Friday, 2013-08-09

*** stevebaker has quit IRC00:00
*** zul has quit IRC00:01
sdaguefungi: and cheetah blood00:03
sdagueok, time to make dinner00:04
*** stevebaker has joined #openstack-infra00:06
jgriffithmordred: NM, got it fixed :)00:08
fungijgriffith: was this the cinder/nova dependency installation issue you pasted earlier? what did it turn out to be?00:09
*** krtaylor has quit IRC00:09
jgriffithfungi: yes..00:09
jgriffithfungi: I'm not exactly sure what the root was... but00:10
jgriffithI discovered somethign funky going on with python packages being cached somewhere or something00:10
jeblairsdague: this is going to mess up your plans for the zuul status page:
jgriffithfungi: I manuall rm -rf dist-packages/openstack, setuptools, discover, oslo and pbr00:11
jgriffithincluding eggs00:11
jgriffiththen started over with pip install and low and behold it worked00:11
jgriffithwish I wold've light bulbed exactly what it was, but I *think* it was dist-packages00:12
jgriffitherr... discover00:14
*** vipul is now known as vipul-away00:30
*** cppcabrera has joined #openstack-infra00:32
sdaguejeblair: heh00:33
*** Ryan_Lane has quit IRC00:35
*** vipul-away is now known as vipul00:35
*** david-lyle has quit IRC00:36
*** david-lyle has joined #openstack-infra00:38
fungiaah, so an older cached version, maybe00:40
*** sdake_ has joined #openstack-infra00:40
mroddenits nova client...00:40
fungimrodden: right, i think there's a review open to uncap that in novaclient00:42
mroddenso why is pip installing a version of requests that conflicts with that...00:42
ianwmrodden: you don't have an old version in /tmp/pip-build do you?00:43
fungimrodden: likely pip was also installing something else depending on requests but uncapped, and decided to process that one first00:44
fungiit's fairly mindless in that it doesn't evaluate all transitive dependencies and run a solver to find the intersection00:46
mroddenwow look at that00:46
mroddeni have 2.14.000:47
mroddenand its removed in .100:47
mroddenfungi: yeah thats what i would expect... kind of scary it doesn't00:47
mroddeni wonder if its something with our multiple requirements files00:48
fungiso if you're expecting it to find solutions like yum/apt do, get used to disappointment ;)00:48
mrodden-r requirements, -r test-reqs.txt00:48
mroddeni would think its a pretty well solved problem00:49
mroddeni mean almost every package system has something like that i would think00:49
clarkbI have filed a bug with upstream pip and dstufft thinks it may be fixed in 1.500:49
mroddenif not, could borrow the portage tree one from Gentoo :)00:49
dstufftwhich bug is this00:49
clarkbdstufft: pip dependency resolution00:49
clarkb9XX /me finds it00:50
dstufftah right00:51
dstufftI want to get that into 1.5, it might not be though :[ Dpends on if the work gets done in time for the 1.5 timeframe00:51
mroddenwell that did it00:52
mroddenmade sure i had 2.14.1 and tests pass now...00:52
*** dina_belova has joined #openstack-infra00:54
fungimrodden: well, remember that installing a package with pip doesn't guarantee its dependencies get installed with pip. python has multiple package installation systems which comingle. i get the impression pip can't be absolutely certain what any particular package's dependencies might be, particularly prior to retrieving it00:54
mroddenonly 6 hours to find and fix that issue... but i got to learn about neutron, why sys.exit is bad, how stevedore works (kinda), and that pip doesn't really resolve dep conflicts for you00:55
fungiunlike, say, apt which has a database of all dependency relationships available to it in advance00:56
mroddeni think i just grabbed requests before novaclient came up in the list with the <1.2.3 requirement00:56
*** cppcabrera has left #openstack-infra00:56
mroddenportage/emerge would build its dep graph on the fly00:57
mroddeni think it still might00:57
mroddencould probably use a DB as a cache00:57
mroddenrequirements only change on newer versions00:58
fungithough it still probably has an enumerable means of discovering those dependencies00:58
mroddenoh wait... nvm... build time configuration flags change the graph00:58
mroddenapt-get doesn't have to deal with that00:58
*** gyee has quit IRC00:58
*** dina_belova has quit IRC00:58
fungiwhereas python packages *can* run arbitrary code at installation time to determine subsequent dependencies01:00
mroddenfungi: right01:00
mroddenportage does the same thing01:00
*** cody-somerville has quit IRC01:00
mroddensince at 'install' time in Gentoo you are actually building stuff, and depending upon waht build configuration flags you have, you might say need to download and compile boostlibs if you don't have them, vs maybe if you have the flag off it doesn't need that01:01
mroddenfyi avoid having to compile boostlibs on gentoo, it takes a few hours at least01:02
mroddenamazingly enough, its actually faster to do taht than compiling v8 and chromium01:02
mroddenthat takes forever01:02
mroddeni have a core i7 at home too01:03
lifelessdon't worry, they will rewrite it in go soon :P01:03
mroddengo was built for servers....01:03
mroddeni dont understand the fascination with it01:04
lifelessmrodden: I know, right :)01:04
mroddenwould have been great for the API levels in OpenStack though01:04
*** enikanorov-w_ has joined #openstack-infra01:10
*** enikanorov-w has quit IRC01:10
lifelessmebbe :)01:15
openstackgerritMathieu Gagné proposed a change to openstack-infra/config: Add stable/grizzly branch to gerribot for puppet
mroddenidk, its not like python is slow...01:15
mroddenand i think go uses "goroutines" which are essentially greenthreads, which is what we are doing now01:16
*** cody-somerville has joined #openstack-infra01:16
lifelessI don't know if they've changed this01:16
mroddencould be wrong about that though, its been awhile since i looked at it01:16
lifelessbut by default go uses one CPU w/greenthreads01:16
lifelessand for 2-cpu's it gets slower.01:16
*** UtahDave has joined #openstack-infra01:17
mroddenthats what i thought too01:18
mroddennot really all that great if thats still true01:19
*** mrodden has quit IRC01:23
dstufftfungi: generally pip downloads a package, runs a command to get it's dependencies, downloads those depenendencies, runs a command to get their dependenencies, until it builds a graph, and then it goes through and installs things. The two exceptions to that are setup_requires which will get installed locally (not globally) by setuptools during the first command, and people who do wierd things in their setup.py01:24
dstufftBut wierd things, things like automatically installing something when the is executed (vs when install is run), or overwriting the egg-info command with something else01:24
fungiright. bad ideas now that there are better alternatives, but these things still happen at least for now01:26
dstufftWheels don't have that problem though, given that they don't have a setup.py01:28
dstufftthey have some other problems that still need fixed :/01:28
dstufftAnd Sdist 2.0 whenever it gets done won't have that problem either01:28
* fungi can't wait!01:29
jheskethHowdy. Are there any Zuul devs in here?01:30
jeblairjhesketh: yes01:31
jheskethjeblair: Is it possible for a gearman worker to update the URL after it has accepted a job and given an initial one?01:31
jheskethsending another WorkData packet doesn't seem to have any effect01:31
jeblairjhesketh: that's correct, it only saves the url and build number from the first one01:33
*** Ryan_Lane has joined #openstack-infra01:33
jeblairjhesketh: i _think_ it would be okay to change that though01:34
jheskethjeblair: right, it kinda makes sense to be able to change it in my mind01:34
jheskethspecifically because once my worker is done with a log file I want to push it away (say onto a swift store or something) and then I'll need the URL to point there01:35
*** yaguang has joined #openstack-infra01:35
jeblairjhesketh: indeed; we want the same thing (and currently accomplish it by setting the log urls in configuration); but having the worker correctly report it is even better01:36
*** markmcclain has quit IRC01:37
jeblair(all our workers are jenkins atm, so that's a bit harder for us)01:37
*** nayward has joined #openstack-infra01:37
jeblairjhesketh: do you want to make that change?01:37
jheskethyep, I'll give it a go01:37
*** xchu has joined #openstack-infra01:37
jheskethbefore I do though, would you mind clarifying how the build numbers work please jeblair?01:37
jeblairjhesketh: cool; i'd probably just put a non-empty sanity check in there to make sure a real value doesn't get overwritten with a null or empty one later on.01:38
