Thursday, 2017-03-02

*** jwcroppe has joined #openstack-powervm00:28
*** thorst has joined #openstack-powervm00:31
*** thorst has quit IRC00:36
*** thorst has joined #openstack-powervm00:41
*** seroyer has joined #openstack-powervm00:54
*** thorst has quit IRC01:01
*** smatzek_ has joined #openstack-powervm02:09
*** smatzek_ has quit IRC02:09
*** smatzek_ has joined #openstack-powervm02:10
*** smatzek_ has quit IRC02:22
*** seroyer has quit IRC02:41
*** esberglu has joined #openstack-powervm02:43
*** esberglu has quit IRC02:48
*** thorst has joined #openstack-powervm03:50
*** thorst has quit IRC03:50
*** thorst has joined #openstack-powervm04:32
*** thorst has quit IRC04:32
*** esberglu has joined #openstack-powervm04:32
*** esberglu has quit IRC04:36
*** thorst has joined #openstack-powervm05:03
*** thorst has quit IRC05:16
*** k0da has joined #openstack-powervm05:34
*** k0da has quit IRC05:54
*** esberglu has joined #openstack-powervm06:21
*** esberglu has quit IRC06:26
*** edmondsw has joined #openstack-powervm06:32
*** edmondsw has quit IRC06:56
*** edmondsw has joined #openstack-powervm07:24
*** edmondsw has quit IRC07:29
*** edmondsw has joined #openstack-powervm07:29
*** edmondsw has quit IRC07:34
*** esberglu has joined #openstack-powervm08:10
*** esberglu has quit IRC08:15
*** k0da has joined #openstack-powervm08:41
*** thorst has joined #openstack-powervm09:13
*** thorst has quit IRC09:18
*** BorD__ has quit IRC09:22
*** BorD__ has joined #openstack-powervm09:24
*** esberglu has joined #openstack-powervm09:59
*** esberglu has quit IRC10:03
*** k0da has quit IRC10:36
*** thorst has joined #openstack-powervm11:14
*** thorst has quit IRC11:19
*** edmondsw has joined #openstack-powervm11:20
*** edmondsw has quit IRC11:25
*** k0da has joined #openstack-powervm12:12
*** smatzek_ has joined #openstack-powervm12:20
*** smatzek_ has quit IRC12:21
*** smatzek_ has joined #openstack-powervm12:21
*** thorst has joined #openstack-powervm12:45
*** jwcroppe has quit IRC13:08
*** seroyer has joined #openstack-powervm13:08
*** jwcroppe has joined #openstack-powervm13:23
*** esberglu has joined #openstack-powervm13:24
*** jpasqualetto has joined #openstack-powervm13:24
openstackesberglu: Error: A meeting name is required, e.g., '#startmeeting Marketing Committee'13:30
esberglu#startmeeting powervm_ci_meeting13:30
openstackMeeting started Thu Mar  2 13:30:21 2017 UTC and is due to finish in 60 minutes.  The chair is esberglu. Information about MeetBot at
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.13:30
openstackThe meeting name has been set to 'powervm_ci_meeting'13:30
esberglu#topic OOT CI13:33
esbergluOut of tree CI is still looking good13:33
esbergluAll of the compute nodes are back online, neutron and ceilometer are enabled on the silent pipeline again13:33
esbergluI have patches out for a few things13:35
esbergluOne to only enable CI runs on the branches we want13:35
esbergluAnd one to use the pvc-images server13:36
esberglu4935 and 492313:36
efriedYou got that SSP sorted?13:36
esberglu#topic IT CI13:38
esbergluIn-tree CI is also looking good13:38
esbergluNot much action here until we start getting changes merged13:38
esbergluI also have a change up13:39
esbergluWhich lets us add additional tests to the whitelist. I want to deploy with that on staging today13:39
efried[I +1ed 4935 and 4923 - would like thorst or adreznec to approve.]13:40
esbergluAnd continue work I started determining which tests we can add to the IT whitelist for some of our later changesets13:40
efried[ditto 4918]13:41
efriedWhy are we patching in the power_opts stuff?13:41
efriedIs that just needed for cherry-pick compatibility purposes?13:41
thorstI thought to help with stability (or a thought that it might help with stability)13:42
efriedWhere did we land on the idea of including that in
esbergluYeah what thorst said13:43
thorstwe didn't...yet...because of the whole hubub getting actually working13:43
efriedI just don't like the idea of piling on some nontrivial subset of the (.5 - .4) delta.13:45
efriedIt may make the CI look better to the outside world, but it reduces my confidence in the sanity of what we're actually testing.13:45
