contributor Survey: https://goo.gl/forms/SYVo9gbacsMfe8Ov2
15:51:10 <trown> #topic chair for next meeting
15:51:18 <trown> who wants it :)
15:51:57 <amoralej> i can do it
15:52:14 <trown> #action amoralej to chair next meeting
15:52:26 <trown> thanks amoralej
15:52:27 <trown> #topic open floor
15:56:51 <amoralej> nothing from my side
15:57:59 <trown> k
15:58:03 <trown> #endmeeting