15:01:01 <PagliaccisCloud> #startmeeting RDO meeting - 2019-02-20
15:01:02 <openstack> Meeting started Wed Feb 20 15:01:01 2019 UTC and is due to finish in 60 minutes.  The chair is PagliaccisCloud. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:05 <openstack> The meeting name has been set to 'rdo_meeting___2019_02_20'
15:01:27 <PagliaccisCloud> #topic rollcall
15:02:15 <baha> o/
15:02:23 <PagliaccisCloud> #chair baha
15:02:23 <openstack> Current chairs: PagliaccisCloud baha
15:03:36 <jpena> o/
15:04:11 <PagliaccisCloud> #chair jpena
15:04:12 <openstack> Current chairs: PagliaccisCloud baha jpena
15:06:24 <PagliaccisCloud> ready to jump into the first topic?
15:07:30 <baha> I'd say go for it
15:07:48 <PagliaccisCloud> #topic A quick question on some jinja2 issues
15:07:52 <PagliaccisCloud> #link https://review.rdoproject.org/r/#/c/18819/
15:08:00 <mjturek> o/
15:08:07 <PagliaccisCloud> @chair mjturek
15:08:12 <PagliaccisCloud> h*ck lol
15:08:16 <baha> It's like we summoned him
15:08:22 <PagliaccisCloud> #chair mjturek
15:08:23 <mjturek> :)
15:08:23 <openstack> Current chairs: PagliaccisCloud baha jpena mjturek
15:09:14 <PagliaccisCloud> alright baha, you're up :D
15:09:59 <baha> This is something that's been causing mjturek and I some confusion. Basically, when the two expressions we put into the upstream-builders .yml file ( {{ ansible_distribution|lower }} {{ ansible_distribution_major_version }} ) are run, rather than getting "centos" and "7", we get "centos" and "" out the other end.
15:10:43 <baha> I'm wondering if it's a syntax error, or if it might be an environment error, or if it's just in the wrong place... or what a proposed solution might be
15:11:28 <mjturek> yeah we could really use a fresh set of eyes here
15:12:26 <jpena> not sure about it (since it's ansible inside jjb), but have you tried putting the whols $WORKSPACE... thing between quotes?
15:12:49 <jpena> like script: "$WORKSPACE .... _major_version }}"
15:13:12 <mjturek> hmmm might be worth a shot!
15:13:42 <mjturek> buuuuut
15:13:43 <jpena> I think jjb did some mangling to curly brackets... That's another alternative
15:14:27 <mjturek> hmmm we'll play around with the parsing I suppose... but not sure why it didn't just use the defaults that baha set
15:14:59 <baha> Yeah, I would have expected the defaults set in the script file in the same patch, or else some strangely garbled URL
15:16:08 <mjturek> let me grab the review.rdoproject.org script it hits
15:17:13 <mjturek> https://github.com/rdo-infra/review.rdoproject.org-config/blob/master/ci-scripts/tripleo-upstream/build-containers-images.sh#L26
15:17:30 <rdogerrit> Sorin Sbarnea proposed rdo-jobs master: Adds tripleo-build-containers-fedora-28 non voting  https://review.rdoproject.org/r/18913
15:18:22 <mjturek> so distro seems to be built wrong
15:18:32 <mjturek> $DISTRO*
15:23:01 <mjturek> it's not ansible
15:23:22 <mjturek> http://paste.openstack.org/show/745473/
15:23:28 <mjturek> output is "centos"
15:24:08 <jpena> mjturek: oh, of course
15:24:20 <jpena> that's using DISTRO_VERSION instead of DISTRO_RELEASE
15:24:25 <jpena> see line 9 and 12
15:24:36 <mjturek> ...oh my gosh
15:24:41 <mjturek> baha: we're dumb ^
15:24:52 <baha> That makes a load of sense
15:25:00 <baha> Thank you very much for the catch jpena
15:25:01 <mjturek> thanks jpena... sorry for that :)
15:25:19 <jpena> no prob, it's the kind of issue that needs a fresh pair of eyes
15:25:26 <PagliaccisCloud> teamwork! \o/
15:25:42 <mjturek> indeed :) thanks for letting us debug in the meeting!!