efriedDo we have any actual evidence or other reason to believe these patches will add stability, or is it just a hunch?13:46
thorstadreznec: did the release go out?  Or can we add this bit in?13:47
esbergluRelease went out I believe13:47
efriedHave we proposed the global reqs update to pike yet?13:48
efriedor was tonyb going to do that?13:48
*** tjakobs_ has joined #openstack-powervm13:49
esbergluNot sure13:51
esbergluDoesn't look like it13:51
efriedI'll do it.13:53
efried#action efried to propose requirements bump13:53
esbergluSo do we want to remove the patches we are currently piling onto for in tree?13:53
esbergluFor the reason efried detailed above?13:54
efriedUpdate: requirements project is already done:
*** tjakobs_ has quit IRC13:56
efriedesberglu I'd say just kill 4931 and I'll be satisfied.13:58
efriedThose are straight up .5 changes.13:59
efriedThey're not going to lend stability or fix any bugs.13:59
efriedAnd we're already getting burn-in on them with OOT.13:59
esbergluOkay that's good with me13:59
esbergluAnything have final topics?14:00
efriedWe should try to keep in-tree as close as possible to the version in global requirements.14:00
esbergluYes I agree14:00
*** tblakes has joined #openstack-powervm14:00
esbergluLast thing I had was OSA CI. Still working on rebasing the playbooks for that. Nothing else to report14:01
efriedadreznec note -1 on 493214:08
efried(You can mark that down as me disagreeing with thorst)14:08
adreznecOops, for some reason my calendar invite for this didn't pop up. What's broken on 4932?14:08
adreznecThe g-r bump already went out btw14:08
adreznecThey pushed it in last night and bumped pbr to >2 after that14:10
adreznecwhich broke a bunch of other projects thanks to that hacking requirement14:10
adreznecWe should be good to go on that front with 4932 and the correspoding g-r bumps on the powervm projects14:10
efriedadreznec Take a look at my comments.  If they're all bogus, feel free to merge.14:11
adreznecWill do14:11
efriedBut if we're aligning with g-r, I don't see a reason for our pbr req to be downrev.14:12
*** mdrabe has joined #openstack-powervm14:12
efriedAnd the lxml thing bugs me.  I think we should be able to keep our floor higher.14:12
adreznecYeah, that wasn't merged yet when I proposed that14:12
adreznecand hmm ok, I'll have to look at the other comment14:12
efriedAny other topics for the CI meeting?14:13
adreznecDoesn't sound like it14:14
efriedesberglu call it?14:22
openstackMeeting ended Thu Mar  2 14:22:44 2017 UTC.  Information about MeetBot at . (v 0.1.4)14:22
openstackMinutes (text):
*** edmondsw has joined #openstack-powervm14:32
efriedesberglu Use this one to test the branch selection once that's active:
efriedExpect that guy to run (and pass) OOT, but not run in-tree.14:36
*** tblakes has quit IRC14:39
*** esberglu has quit IRC14:43
*** nbante has joined #openstack-powervm14:47
*** tblakes has joined #openstack-powervm14:57
*** esberglu has joined #openstack-powervm15:01
*** dwayne has joined #openstack-powervm15:03
*** tjakobs_ has joined #openstack-powervm15:37
efriedthorst You want those to be warnings or info?16:17
thorstefried: looking...I just know they were different16:28
thorstinfo seems fine to me16:28
thorstthey're part of an exception path, but they're info about the exception path.16:29
*** mdrabe has quit IRC16:31
*** mdrabe has joined #openstack-powervm16:36
*** seroyer has quit IRC16:38
*** seroyer has joined #openstack-powervm16:51
*** apearson has joined #openstack-powervm16:58
*** jpasqualetto has quit IRC17:23
*** jpasqualetto has joined #openstack-powervm17:24
thorstadreznec: around?17:24
adreznecthorst: sup17:24
thorstyou said earlier that vxlan tunnels can be multicast or you have docs on that?17:25
thorstgoogle is not finding it after 5 min of searching...figured I'd ask as I'm probably using the wrong wording  :-)17:25
adreznecwhich part are you looking for docs on? unicast or multicast?17:27
thorstI want to try unicast.17:27
thorstwe're having issues with multicast17:27
adreznecHere's some stuff on a multiple server unicast setup -
adreznecHad been looking at something else too, trying to find the bookmark17:29