jeblairjhesketh: it can be any id that uniquely identifies that build of that job01:39
jeblairjhesketh: it comes from jenkins, where builds are identified as job-name + a monotonically increasing build number (trying to find a jenkins job by a zuul uuid is really difficult)01:39
jheskethhmm okay01:40
jeblairjhesketh: but zuul doesn't really care, so you you pass the zuul uuid back as the build number if you wanted01:40
jheskethso does zuul use it for anything?01:40
jeblairit uses it to abort jobs that are in progress01:40
jheskethyeah so I was confused by why zuul needed a number and didn't use it's own uuid01:40
jheskethor the job uuid01:40
*** mrodden has joined #openstack-infra01:41
*** nayward has quit IRC01:41
jeblairjhesketh: here is where it's used:
jeblairit calls "stop:[manager-name]" where manager-name is something else supplied in the workdata packet -- that's to route the stop job to the right gearman worker01:42
jeblairand passes the build name and number (supplied by the worker in the workdata packet) as json arguments01:42
jheskethyep cheers :-)01:43
jeblairjhesketh: good luck!01:43
*** emagana has quit IRC01:44
*** UtahDave has quit IRC01:44
*** sarob_ has joined #openstack-infra01:45
*** sarob has quit IRC01:48
*** sarob_ has quit IRC01:50
*** sdake_ has quit IRC01:53
*** dina_belova has joined #openstack-infra01:54
*** UtahDave has joined #openstack-infra01:56
lifelessruh roh - No distributions at all found for repoze.lru>=0.3 (from Routes>=1.12.3->-r /opt/stack/neutron/requirements.txt (01:57
lifelessis installing neutron known-broken ?01:57
lifelessor perhaps this is the more likely 'you used pip, from across the world'''01:58
*** thomasm has joined #openstack-infra01:59
*** dina_belova has quit IRC01:59
*** yaguang has quit IRC02:00
jheskethjeblair: it looks like the 'name' field in the data is unused?02:01
*** erfanian has joined #openstack-infra02:03
*** Ryan_Lane has quit IRC02:04
*** yaguang has joined #openstack-infra02:08
*** sarob has joined #openstack-infra02:12
*** mrodden1 has joined #openstack-infra02:14
*** mrodden has quit IRC02:15
*** HenryG has joined #openstack-infra02:20
*** UtahDave has quit IRC02:27
*** annegentle has quit IRC02:28
*** nati_ueno has quit IRC02:32
*** xchu has quit IRC02:33
*** melwitt has quit IRC02:33
*** jerryz has quit IRC02:34
*** jerryz has joined #openstack-infra02:44
*** UtahDave has joined #openstack-infra02:46
*** UtahDave has left #openstack-infra02:51
*** xchu has joined #openstack-infra02:51
*** markmcclain has joined #openstack-infra02:55
*** dina_belova has joined #openstack-infra02:55
*** jfriedly has joined #openstack-infra02:56
*** yaguang has quit IRC02:56
*** yaguang has joined #openstack-infra02:58
jfriedlyHi there guys.  I'm trying to build a debian package of python-cinderclient v1.0.5.  Has anyone done that already?02:58
*** dina_belova has quit IRC03:00
jfriedlyI found the 1.0.4 package on precise-proposed/havana, but I need to be able to update volume statuses with the cinderclient, and that's only in v1.0.503:00
*** nayward has joined #openstack-infra03:01
*** thomasem has joined #openstack-infra03:01
clarkbzigo or zul may have built a package03:01
jfriedlyclarkb:  Thanks, I'm looking around on Zuul now.  What/who is zigo?03:03
*** thomasm has quit IRC03:04
*** mordred has quit IRC03:04
clarkbzigo is doing debians openstack packaging03:05
clarkband zul not Zuul. There is a difference :)03:05
*** nayward has quit IRC03:05
jfriedlyOh, zul and zigo are both real people lol.  I'll see if I can find one of them online tomorrow maybe03:06
jfriedlyWould they be in this channel?03:06
clarkbI think zul is but not zigo03:07
*** mordred has joined #openstack-infra03:07
jfriedlyalright, thanks a bunch03:07
*** jrex_laptop has quit IRC03:13
yaguanghi all, can any help figure out why this happens [ERROR] /opt/stack/new/devstack/exercises/ server didn't become active within 180 seconds03:16
yaguangthe full log is here
clarkbyaguang: have you looked through any of the logs to see?03:22
yaguangclarkb, the nova-api log shows  a lot of fails of ec2 requests03:24
*** annegentle has joined #openstack-infra03:25
clarkband a little earlier
*** anteaya has quit IRC03:27
yaguangclarkb, this may be caused by my code change, thanks clarkb.03:28
openstackgerritJoshua Hesketh proposed a change to openstack-infra/zuul: Allow build info to be updated
jheskethjeblair: This allows the WorkData to be updated
clarkbyaguang: because the failure seems to be consistent I think that is the case03:28
clarkbyaguang: the tests have run 3 times on that patchset and neutron fails each time03:29
yaguangclarkb, I'll recheck my patch first03:30
clarkbyaguang: if you look at the second traceback I linked it starts with a line your patch has changed03:33
yaguangclarkb, I think I have find the case, thanks03:36
*** nayward has joined #openstack-infra03:37
*** vogxn has joined #openstack-infra03:37
*** nayward has quit IRC03:43
*** SergeyLukjanov has joined #openstack-infra03:46
*** reed has quit IRC03:55
*** dina_belova has joined #openstack-infra03:55
*** jerryz has quit IRC03:57
*** xchu has quit IRC03:58
*** dina_belova has quit IRC04:00
*** emagana has joined #openstack-infra04:09
*** SergeyLukjanov has quit IRC04:11
*** dina_belova has joined #openstack-infra04:14
*** Ryan_Lane has joined #openstack-infra04:23
clarkbyaguang: looks like it is passing now. Did it just need a rebase? I don't see anything else that changed04:28
openstackgerritA change was merged to openstack-infra/jenkins-job-builder: Added Robot Framework reports publisher
*** dina_belova has quit IRC04:29
*** jfriedly has quit IRC04:29
*** erfanian has quit IRC04:33
*** sarob has quit IRC04:35
*** sarob has joined #openstack-infra04:36
*** emagana has quit IRC04:40
*** sarob has quit IRC04:40
*** afazekas_ has joined #openstack-infra04:47
*** afazekas_ has quit IRC04:48
*** sarob has joined #openstack-infra04:48
*** afazekas_ has joined #openstack-infra04:48
*** sarob has quit IRC04:49
*** sarob has joined #openstack-infra04:49
*** sarob has quit IRC04:53
*** yamahata has quit IRC04:57
*** yamahata has joined #openstack-infra04:57
openstackgerritJoshua Hesketh proposed a change to openstack-infra/zuul: Allow build info to be updated
*** boris-42 has joined #openstack-infra05:01
*** emagana has joined #openstack-infra05:04
*** xchu has joined #openstack-infra05:12
*** markmcclain has quit IRC05:13
*** Ryan_Lane has quit IRC05:16
yaguangclarkb, I changed my code , so it passes now05:16
*** yaguang has quit IRC05:17
*** dripton has quit IRC05:21
*** dripton has joined #openstack-infra05:25
*** marun has quit IRC05:26
*** marun has joined #openstack-infra05:27
*** nicedice_ has quit IRC05:28
*** emagana has quit IRC05:29
*** yaguang has joined #openstack-infra05:41
openstackgerritJoshua Hesketh proposed a change to openstack-infra/zuul: Allow build info to be updated
*** yaguang has quit IRC05:54
*** emagana has joined #openstack-infra06:00
*** yamahata has quit IRC06:00
*** yamahata has joined #openstack-infra06:01
jheskethAlex_Gaynor: thanks for your review on 4102006:03
jheskethwith regard to the private attribute, wouldn't it map to the same as line 382 in the original patch? (
Alex_Gaynorhmm, if they're all in the same class, it actually will work, but that sounds like a really messy approach06:05
clarkbAlex_Gaynor: that is unrelated to this change. If something needs to be addressed there I would do it in a different change06:05
jheskethyeah, I agree06:05
clarkbjhesketh: I have another comment too06:05
Alex_GaynorOk, sounds reasonable to me then.06:05
clarkbwith what I think is an even better alternative to Alex_Gaynor's suggestion >_>06:06
jheskethI think that it was made private because it should never be changed.. however I can't see a reason why a job couldn't allocate a different manager06:06
jheskethclarkb: I actually had an initial implementation like this but using setattr06:08
jheskeththat didn't work so I reverted to closer to the original file06:08
jheskethI don't suppose you would have any insight into why setattr wouldn't work rather than
*** emagana has quit IRC06:09
clarkbnot without more context, was the one of the earlier patchsets in gerrit?06:09
jheskethclarkb: patchset 2 does it (although before pushing that I had the one loop for both the initial assignment and following ones)06:10
jhesketh(but that's irrelevant.. I'm not sure why patchset 2 didn't actually work)06:10
clarkbjhesketh: I do not know why that would not work06:14
*** yaguang has joined #openstack-infra06:14
clarkbjhesketh: actually Alex_Gaynor's link may explain it06:18
clarkbat least for __gearman_manager06:18
jheskethclarkb: actually I think you're right.. it did seem most paranoid on that attribute06:23