15:25:48 <rdogerrit> Sorin Sbarnea proposed rdo-jobs master: Adds linting job  https://review.rdoproject.org/r/17691
15:26:49 <PagliaccisCloud> sweet, next topic
15:27:26 <PagliaccisCloud> #topic Hash GC and invalidating hashes less than 60 days old
15:27:37 <amoralej> jschlueter, ^
15:27:52 <PagliaccisCloud> #chair amoralej
15:27:53 <openstack> Current chairs: PagliaccisCloud amoralej baha jpena mjturek
15:29:20 <jschlueter> So if I understand issue correctly, the hash for a promoted hash from earlier in the month was invalidated after we had several promotions and Garbage Collection removed several hashes (real files from last year)
15:29:27 <rdogerrit> Adam Kimball created rdo-infra/ci-config master: Fix DISTRO_VERSION variable for upstream ppc64le containers job  https://review.rdoproject.org/r/18924
15:29:46 <PagliaccisCloud> #chair jschlueter
15:29:47 <openstack> Current chairs: PagliaccisCloud amoralej baha jpena jschlueter mjturek
15:30:17 <amoralej> jschlueter, yes, so current purging mechanism can lead to invalid hash repos
15:30:42 <amoralej> if they are not symlinked to a set of known *protected* symlinks
15:30:43 <jschlueter> I realize the current model works and is intended to cleanup unused builds but it also can result in hashes that are recently created being invalid for any type of regression testing
15:30:59 * jschlueter nods
15:31:36 <amoralej> short term, the easiest way to protect a hash is to create a symlink pointing to it and add it to the list of protected
15:31:50 <amoralej> but yes, having inconsistent hash repos is not good
15:32:00 <amoralej> and something to look for better alternatives
15:32:07 <jschlueter> s/inconsistent/incomplete/
15:32:14 <amoralej> yes, correct
15:32:48 <amoralej> we could look for a more smart cleaning or move to hardlinks
15:32:52 <jschlueter> that's all from me, just wanted to raise the issue
15:33:30 <jpena> #action jpena to groom a DLRN story to avoid unconsistent repos due to purges
15:33:45 <amoralej> thanks jpena for taking care
15:34:42 <amoralej> this may have collateral damage, so we need to be careful
15:35:28 <jschlueter> amoralej, jpena: just a thought can we keep a longer series of symlinks to promoted hashes, that get's pruned as they go past 30 or 60 days?
15:35:49 <jschlueter> or total number of them ... right now we keep promoted hash -1
15:36:20 <jschlueter> maybe keep promoted hash back through -10
15:36:55 <jschlueter> should only be a big difference if we have long time between promotions
15:37:09 <jpena> jschlueter: that we have today (current-tripleo and previous-current-tripleo). It's easy to extend that, for example creating a current-import symlink for the repo being imported now, that only changes once we start a new import
15:37:32 * jschlueter nods
15:40:05 <PagliaccisCloud> ready for open floor?
15:41:23 <PagliaccisCloud> oh wait let's do next week's chair first
15:41:35 <PagliaccisCloud> #topic chair for next meeting
15:42:20 <amoralej> i can take it
15:42:47 <PagliaccisCloud> schweet thanks amoralej
15:43:11 <PagliaccisCloud> #action amoralej to chair next week's meeting
15:43:43 <PagliaccisCloud> #topic Open floor
15:44:40 <mjturek> would be nice if we could get this merged! https://review.rdoproject.org/r/#/c/18924/ (thanks for the quick patch baha)
15:45:07 <baha> It just fixes the variable name. D'oh
15:48:24 <baha> Thank you jpena
15:48:52 <rdogerrit> Merged rdo-infra/ci-config master: Fix DISTRO_VERSION variable for upstream ppc64le containers job  https://review.rdoproject.org/r/18924
15:49:00 <amoralej> apevec, ykarel|pto let's merge https://review.rdoproject.org/r/#/c/18919/
15:50:30 <apevec> yes +2
15:51:52 <chandankumar> ykarel|pto: jpena amoralej https://review.rdoproject.org/r/#/c/18917/
15:53:17 <PagliaccisCloud> anything else before we close out?
15:53:59 <PagliaccisCloud> also don't forget to add notes for anything you'd like to discuss at next week's meeting https://etherpad.openstack.org/p/RDO-Meeting
15:57:11 <PagliaccisCloud> alright, giving you all a few minutes back. see everyone next week
15:57:15 <PagliaccisCloud> #endmeeting