thorstis this just containers or openstack generic as well?17:29
adrezneci've never used it with OpenStack, only manual setup, so not sure17:30
adreznecYou definitely don't need to be using containers with it17:30
adreznecIt's just using the kernel vxlan support17:30
adreznecIdk how flexible the neutron config/support for handling how tunnels are set up is though17:31
thorstahh, I was under impression that you had done it with openstack17:35
*** jpasqualetto has quit IRC17:35
openstackgerritEric Fried proposed openstack/nova-powervm master: In-tree backports: TaskFlow base
efriedthorst and ^^17:43
*** jpasqualetto has joined #openstack-powervm17:48
efriedthorst 4919 should be ready for your +2 now.17:50
thorstefried: first two looked good18:06
thorstlooking at 491918:07
*** apearson has quit IRC18:24
*** nbante has quit IRC18:25
*** edmondsw has quit IRC18:29
*** edmondsw has joined #openstack-powervm18:32
*** edmondsw has quit IRC18:36
*** thorst has quit IRC19:09
*** thorst has joined #openstack-powervm19:10
openstackgerritMerged openstack/nova-powervm master: In-tree backports: TaskFlow base
*** edmondsw has joined #openstack-powervm19:14
*** thorst has quit IRC19:14
*** edmondsw has quit IRC19:47
openstackgerritEric Fried proposed openstack/nova-powervm master: Move _get_vg_uuid to pypowervm's find_vg
*** edmondsw has joined #openstack-powervm19:50
*** thorst has joined #openstack-powervm20:35
thorstefried: good with me pushing up 4919?20:40
thorstI just +2'20:40
efriedthorst Yup, make it happen.20:40
thorstcool...on it20:40
thorstymadhavi will have a nice surprise tomorrow morning when she see's its merged.20:41
tonybefried: I offered to do it if y'all wanted to but I ddin't get a frim "go ahead" so I didn't20:48
efriedtonyb Somebody did.20:49
tonybefried: okay20:49
efriedtonyb Right?20:49
tonybefried: Oh yeah that thing.  I thought you were asking about auto-syncing g-r into the nova-powervm repo20:50
tonybefried: like nova do20:50
efriedtonyb It would be really neat if we could do that (for nova-powervm and for pypowervm also).20:50
efriedWe discussed it briefly.  A question of putting the resources into it.20:51
tonybefried: pypowervm is on GH so we can't do it there20:51
efriedWell "we" can ;-)20:51
tonybefried: but I can write a script that will propose a PR for y'all but it wont be the same20:51
tonybefried: I'd be happy do the scripting if y'all have the permissions20:52
efriedtonyb No, for pypowervm we would probably write the script to propose to our internal gerrit.20:52
efriedor jenkins job, more likely.20:52
tonybefried: Oh I can do that assuming your internal gerrit has .gitreview stup right20:53
efriedtonyb Yeah; or it can be thus finagled.20:53
tonybefried: so I'll do the nova-powervm things and when you're ready I can help you with the pypowervm ones too but y'all will need to work out how you merge them20:54
tonybefried: sound reasonable?20:54
efriedadreznec esberglu Have either of you already made inroads on this front (automatic proposal of change sets to bump pypowervm requirements in line with openstack global reqs)?20:54
efriedtonyb Absolutely!  Thank you for volunteering!20:55
tonybefried: It's what I do20:55
efriedInternally, I'm sure we'll be happy to review & merge them manually for pypowervm.20:55
efriedAt least initially.  We'll see how much volume it generates.20:55
tonybefried: Just run it (manually) once a week the load will be minimal20:58
openstackgerritTony Breeds proposed openstack/nova-powervm master: Manual sync with openstack/requirements repo
tonyb^^ step one, just running tests on steps 2 and 321:06
efriedtonyb Thanks, I added our core team as reviewers.21:06
tonybefried: cool21:07
*** smatzek_ has quit IRC21:10
*** kriskend has joined #openstack-powervm21:10
*** jpasqualetto has quit IRC21:11
efriedtonyb Help me understand this a little bit - does this review come from some kind of bot that's controlled outside of our purview?21:12
efriedCan we customize it a little bit - e.g. have exclusions/exceptions?21:12
tonybefried: Yes21:12
tonybefried: everytime there is a change in openstack/requirements:global-requirements.txt that affect you you'll see a bot generated chnage very similar to the one I just created21:13