jheskethokay, I'll fix something up06:23
*** nayward has joined #openstack-infra06:43
*** dnavale has quit IRC06:47
*** sarob has joined #openstack-infra06:49
*** yolanda has joined #openstack-infra06:57
openstackgerritJoshua Hesketh proposed a change to openstack-infra/zuul: Allow build info to be updated
*** sarob has quit IRC07:01
*** sarob has joined #openstack-infra07:02
*** sarob has quit IRC07:06
*** dina_belova has joined #openstack-infra07:15
*** yaguang has quit IRC07:15
ttxfungi: I'm not pulling logs. I'm looking into jobs, their weather report and the runs in progress07:15
*** dina_belova has quit IRC07:16
ttxfungi/jeblair: used to be a straightforward URL and APi call... Now it looks like a hide and seek game07:16
ttxbut I'm probably not using that setup correctly07:16
*** yaguang has joined #openstack-infra07:21
*** vogxn has quit IRC07:23
*** Ryan_Lane has joined #openstack-infra07:26
*** Ryan_Lane has quit IRC07:34
*** fbo_away is now known as fbo07:38
*** afazekas has quit IRC07:44
*** afazekas_ is now known as afazekas07:44
*** jinkoo has joined #openstack-infra07:48
*** Ryan_Lane has joined #openstack-infra07:49
*** sarob has joined #openstack-infra07:49
*** SergeyLukjanov has joined #openstack-infra07:49
*** Ryan_Lane has quit IRC07:52
*** sarob has quit IRC07:57
*** thomasbiege has joined #openstack-infra07:58
*** Ryan_Lane has joined #openstack-infra08:01
*** shardy has joined #openstack-infra08:01
*** dina_belova has joined #openstack-infra08:08
*** CaptTofu has quit IRC08:13
*** markmc has joined #openstack-infra08:18
*** Ryan_Lane has quit IRC08:19
*** jpich has joined #openstack-infra08:20
*** vogxn has joined #openstack-infra08:20
*** sarob has joined #openstack-infra08:23
*** sarob has quit IRC08:28
*** Ryan_Lane has joined #openstack-infra08:47
*** nayward has quit IRC08:49
*** sarob has joined #openstack-infra08:54
*** dina_belova has quit IRC08:55
*** Ryan_Lane has quit IRC08:57
*** enikanorov has quit IRC08:58
*** sarob has quit IRC08:59
*** enikanorov has joined #openstack-infra08:59
*** jinkoo has quit IRC09:09
*** dina_belova has joined #openstack-infra09:14
*** dina_belova has quit IRC09:14
*** sdake_ has joined #openstack-infra09:18
*** dina_belova has joined #openstack-infra09:20
*** dina_belova has quit IRC09:36
*** jhesketh has quit IRC09:40
*** mrmartin has joined #openstack-infra09:40
*** psedlak has joined #openstack-infra09:45
*** sarob has joined #openstack-infra09:55
*** CliMz has joined #openstack-infra09:58
*** sarob has quit IRC09:59
*** odyssey4me has joined #openstack-infra10:02
*** dina_belova has joined #openstack-infra10:13
*** dina_belova has quit IRC10:15
*** dina_belova has joined #openstack-infra10:15
*** jjmb has quit IRC10:18
*** davidlenwell has quit IRC10:19
*** davidlenwell has joined #openstack-infra10:22
*** xchu has quit IRC10:26
*** dina_belova has quit IRC10:27
*** dina_belova has joined #openstack-infra10:41
*** dina_belova has quit IRC10:41
openstackgerritA change was merged to openstack-infra/reviewday: Rename quantum to neutron
*** vogxn has quit IRC10:45
*** yaguang has quit IRC10:48
*** sarob has joined #openstack-infra10:55
*** rscottcoyle has quit IRC10:56
*** mirrorbox has quit IRC10:56
*** CaptTofu has joined #openstack-infra10:57
*** sarob has quit IRC11:00
*** dina_belova has joined #openstack-infra11:06
*** rscottcoyle has joined #openstack-infra11:07
*** BobBall_AFK is now known as BobBall11:08
openstackgerritSean Dague proposed a change to openstack/requirements: use dev-requirements for global sync
openstackgerritSean Dague proposed a change to openstack/requirements: add option for adding a suffix to output files
*** vogxn has joined #openstack-infra11:15
*** vogxn has quit IRC11:17
*** odyssey4me has quit IRC11:24
*** fifieldt has quit IRC11:27
*** woodspa has joined #openstack-infra11:29
*** odyssey4me has joined #openstack-infra11:30
*** dina_belova has quit IRC11:33
*** dina_belova has joined #openstack-infra11:42
*** weshay has joined #openstack-infra11:44
*** ruhe has joined #openstack-infra11:45
*** ArxCruz has joined #openstack-infra11:54
*** sarob has joined #openstack-infra11:56
*** dprince has joined #openstack-infra11:56
*** dina_belova has quit IRC11:57
openstackgerritSean Dague proposed a change to openstack-infra/config: float the progress bars and statuses to the right
openstackgerritSean Dague proposed a change to openstack-infra/config: float the progress bars and statuses to the right
sdaguejeblair: ok, when you get a chance, I think that change ^^^^^^^^^^ should be compatible with both of our sensibilities. :)11:59
*** sarob has quit IRC12:00
*** ruhe has quit IRC12:01
*** dkranz has joined #openstack-infra12:03
*** SergeyLukjanov has quit IRC12:08
*** jjmb has joined #openstack-infra12:09
*** mjfork has joined #openstack-infra12:13
*** dina_belova has joined #openstack-infra12:28
*** vogxn has joined #openstack-infra12:29
*** ruhe has joined #openstack-infra12:31
*** beagles_biab is now known as beagles12:33
*** mrmartin has quit IRC12:35
*** dina_belova has quit IRC12:36
*** xchu has joined #openstack-infra12:40
*** thomasbiege has quit IRC12:41
*** adalbas has joined #openstack-infra12:45
*** dprince has quit IRC12:54
openstackgerritA change was merged to openstack-infra/storyboard: Introducing project groups
*** derekh has joined #openstack-infra12:56
*** sarob has joined #openstack-infra12:57
*** ruhe has quit IRC12:58
*** sarob has quit IRC13:01
*** dprince has joined #openstack-infra13:05
*** vijendar has joined #openstack-infra13:06
lifelesshuh, why isn't distribute mirrored by run-mirror ?13:08
mordredlifeless: good question.13:09
lifelessor another way, whats the best way for me to add a single package to a run-mirror mirrored mirror ?13:09
*** markmc has quit IRC13:09
mordredlifeless: well, I do that in the integration script - by injecting a package in between the two runs13:10
*** dprince has quit IRC13:10
lifelessbetween the no-process and no-downloads ?13:12
lifelessmordred: btw is process needed for file:/// usage?13:12
mordredyeah - process is what takes the download dir and makes it into a mirror structure13:12
mordredso - download just produces a giant directory of files that looks sort of a like a PIP_DOWNLOAD_CACHE13:13
*** russellb is now known as rustlebee13:14
mordredhowever, hrm - if one were to just simply download into the structure in the first place, it would be easier to incrementally run and add new packages13:14
mordreddirectly into the mirror13:14
*** mriedem has joined #openstack-infra13:14
mordredwe _used_ to make html indexes, which is the reason that the code is structured that way - we didn't have context to make the index until the end13:14
*** zul has joined #openstack-infra13:15
lifelessmordred: that is why I'm asking, yes.13:16
lifelessmordred: also - glance breaks pbr test runs atm, setuptools-git something or other.13:16
mordredsigh. that's the 'we don't grab transient easy_install depends' problem13:17
mordredI can patch around that - I'd really love it if people would finish merging the pbr update patches though13:18
openstackgerritDevananda van der Veen proposed a change to openstack/requirements: Add pyghmi
lifelessmordred: well, we can't land any pbr patches atm :)13:19
devanandamordred: g'morning!13:19
lifelessmordred: gnight!13:19
*** sandywalsh has quit IRC13:19
*** thomasbiege has joined #openstack-infra13:19
mordredmorning devananda !13:19
mordredgnight lifeless13:19
HenryGAre these "Add dependency" instructions for git review still correct?13:19
devanandamordred: have a few to chat?13:20
HenryGWhen I get to "git review -R" I get a very scary long list of outstanding commits.13:20
devanandaHenryG: yes, that wiki is correct. I think that suggests you diverged from upstream a while back13:21
HenryGdevananda: thanks. Could it be the review I am depending on is diverged?13:22
devanandaHenryG: you'd see that in the dependencies of that review, if it were13:23
*** changbl has quit IRC13:25
devanandaHenryG: check the SHA of the commits in your local tree vs. the review you think you are depending on13:25
devanandaHenryG: if they're not the same SHA, then when you git review -R, you'll overwrite // update the review "ahead" of yours, which, it sounds like, is not what you want to do13:25
*** sarob has joined #openstack-infra13:27
HenryGdevananda: yeah, how did that happen? I guess I start from scratch13:27
*** mikal has quit IRC13:28
*** jpeeler has joined #openstack-infra13:28
*** mikal has joined #openstack-infra13:29
devanandaHenryG: leave thta branch aside and you can cherry-pick from it13:30
HenryGdevananda: I think I am confused about how to start13:31