tonybefried: let me find you some examples21:13
tonybefried: that's the nova ones21:14
efriedThe thing is, because of the way our development stream coordinates between nova-powervm and pypowervm, we (at least) would need to exclude pypowervm from any bot updates.21:14
tonybefried: It's an all or nothing thing21:14
tonybefried: *you* will only update pypowervm in openstack/requirements:global-requirements.txt when you're ready21:15
tonybso it wont chnage unless you do it over there21:15
tonybefried: so when y'all release a new pypowervm it will within 24hours get added to upper-constratints.txt and you can start using it in the gate21:16
tonybefried: once you're happy that it works if needed you can do the raise the minimum in openstack/requirements:global-requirements.txt if needed21:16
tonybefried: It'll be good practise for when the powervm driver is merged into nova as you'll have *exactly* the same workflow there21:18
*** kjw3 has joined #openstack-powervm21:19
*** tlian has joined #openstack-powervm21:20
efriedtonyb Okay, I still don't fully grok who owns the bot or how it's run - but presumably human interaction is still required before the auto-proposed changes actually merge.21:25
tonybefried: requirement/Infra own the bit21:26
efriedtonyb And presumably it's a change set like any other, so if we need to go in and tweak something before we merge it, we can still do that.21:26
efriedtonyb Okay, then ^^ is especially important.21:26
tonybefried: Yeah core reviewers still need to merge the changes21:26
tonybefried: You can change it *BUT* the bot is dumb so if it suggest $thing and you say no, it'll just re-propose the same thing next time21:27
tonybefried: when the bot isn't doing what you want the *souce* data is openstack/requirements:global-requirements.txt21:27
efriedOh.  Next time that line is touched, or next time it runs?21:27
tonybefried: next time openstack/requirements:global-requirements.txt is touch the bot will re-run21:27
efriedYeah, that'd be fine.  As long as it doesn't just keep generating reviews for changing that one line over and over.21:28
efriedCause we'll routinely be removing the constraints from pypowervm in the master branch for the forseeable future.21:28
*** tlian has quit IRC21:29
tonybefried: that last line worries me21:29
tonybif you're rapidly changing things in *requirements.txt then a) you don't wnat auto sync and b) you're not really ready for integration in nova21:30
*** edmondsw has quit IRC21:31
thorstsorry...I've been catching up here.21:32
efriedtonyb Well, it's essentially because we don't support nova-powervm (out-of-tree compute driver) for a given release until that release is cut; so we're developing pypowervm in parallel.21:32
thorstthe stuff in nova is not changing rapidly at all...21:32
efriedThe in-tree driver is going to lag for a while, and will always be against an official release of pypowervm.21:32
*** jpasqualetto has joined #openstack-powervm21:33
thorstyeah, I want to revisit that for the out of tree though.21:33
thorstI think adreznec, efried and I should chat...I'm not sure oot needs that approach.21:34
*** edmondsw has joined #openstack-powervm21:35
efriedAlso tonyb FYI, the goal is for our g-r pypowervm version to change once per release.  Unless something extreme happens, like this pbr thing.21:35
tonybefried: okay, that's fine but you're free to chnage it on *master* anytime up the RC1 so you don't need to stress to much about the 1 pypowervm per cycle21:36
thorstthat helps.21:38
efriedtonyb Except we didn't really want to be cutting a new pypowervm that frequently.  But realistically, we're only talking like once a month.  So if that kind of velocity is acceptable in the master requirements project...21:38
efriedYeah, what thorst said.21:38
tonybefried: Yeah until RC1 once a month is fine21:38
tonybefried: You only need to update openstack/requirements:global-requirements.txt when you *require* a new feature from ppypowervm.  Not just 'cause it was released21:39
efriedtonyb Understood.  It's just... that'd be the only reason we would cut a new release.21:40
thorstefried: RC1 is Aug 721:40
efried...other than normal release boundaries.21:40
tonybefried: okay that's somewhat strange but it's up to y'all21:40
efriedtonyb This may be TMI, but we try to hide as much PowerVM-specific esoterica in pypowervm so the nova cores don't have to look at it and try to grok it when looking at our community code.21:41
thorstmeh, that's not TMI.  :-)21:42
efriedOkay, but this might be...21:42
tonybefried: Yeah that's fine.  You'll just find it takes longer for things to land if  you can't work with multiple releases of pypowervm21:43
* thorst drum roll21:43
efriedSo when we need to support a new feature, code goes into multiple parts of the stack, sometimes as low as the Power firmware, and has to trickle up until it eventually gets into the community code.21:43
efriedSo if we were to skip pypowervm in that chain, there would be some fairly incomprehensible stuff going into nova(-powervm)21:43
tonybas you'll tie you nova features/bugfixes to the pypowervm cycle which you've just said will be 1/month so you'll only have say 5 windows to merge code into nova per cycle21:44
efriedThat should be more than enough.21:44
tonybas opposed to relgualr small increments21:44
efriedIt's better than having to lag a whole release or two.21:44
tonybokay if you're cool with that21:44
efriedWe are if y'all are.21:45
tonybI guess if it's a problem once your integrated you can address it internally and alter the process if needed21:45
* tonyb needs to take kids to school 21:46
efriedThanks for all the help today tonyb21:46
tonybefried: np21:46
openstackgerritMerged openstack/nova-powervm master: In-tree backports: Redundant args
openstackgerritEric Fried proposed openstack/nova-powervm master: Manual sync with openstack/requirements repo
efriedthorst adreznec esberglu ^^ Manual edit to remove pypowervm version spec.  Do we want to merge this?22:03
efried...assuming it passes our CI22:03
adreznecefried: It should, I pushed up an almost identical change yesterday... and forgot to add reviewers, of course22:06
* adreznec sighs22:06
adreznecI'll abandon mine though22:06
adreznecTony's has the updated pbr>2 req in it22:07
*** thorst has quit IRC22:08
efriedIf I understood correctly, we'll have to edit every time to do this.  Until we get set up for a cadence with real pypowervm sub-releases in cycle.22:08
efriedWhich needs to be on the agenda for an internal discussion soon.22:08
adreznecthorst and I had been pushing them in the past22:08
efriedCause we can't have it take 24 hours of talkytalk every time we need to do that.22:08
adreznecwell most of the talk this time was because we'd been neglecting pypowervm22:09
efriedas far as global requirements?22:09
adreznecnot really anything to do with the *-powervm project requirements22:10
adreznecIt's not an openstack project, so nobody had been keeping it in sync with g-r22:10
adreznecWhich didn't matter until we put it in requirements...22:10
adreznecAnd now all of a sudden it matters22:10
efriedOkay, so hopefully we'll bot that, and not have so much churn next time.22:10
adreznecAh, one other thing from mine that bump that isn't in the one you just edited for nova-powervm was an update to our hacking requirement22:11
adreznecNot sure why that didn't get picked up in the manual sync one...22:12
efriedcould it be the blank line?22:13
adreznecBut hacking<0.12,>=0.11.0 definitely doesn't match
efriedthat would be... weird.22:13
openstackgerritEric Fried proposed openstack/nova-powervm master: Manual sync with openstack/requirements repo
efriedadreznec edited22:14
*** thorst has joined #openstack-powervm22:36
*** tblakes has quit IRC22:37
*** thorst has quit IRC22:41
*** seroyer has quit IRC22:44
tonybefried: for the records here are the 3 parts of the process with appropriate Depends-Ons
*** mdrabe has quit IRC23:06
*** kriskend_ has joined #openstack-powervm23:07
*** dwayne has quit IRC23:08
*** kriskend has quit IRC23:11
*** seroyer has joined #openstack-powervm23:17
*** thorst has joined #openstack-powervm23:17
*** thorst has quit IRC23:19
*** kriskend_ has quit IRC23:25
*** seroyer has quit IRC23:47
*** jpasqualetto has quit IRC23:56
*** jwcroppe has quit IRC23:59
*** jwcroppe has joined #openstack-powervm23:59

Generated by 2.14.0 by Marius Gedminas - find it at!