*** dansmith is now known as Steely_Dan13:32
*** sarob has quit IRC13:32
*** sandywalsh has joined #openstack-infra13:32
HenryGDo I start from master and fetch the refs I depend on?13:32
mordredzomg. my longest patch is going in today:13:34
mordredUploadedApr 21, 2013 3:00 AM13:34
mordredUpdatedAug 9, 2013 10:34 AM13:34
mordredHenryG: which refs do you depend on?13:35
mordreddevananda: ola13:35
*** mjblack has joined #openstack-infra13:38
*** prad has joined #openstack-infra13:39
mordredttx, notmyname: ping13:42
mordredHenryG: and you have your patch on top of that patch?13:43
notmynamemordred: ack13:43
*** CliMz has quit IRC13:43
mordrednotmyname: a) what are you doing awake?13:43
mordrednotmyname: b) I wanted to talk to you and ttx about merging milestone-proposed branches back into master post-release13:44
mordredI think we should do it across the board in the general case, but swift might be a gentler project to explore how it might work13:44
notmynameI've started getting up at 6am to do work, especially openstack stuff like reviews13:44
HenryGmordred: That is what I want, but I have started over. Deleted the branches and checked out master and pulled latest.13:45
mordredbecause the chances of master diverging in a huge way during the m-p period is, I think, smaller13:45
* notmyname hasn't done any reviews yet this morning13:45
mordrednotmyname: that's smart13:45
notmynamemordred: thinking a few steps ahead, this should solve the pbr version issue, right?13:45
mordredHenryG: ok. if you do that - just git review -d 40636, make your patch, and then git review -R and it should only show you two patches13:46
mordrednotmyname: yes. it will definitely solve the pbr issue13:46
mordredbut also I kinda think having the tags exist in the master history would be nice13:46
HenryGmordred: do I need to specify --amend for commit?13:47
Alex_Gaynormordred: Are the docs at still up to date?13:47
notmynamethe only changes we normally make to milestone-propose that aren't in master are the final versioning marker13:48
mordredHenryG: no. you are not ammending the commit13:48
*** dprince has joined #openstack-infra13:48
mordrednotmyname: right - and if you move to pbr, then you might not even need a m-p branch in most cases13:48
*** Guest65861 has quit IRC13:48
notmynameso I can't imagine that this would actually cause any impact at all once that's resolved. especially since patches on m-p should be on master anyway13:49
*** jpeeler has quit IRC13:49
notmynamemordred: so how are you imagining this works? time for a release, and we make an -rc tag? then add the final tag when the release is finalized?13:50
fungittx: mmm, probably the safest way to get the logs you want is to construct urls to for the tags in question13:51
fungittx: let me see if i can come up with a good mapping you can construct easily13:51
mordrednotmyname: yes - although now that you say that I'd like to verify that if a single commit has two tags (like if you tag it rc1 and then actually have no more changes) that we produce the best version13:51
fungittx: specifically you want to be able to find the jenkins console log containing the checksum of a tarball generated froma particular tag on a particular project, right?13:52
mjblackso I need to do the walk of shame....I accidentally did the USG/System CLA's and need to have them removed13:52
mordredI think we use the latest tag - but I'd like to make sure13:52
mordredfungi: logstash?13:52
fungimjblack: gimme a couple minutes and i'll administer your spanking^W^W^Wclear that from the database13:52
mjblacklol thank you13:53
HenryGmordred: thanks. Should I change the --topic ?13:53
mordredHenryG: probably, yes13:53
*** burt has joined #openstack-infra13:53
fungimordred: maybe, but i'd rather we went with a lookup solution which didn't magically expire in two weeks13:53
HenryGmordred: So I get that huge long list of outstanding commits again.13:54
*** anteaya has joined #openstack-infra13:54
mordrednotmyname: yes. that works13:55
fungiHenryG: are you maybe submitting to a different remote branch?13:55
fungiHenryG: what is the change number?13:56
HenryGfungi: I don't have a change number yet. Have not dared to submit the review.13:57
fungiHenryG: no, i mean what's the number of the change you're trying to depend this one on top of?13:57
notmynamemordred: so if m-p goes away (ie replaced with a tag) via pbr, then the idea of merging m-p doesn't make too much sense (ie it doesn't apply). so the question seems to only matter before pbr13:58
*** sarob has joined #openstack-infra13:58
HenryGfungi: 47c0d2ebd55d4c5de2c3e914a7c3f91a8d31a20d13:58
fungiHenryG: for example, if you git review -d some change number which is a review for stable/grizzly and then commit on top of that and try to git review to master, this can happen13:58
mordrednotmyname: yes. which means specifically to enable you to consider the pbr change13:58
HenryGfungi: aha, I did not check that13:58
fungimmm, though 47c0d2e says neutron master13:59
mordrednotmyname: we might want to merge your last m-p branch in, which should get the last tag into your master13:59
mordredfungi: the gerrit remote could be out of date13:59
mordredHenryG: can you run "git remote update"13:59
fungimordred: also entirely possible. maybe this is fallout from the repo rename a couple months ago14:00
HenryGfungi: mordred: it fetch some new stuff14:00
mordredHenryG: now your git review -R should not be a scary list14:01
HenryGmordred: yes!14:01
fungiHenryG: also note that suggests running 'git remote update' before you do anything else14:01
fungigood to keep in your back pocket14:02
* HenryG hangs head in shame14:02
*** markmcclain has joined #openstack-infra14:02
fungido not be ashamed. rejoice in your newly acquired knowledge!14:02
*** sarob has quit IRC14:02
HenryGfungi: it is knowledge I learned a few months ago and forgot :(14:03
mordrednotmyname: I don't see a 1.9 tag in your repo at all14:03
mordredoh - weird. I see it on github, but not locally14:04
fungiHenryG: your memory can't be any worse than mine, so don't worry14:04
mordrednotmyname: nm. got it14:04
* fungi has a mind like a steel sieve14:04
HenryGSo just to be triple-sure: git review -R says I have two outstanding commits. Mine and the dependent one. Is that right?14:04
*** jpeeler has joined #openstack-infra14:05
*** beagles is now known as seagulls14:05
fungiHenryG: yep, and as long as the commit sha of that parent commit matches the one in gerrit, gerrit will ignore it and set yours as a child of it14:05
HenryGfungi: thanks! ok here goes...14:05
fungiHenryG: git review just has no way of knowing that will happen, so it warns you about it14:05
*** pentameter has joined #openstack-infra14:06
fungiHenryG: the next release of git-review (maybe today if i get a moment?) will have a more verbose message explaining what that means14:06
*** krtaylor has joined #openstack-infra14:06
notmynamettx: shouldn't we have a 1.9.1 rc tag?14:07
HenryGfungi: cool! And my review looks good. Thanks!14:07
*** rfolco has joined #openstack-infra14:08
sdagueclarkb: once you wake up, I want to figure out what about the log formatter was choking for you14:10
fungimjblack: i just cleared all the mistaken system cla and usg cla signature attempts from gerrit. don't feel bad--there were at least four other people who seem to have made the same mistake in the few weeks since i last cleared it14:10
mjblackthank you!14:11
fungimjblack: you're welcome. happy hacking!14:11
sdagueapparently... I wasn't on the openstack-infra list14:11
*** alaski is now known as lascii14:11
*** afazekas has quit IRC14:11
mordrednotmyname: yea, I don't see one of those14:13
*** mikal has quit IRC14:14
mordrednotmyname: there's what such a thing looks like14:14
*** mikal has joined #openstack-infra14:15
fungisdague: did you get a held-for-moderation notice?14:16
Alex_Gaynornotmyname: I do seem to be, don't I?14:16
notmynameAlex_Gaynor: heh :-)14:16
fungisdague: if so, i can go poking in the mm interface and approve14:16
*** ruhe has joined #openstack-infra14:16
sdaguefungi: no, dkranz asked me to weigh in on something, and I realized I'd never seen the email :)14:16
mordrednotmyname: and it looks like swift-milestone already has permission to propose merge commits14:17
fungisdague: ahh, archives ftw14:17
mriedemmordred: ping14:18
notmynamemordred: since we cut an RC yesterday (planned for final early next week) and have the current dev target of 1.9.2 in the source, it sounds like we need a 1.9.2 tag on master before we can merge pbr14:19
notmynamemordred: I anticipate that our next release will be 1.10.0 (instead of 1.9.2), so perhaps tagging master as 1.9.2 right before pbr merges would be appropriate?14:19
mordrednotmyname: do you? do you not just need a 1.9.1 tag?14:19
*** avtar has joined #openstack-infra14:19
mriedemmordred: nevermind my ping, figured out my questoin14:20
notmynamemordred: we need the 1.9.1 tag anyway. but when pbr is merged, it must produce a version string that sorts after the one from the previous commit, right?14:22
notmynamemordred: eg master's version is at the top of
mordrednotmyname: oh - I get it - your current tarballs are showing 1.9.2-dev because of the current scheme14:23
notmynamemordred: correct14:23
notmynameI dont' particularly like the current scheme, but it's what we have14:23
mordredyeah - I think the pbr tag stuff will be more like what you want14:24
sdaguemordred: this will make you happy, not parallel yet -
sdaguebut at least common runner14:24
mordrednotmyname: what if we made a tag in master on the commit where you changed the version to 1.9.2-dev14:24
mordrednotmyname: called "1.9.2-dev"14:24
notmynamemordred: yes it is. I really like it. the challenge is if the extra dependency is worth it for packagers. ie we're replacing a string in a file with a bit of complexity14:25
mordredthe tarball that gets cut shouldn't surprise anyone14:25
*** cppcabrera has joined #openstack-infra14:25
mordrednotmyname: so, for the _packagers_ we've actually got some nice features to make their lives easier14:25
mordrednotmyname: if they consume the published tarball, the version is encoded in the tarball, so they do not need access to vcs14:26
notmynamemordred: cool. (by packagers, I normally mean people at rackspace and swiftstack who gripe at me when things like this change)14:26
mordrednotmyname: also, they can add SKIP_PIP_INSTALL=true to their env or their package build scripts, and all of the python deps will be ignored14:26
notmynamethat's good to know. are these things documented somewhere?14:27
*** cppcabrera has left #openstack-infra14:27
mordrednotmyname: it is - but there's a bug in the job that uploads the docs that I need to fix14:27
*** changbl has joined #openstack-infra14:27
*** thomasbiege has quit IRC14:28
notmynamemordred: ok. but the docs come from the source?14:28
mordredyes - although it seems like SKIP_PIP_INSTALL may not have made it into the docs...14:28
* mordred goes to do a doc cleanup patch14:28
mordredACTUALLY - I will go fix the doc job, _then_ do a doc cleanup patch14:29
notmynamemordred: so don't we need the 1.9.2-dev patch for pbr (independent of when a 1.9.1 patch happens)?14:30
*** ruhe has quit IRC14:30
mordrednotmyname: yes. I think that doing the 1.9.2-dev tag will be the cleanest way to move forward14:30
notmynameI leave a note on the gerrit review14:31
openstackgerritMonty Taylor proposed a change to openstack-dev/pbr: Update requirements in integration test
*** adalbas has quit IRC14:33
notmynamemordred: so we don't need the 1.9.0 tag merge patch, correct. or rather, it neither helps nor really hurts us14:35
mordrednotmyname: right. neither helps nor hurts14:35
*** datsun180b has joined #openstack-infra14:35
*** datsun180b has quit IRC14:35
notmynamemordred: kk. then I'd prefer we not deal with it. I can -2 or you can abandon it14:36
*** afazekas has joined #openstack-infra14:36
mordredI'll abandon14:36
*** changbl has quit IRC14:37
*** ruhe has joined #openstack-infra14:37
*** seagulls is now known as seagulls_at_dump14:39
*** xchu has quit IRC14:45
*** yolanda has quit IRC14:46
*** adalbas has joined #openstack-infra14:48
openstackgerritMonty Taylor proposed a change to openstack-dev/pbr: Added documentation for packagers
*** mrodden1 has quit IRC14:50
mordrednotmyname: ^^14:51
mordrednotmyname: thanks - you always ask useful questions14:51
notmynameglad to help :-)14:51
sdaguemordred: when you get a chance14:53
sdagueand the one on top of it is rechecking because of neutron race bug14:53
sdaguebut for your consideration as well14:53
mordredsdague: looks good. I tihnk we shoudl land it14:56
*** afazekas has quit IRC14:56
sdaguego for it14:56
sdagueonce the one on top lands, I'll go fix devstack to use it14:57
sdaguedtroyer: you up yet? I'm getting an odd permissions issue with fresh devstack with the new pip patch14:57
ttxnotmyname: we do tag RCs on the synchronized release... but not so much on intermediary releases. That's because the process is slightly different14:58
notmynamettx: ah ok14:58
ttxnotmyname: if you look at 1.9.0 there wasn't an RC tag either14:58
ttxthis is more like a frozen branch than a "real" RC14:58
*** sarob has joined #openstack-infra14:58
*** BobBall has quit IRC14:58
*** changbl has joined #openstack-infra14:58
*** BobBall has joined #openstack-infra14:59
openstackgerritMonty Taylor proposed a change to openstack-infra/config: Fix pbr doc jobs
ttxit's more lightweight, no need to open new milestones and stuff15:00
mordredfungi: ^^15:00
fungimordred: i'll take a gander15:00
ttxfungi: I don't really want the logs. I want to access the job parameters through the Jenkins API15:00
fungisorry should have said "i reckon i'll take a gander"15:00
mordredttx: any thoughts on the earlier discussion of using swift as a test case for merging m-p branches back to master to make tag history happy?15:01
ttxfungi: see
fungittx: ahh, okay... i'll see if i can come up with a way to route that15:01
*** mrodden has joined #openstack-infra15:01
fungittx: since you need to know which jenkins master handled that particular run of the job15:02
ttxfungi: I also look at the job itself to see the weather report on it... it's more difficult to get a feel of the health of a job now ?15:02
*** burt has quit IRC15:02
ttxfungi: or is there some consolidated dashboard somewhere ?15:02
fungittx: and also, bonus points if we can abstract that away from jenkins since tarball building is probably one of the first things we'll want to unjenkinsify15:02
ttxmordred: looking15:03
*** sarob has quit IRC15:03
fungittx: well, we have per-job success/failure counts in graphite now, so we can probably get you a global view of that without jenkins15:03
mordredttx: tl;dr - pbr versions get weird if the tag is not on the master branch, but sometimes the swift tags are on m-p15:03
mordredmornin jeblair15:04
openstackgerritElizabeth Krumbach Joseph proposed a change to openstack-infra/config: Enable ssl module in httpd on
*** vogxn has quit IRC15:04
pleia2fungi: ^^ that should do it, our apache module seems to have support for ubuntu/debian a2enmod but not identical for redhat/centos15:04
sdaguejeblair: ok, going to jump on you to take a look here - - I think I addressed your concerns, and I like the ui way better :)15:04
*** bnemec is now known as oldben15:05
pleia2and I think the problem we had was the module anyway, it wants to control everything in /etc/httd so it deletes ssl.conf if it was created by installing mod_ssl (which is installed in hpcloud centos, not sure about rackspace)15:05
jeblairsdague: okay, i'll take a look locally15:05
pleia2anyway, made sure mod_ssl is installed too, just to be sure :)15:05
mordredfungi, pleia2: do we still need to hup gerrit to get the replication change picked up?15:06
jeblairfungi, ttx: yeah, i agree the better place to look for overall job health would be graphite at this point.  basically like that neutron graph.15:06
pleia2mordred: yes15:07
fungimordred: not sure if a hup will suffice. does it maybe only reread those configs on restart?15:07
mordredfungi: sorry, that's what I meant15:07
ttxmordred, notmyname: a few questions15:07
jeblairfungi, ttx: and i think the wait for tarball script may need to read zuul's status.json to find out who's running the job (better than hardcoding jenkins/01/02 in there, which may change)15:08
ttxmordred, notmyname: explain "if m-p goes away (ie replaced with a tag) via pbr"15:08
ttxhow would you backport bugfixes to a tag ??15:08
pleia2mordred: should be fun times! signed up15:08
fungimordred: but yes, we haven't kicked gerrit to force it to notice that configuration change. i can't remember if we decided automatic restarting of gerrit was a bad idea15:08
jeblairfungi, ttx: i've been thinking about adding a generalized zeromq server to zuul; maybe this would be a good case for it.15:09
mordredttx: I think that was intended to mean "if the commit in m-p which changes the version" goes away15:09
ttxmordred: ok, like what we do elsewhere, right ?15:09
fungijeblair: yes, i'd love to see this particular dependence on the jenkins api disappear15:09
ttxmordred: other question: why would you need a 1.9.2-dev tag ?15:09
pleia2I'm heading out to meet up with some fosscon folks for lunch (in town and all, said I'd help them with some things), laptop is coming along so I'll check in in a bit15:09
mordredttx: in this case purely to make the transition work15:09
ttxmordred: ok15:10
mordredttx: we would not need similar ones in the future15:10
*** _TheDodd_ has joined #openstack-infra15:10
ttxso this is about (1) aligning swift to how we do version/tagging on other projects and (2) try merging m-p tags back to master after release15:10
mordredttx: but they've been installing things labeled 1.9.2-dev already - so if we moved to post-tag versioning on top of 1.9.1 it would be a very weird transition15:10
ttxmordred: ^ ?15:10
*** mjblack has left #openstack-infra15:11
mordredttx: (1) halfway - I dont think we're looking at pre-release versioning for them - but otherwise yes15:11
*** mjblack has joined #openstack-infra15:11
*** markmcclain has quit IRC15:11
ttxmordred: aren't they doing pre-release versioning currently ?15:11
*** mjblack has left #openstack-infra15:11
mordredttx: (2) yes - and swift is a good candidate to try, I think, because the amount of change typically found on their master during an m-p period should be low15:11
mordredttx: no. the version in their tree does not necessarily correspond to the next version15:12
mordredttx: for instance, right now their tree says 1.9.2 but the next release notmyname thinks will be 1.1015:12
openstackgerritwill soula proposed a change to openstack-infra/jenkins-job-builder: Adding support for the Warnings plugin
mordredso post-tag versioning will actually get him closer to what he's trying to do, which is 'pick the version at release time'15:13
ttxmordred: actually, if notmyname was 100% sure the next one is 1.10, the tree would say 1.10.15:13
ttxThe problem is, he doesn't really know right now15:13
mordredttx: I don't know that he's 100% sure15:13
mordredit _could_ also be 1.9.215:13
ttxso 1.9.2 is a... secure guess15:14
ttxmordred: so how would that work ?15:14
mordredttx: same as how the client libs work15:15
* ttx thinks about that15:15
mordredttx: in the general case, when he releases and tags 1.10, the next commit will produce 1.10.a1.g23424515:15
mordredhe could also do something similar to now by adding the versoin field to setup.cfg and doing a version bump in m-p15:16
ttxyou mean 1.10.1.g23234515:16
*** _TheDodd_ has quit IRC15:17
ttx(he should tag 1.10.0 actually to get
ttx(aX is only used in pre-versioning iiuc)15:18
ttxmordred: ^15:18
mordredoh. yes. you're right15:19
ttxok, just doublechecking :)15:19
ttxmordred: yes, this all sounds sane and we can experiment with merging back on 1.9.115:20
mordredttx: also - someone, possibly dstufft or Alex_Gaynor pointed out that there is a specific suffix we should be using instead of a for our things15:20
mordredttx: but I think we can wait to fix that until icehouse15:20
mordredttx: awesome15:20
annegentlemordred: Q for ya on your boot camp dinner logistics when you get a mo15:20
mordredannegentle: shoot!15:20
annegentlemordred: was your restaurant within walking distance? Since it was NY I'm guessing so.15:21
annegentlemordred: trying to figure out if I need people movers in Mountain View for dinner15:21
mordredttx: ... apparently we should probaby be using devX instead of aX for our pre-versioning15:22
mordredannegentle: yeah - it was within walking distance15:22
annegentlemordred: okay15:22
mordredannegentle: one of the few parts about having it in manhattan that was less work :)15:22
ttxthe world could benefit from standardized version comparison15:22
zuloh goody another version change15:23
openstackgerritwill soula proposed a change to openstack-infra/jenkins-job-builder: Adding support for the Warnings plugin
mordredzul: you probably won't notice this one15:23
*** rcleere has joined #openstack-infra15:23
mordredzul: or who knows, maybe you will15:23
ttxmordred: I'm unclear on the benefits of merging tags back though. github's new tag list doesn't sound confused15:23
mordredzul: we don't do it right now15:23
zulmordred:  i bet i will ;)15:23
ttxexample at, search for 2013.2.b215:24
* ttx reads the bug again15:24
mordredbut in the master tree, it does not show up in ones history15:24
mordredI don't care about the github ui15:24
mordredI care about being able to calculate the most recent tag in the history that came before the current commit15:25
ttxmordred: trying to understand the use case15:25
*** sandywalsh has quit IRC15:25
mordredttx: pbr versionin15:25
clarkbsdague re log formatter request a file that should 404 and you get a 50015:25
mordredttx: for swift15:25
mordredfor one15:25
notmynamemordred: ttx: sorry, my wife took cinnamon rolls out of the oven, and I had to do something about it. jsut caught up now15:25
ttxmordred: "most recent tag in the history that came before the current commit" -- you mean git tag --contains doesn't work unless we merge back to master ?15:25
mordredif we don't merge them back, their current revision would be something like
clarkbsdage eg logs/syslog.txt in a unitest dir15:25
mordredttx: yes15:26
mordredthat is correct15:26
ttxmordred: that sounds like a useful enough use case15:26
ttxi'm easily convinced. Especially with the smell of cinnamon rolls15:26
ttxnotmyname: tl;dr: lgtm15:26
ttxnotmyname: was just worried by the  "if m-p goes away (ie replaced with a tag) via pbr" statement15:27
notmynamettx: ya, so if we do have a backport, we'd end up branching from that tag and doing the work. and tagging that branch (right mordred?)15:27
ttxwe'd still do m-p, even with version-using-tag15:28
ttxi.e. we would cut a release branch in all cases, tag on that... then merge it back to master15:28
ttxthat way it's still the same process used everywhere15:29
ttxadditional benefit is that we get a tarball that is "the latest proposed release candidate"15:29
*** changbl has quit IRC15:29
notmynamettx: ok. but in reality, that branch won't have any changes on it if nothing is found during QA right?15:30
*** vogxn has joined #openstack-infra15:30
*** vijendar has quit IRC15:31
ttxnotmyname: you don't really NEED it given how many backports are ususally done on swift release branches... but that way it's aligned on more busy-just-before-release projects15:31
ttxnotmyname: correct15:31
*** nayward has joined #openstack-infra15:31
ttxnotmyname: I don't think you need a tag at the release branch point either (same reason that you don't have a 1.9.1-rc tag right now)15:32
ttxWhen we do formal RCs (a milestone in LP, a -rc1 tag in git) then we do use tag on RCs... but those are hopefully long-lived RCs that we coordinate at common release time15:33
notmynamettx: so `git co master && git co -b milestone-proposed && && git tag 1.9.2`15:33
ttx&& merge-back-to-master-magic-command15:33
ttx&& remove-mp-branch15:34
ttx(or rename it stable/*15:34
ttxnotmyname: I shall run most stages for you anyway, I have scripts for almost everything15:35
ttxmordred, notmyname: 1.9.1 release is planned for Tuesday so that means having the mechanics for merging back ready by Tuesday15:36
*** ^d has joined #openstack-infra15:37
*** rnirmal has joined #openstack-infra15:38
*** zul has quit IRC15:39
notmynamettx: FWIW tagging the same commit as the branch (ie no commits on the branch) shows it on master too15:39
sdagueclarkb: do you have an example?15:39
sdague 404s correctly, for instance15:39
notmynamettx: so yes, we're in agreement: looks good and should work with no anticipated issues15:40
notmynamettx: mordred: for 1.9.1, we don't need the merge semantics15:41
notmynamettx: mordred: or at least it's independent of that release15:41
*** sandywalsh has joined #openstack-infra15:41
ttxnotmyname: ok15:41
mordrednotmyname, ttx: I concur15:42
notmynamettx: what we need is a tag on master that reflects the current version in master before the pbr branch merges15:42
ttxnotmyname: right, for the transition.15:42
notmynameand to avoid confusion with the current RC and release, should we do this next week?15:42
ttxnotmyname: +115:43
notmynamemordred: ok with you?15:43
ttxmordred, notmyname: so I'm traveling on Monday and starting Tuesday will be at UTC-3. Together with notmyname getting up at 6am every day that should work well.15:43
notmynamettx: ya, next tuesday, if I do any openstack stuff, will have to be early. I've got an all-day event in San Jose for open compute and will need to drive down early15:44
notmynameof course ;-)15:45
notmynamettx: but aside from final=True, I don't think you need me for much15:46
clarkbsdague I did before but am on my phone now15:46
ttxnotmyname: agreed15:47
*** avtar has quit IRC15:48
sdagueclarkb: no worries, if you can find me an example, that would be super handy. I thought I handled the case, but I'm sure there are bugs15:48
clarkbsdague I will try reproducing when at a real computer again15:49
*** dripton is now known as dripton_afk15:51
mordrednotmyname: yup!15:51
openstackgerritEmilien Macchi proposed a change to openstack-infra/jeepyb: Update Launchpad projects for puppet modules
clarkbsdague got it on my phone15:52
*** boris-42 is now known as sirob-4215:52
clarkbI think the .gz is important15:52
EmilienMmordred: clarkb: Hello, I've udated jeepyb for new LP pages on puppet. Do I need to do something else ? like authorizing a user in the group ?15:52
mordrednotmyname, ttx: just to be clear - we will release 1.9.1, merge 1.9.1 back to master, tag the revision in master that made the version 1.9.2-dev as 1.9.2 dev, and then we will consider the pbr patch15:53
mordred(all those things in that order)15:53
clarkbEmileinM yes the openstack hudson user needs to be a bug driver15:53
EmilienMclarkb: thx !15:53
sdagueclarkb: nice15:54
sdagueok, I can work on that15:54
notmynamemordred: that will but the merge commit "after" (in the DAG sense) the 1.9.2-dev tag15:54
ttxmordred: sounds good15:54
*** emagana has joined #openstack-infra15:54
openstackgerritA change was merged to openstack/requirements: use dev-requirements for global sync
*** dprince has quit IRC15:56
jeblairmordred: your final comment on this etherpad is particularly insightful:
*** sarob has joined #openstack-infra15:58
notmynamemordred: I don't really care where the 1.9.2-dev commit is. get approvals on the pbr patch and then let's tag master right before it lands. or whatever15:59
mordrednotmyname: great15:59
mordredjeblair: I try to be helpful15:59
notmynamemordred: I think we've got a good enough sense of the issues to pay attention and not let it break. :-)16:00
jeblairmordred: i hate to say this, but there was a lot of back and forth in the scrollback.  when you have a minute, can you give a short summary of what is changing?16:00
*** vijendar has joined #openstack-infra16:02
*** ruhe has quit IRC16:02
*** sarob_ has joined #openstack-infra16:04
*** sarob has quit IRC16:04
mordredjeblair: sure. on the tagging thing with swift?16:06
ttxjeblair: agree on using graphite for job health16:06
ttxjeblair: so I guess I just need to solve the API call thing16:06
*** comstud is now known as bearhands16:07
mordredjeblair: sure. it's actually simple. once the next swift release is cut, we're going to start merging swift's m-p branches back into master once we're done with them16:07
mordredjeblair: so that git tag --contains will work as expected16:07
mordredjeblair: then, once that's done, we're going to add a 1.9.2-dev tag, purely as a transitional helper16:07
mordredjeblair: so that landing the pbr patch will not cause them to jump backwards in version number space16:07
mordredwe wil not make subsequent -dev tags16:08
jeblairttx: you might look at dansmith's tool:
jeblairttx: which grabs info from zuul's status.json.  i would look for the tarball jobs in the release pipleine, and then hit the jenkins api at the urls associated with them (and the rest of the script should be the same)16:09
jeblairttx: that should work for now, and we can do something more elegant in the future16:09
ttxjeblair: ok, thx for the tip16:09
ttxjeblair: may I ask why we use this multi-master madness ? load ?16:10
jeblairmordred: i don't think i follow the '1.9.2-dev' bit16:10
ttxjeblair: I think you didn't post about it yet :)16:10
jeblairttx: yes, load.  and that's true, i haven't.  :(  also, it's an active-active system, so we can upgrade jenkins with no downtime16:11
ttxjeblair: sounds like a good thing to have.16:11
*** markmcclain has joined #openstack-infra16:12
* ttx calls it a week. back on tuesday.16:12
jeblairttx: have a nice long weekend!16:12
anteayabye ttx16:12
ttxha. One more email to post.16:12
jeblairmordred: what does a 1.9.2-dev tag solve?16:13
mordredjeblair: they currently have their in-tree version set to 1.9.2-dev16:14
mordredwhich means that installations of it currently are getting 1.9.2 as a version16:14
mordredjeblair: if we move them to pbr, their versions will start to be
mordredwhich is less than what they're installing now16:15
jeblairmordred: gotcha16:15
*** ^d has quit IRC16:16
*** ^d has joined #openstack-infra16:19
*** dripton_afk is now known as dripton16:19
*** zul has joined #openstack-infra16:19
*** devananda is now known as vandervenv16:24
*** pcrews has quit IRC16:24
*** mriedem has quit IRC16:25
*** BobBall is now known as BobBall_AFK16:25
*** dprince has joined #openstack-infra16:27
*** krtaylor has quit IRC16:29
*** reed has joined #openstack-infra16:29
*** vijendar has quit IRC16:32
*** derekh has quit IRC16:34
fungittx: here's a quick and dirty routine to pull your jenkins url from the zuul status json based on a git show-ref on the tag name...
*** vogxn has quit IRC16:35
*** dprince has quit IRC16:36
fungittx: the example runs there were of branch tarball jobs because i didn't have time to wait and catch a pre-release or release tag16:37
mgagnemordred: as a LP project group admin, can you edit all info as you would be able to do with a normal project? (description, summary, admin team, etc.)16:37
*** mriedem has joined #openstack-infra16:38
*** sirob-42 has quit IRC16:40
*** seagulls_at_dump is now known as seagulls16:42
*** burt has joined #openstack-infra16:43
*** jpeeler has quit IRC16:43
jeblairmgagne: i don't believe so; we make all our projects owned by a single group for that purpose16:44
mgagnejeblair: we don't anymore =)16:44
jeblairmgagne: i don't believe so; we make all our projects owned by a single _team_ for that purpose16:44
mgagnejeblair: we will now have a project per module16:44
jeblair^ (slightly more clear)16:44
uvirtbotjeblair: Error: "(slightly" is not a valid command.16:44
mgagnejeblair: oh ok, I thought you were referring to the review I commented on16:45
jeblairmgagne: no, was answering your mordred question16:45
mgagnejeblair: I therefore have to ask LP support each time I want to update a field? How inconvenient...16:46
jeblairmgagne: er no...16:46
jeblairmgagne: you just be in the team that owns each project16:46
mgagnejeblair: I will be. My question is regarding the fields in the LP project group like the description, not the projects themselves.16:47
jeblairmgagne: oh... i think the project group has an owner too16:47
zulso im looking at python-keystoneclient for porting to python3, i need to use some temporary workarounds (using mox3 and a newer oslo.config) should i throw that in a test-requirements-py3/requirements-py3.txt16:47
jeblairmgagne: eg:  maintainer: openstack admininstrators16:48
mgagnejeblair: yes, initial request for the creation of the group requires to contain all kind of info and I'm worried I won't be able to update those fields by myself later on.16:48
*** jog0 is now known as flashgordon16:48
jeblairmgagne: i haven't made a group (mordred is one of the few ppl who have), but i think if you specify that you or a team you are an admin of is the maintainer, you should be able to edit whatever.16:49
*** nicedice_ has joined #openstack-infra16:49
jeblairmgagne: i will verify that i can edit the things you specifically listed if you want16:49
mgagnejeblair: cool, that would make me reassured16:50
jeblairmgagne: yes, the 'change details' page looks just like a regular project, with edit fields for description, summary, etc.  there is also a 'change maintainer' page that works the same.16:50
mgagnejeblair: cool! thanks!16:50
*** ruhe has joined #openstack-infra16:52
*** ruhe has quit IRC16:52
*** jpich has quit IRC16:54
*** nati_ueno has joined #openstack-infra16:56
openstackgerritEmilien Macchi proposed a change to openstack-infra/jeepyb: Update Launchpad projects for puppet modules
*** jfriedly has joined #openstack-infra17:00
*** ruhe has joined #openstack-infra17:00
*** reed has quit IRC17:01
*** ruhe has quit IRC17:05
*** dprince has joined #openstack-infra17:07
*** krtaylor has joined #openstack-infra17:11
*** jerryz has joined #openstack-infra17:15
openstackgerritA change was merged to openstack/requirements: Revert "remove netifaces as a requirement"
openstackgerritA change was merged to openstack-infra/jenkins-job-builder: Added pre-scm-buildstep wrapper
*** fbo is now known as fbo_away17:26
*** ewindisch has joined #openstack-infra17:30
fungimordred: when you get a moment, can you confirm you're okay with the tweak i had to make to to work around that jjb parser issue?17:31
*** mikal has quit IRC17:41
*** gyee has joined #openstack-infra17:41
*** mikal has joined #openstack-infra17:42
*** pcrews has joined #openstack-infra17:43
*** vipul is now known as vipul-away17:49
*** ArxCruz has quit IRC17:52
*** ArxCruz has joined #openstack-infra17:55
openstackgerritJeremy Stanley proposed a change to openstack-infra/git-review: Migrate to pbr.
*** changbl has joined #openstack-infra17:56
*** jpeeler has joined #openstack-infra17:58
*** jpeeler has quit IRC17:58
*** jpeeler has joined #openstack-infra17:58
jgriffithanybody have any updates on this?
uvirtbotLaunchpad bug 1210520 in devstack "permission denied prettytable-0.7.2.egg-info/top_level.txt" [Undecided,In progress]18:01
jgriffithThe chmod interstingly doesn't work in my case18:01
jgriffithdtroyer: ^^18:02
openstackgerritJeremy Stanley proposed a change to openstack-infra/git-review: Implement integration tests.
dtroyerjgriffith: on precise?  the chmod doesn't fix it but it should let pip run again.   do you have pip 1.4?18:05
jgriffithdtroyer: checking18:06
jgriffithdtroyer: so the chmod actually worked, just still get a keystone failed.  Looking at screen it's running though...18:06
jgriffithdtroyer: lemme kick off again18:06
jgriffithdtroyer: and yes... I have 1.418:07
*** jpeeler has quit IRC18:07
*** jpeeler has joined #openstack-infra18:07
dtroyerI'm doing another fresh precise run…I think this is caused by out of date requirements somewhere but haven't nailed the root cause yet.  When I saw it some combination of reinstalling packages with pip and updating repo requirements fixed it.18:10
dtroyerit's always prettytable too…that should be a clue ;)18:11
jgriffithdtroyer: yeah, that's actually where I started on this yesterday18:11
jgriffithdtroyer: but pip *broke* I ended up nuking everything and starting over18:11
jgriffithdtroyer: anyway... I'll get back to it and see if I can figure out the deal... Prettytable perms are ok with the chmod18:12
jgriffithdtroyer: just still failing to start for some reason (keystone)18:12
jgriffithdtroyer: AHH!  got it18:12
jgriffithnever mind18:12
jgriffithbad localrc fiel18:12
dtroyerwhat was bad?18:13
jgriffithHost IP18:13
jgriffithso even though keystone started the check to connect to it failed :(18:13
jgriffithsorry bout that18:14
*** vijendar has joined #openstack-infra18:17
*** mrmartin has joined #openstack-infra18:17
*** vipul-away is now known as vipul18:18
*** pcm_ has joined #openstack-infra18:29
pcm_Hi - have a problem with pbr package in devstack18:29
pcm_can someone advise how to fix?18:29
pcm_I did a "git review -d 40993' to pick up a change and then tried to stack.sh18:30
pcm_It gave this error:18:30
pcm_pkg_resources.VersionConflict: (pbr 0.5.17 (/opt/stack/pbr), Requirement.parse('pbr>=0.5.20'))18:30
pcm_My "pip freeze" shows:18:30
pcm_-e git+
pcm_I'm not sure how to resolve. Any advice?18:31
vishymordred: ping18:33
clarkbpcm try rebasing the pbr change on tip of master so it has the latest version18:34
pcm_So update the devstack repo to latest?18:34
clarkbno the pbr repo18:35
clarkbI think 40993 may be based on an older version of pbr hence the conflict18:36
pcm_clarkb: can you elaborate? Not sure how to do that (or where the pbr repo is located)?18:36
pcm_clarkb: I was using latest Havana, then pulled review for 33148 an now trying 40993, BTW.18:37
clarkb40993 is version 0.5.17 (or based on it) something else wants 0.5.2018:37
pcm_clarkb: I see there's /opt/stack/pbr18:37
clarkbif you rebase the change onto tip of master pbe when it installs it should install as the latest version18:38
openstackgerritmark mcclain proposed a change to openstack/requirements: update mock minimum to 1.0
pcm_clarkb: Ah I see. BTW pbr is using: 182feb3 Fix pip invocation for old versions of pip.18:39
clarkb(this is my theory at least)18:40
pcm_clarkb: Gave it a shot... looks like some conflicts :(18:42
openstackgerritA change was merged to openstack-infra/jenkins-job-builder: Fix handling of optional parameter description
pcm_clarkb: Do you think it would be easier to run devstack with OFFLINE=False, RECLONE=on and then re-patch 40993?18:43
dtroyerI find RECLONE=yes to be extremely helpful in transitory times like this.  And note, the value isn't normalized, it must be 'yes'.  #NTS fix that…18:46
openstackgerritElizabeth Krumbach Joseph proposed a change to openstack-infra/config: Enable ssl module in httpd on
mrmartinHi - What is the process of adding a new server to ci infrastructure? I've read, it is  clear how can I add it to puppet, but do I need to open a blueprint / bug request on launchpad, or just required to simply submit a patch for review?18:49
fungimrmartin: just a patch for review with a clear commit message should be plenty18:49
pcm_dtroyer: I'll give that a try. Think I should delete /opt/stack to make sure it is clean?18:49
fungimrmartin: and then we'll guide your patch to correctness through the review process18:50
mrmartinfungi: thanks, and do you know who have the permission to create git repository for the project?18:50
*** jpeeler has quit IRC18:50
dtroyerpcm_: that couldn't hurt, especially if you think there's brokenness in /opt/stack/data.  usually is enough though18:51
fungimrmartin: in fact, you do (indirectly). it also happens through submitting a code review against that same openstack-infra/config project18:51
dtroyerwith RECLONE=yes that is...18:51
fungimrmartin: a guide to how we do this for stackforge projects, since they're mostly self-service hosting, is available at
fungimrmartin: for an official project in openstack, openstack-dev or openstack-infra orgs we can do that for you, if you prefer18:52
pcm_dtroyer: giving it a shot. If I comment out the RECLONE line, it defaults to yes, right?18:52
fungimrmartin: though the process is basically still the same18:52
dtroyerit defaults to unassigned which is functionally 'no'18:53
mrmartinfungi: it will be the official community portal, at and groups.openstack.org18:53
fungimrmartin: yep, we can certainly help with that to get things giong faster18:53
fungier, going18:53
mrmartinthis stackforge description is very useful18:54
*** jpeeler has joined #openstack-infra18:54
*** jpeeler has joined #openstack-infra18:54
jeblairmrmartin: yeah, it's basically that process but you can use 'openstack-infra/' instead of 'stackforge/'18:54
mrmartinhi jeblair18:55
mrmartinok, I need to understand the entire process, what I see, that we need to create drupal environment under groups-staging.openstack.org18:56
pcm_dtroyer: Gotcha. Thanks!18:56
*** markmcclain has quit IRC18:57
mrmartinand need to create the project under different places of openstack-infra/config (as described in stackforge page)18:57
mrmartinand also need to create specific gating tasks18:57
fungimrmartin: just for a bit clearer and more concise example, here's where we added the storyboard project recently...
jeblairmrmartin: you'll only need a new git repo when there's code, but i think you just wanted to start with setting up a server and running drupal, right?18:58
mrmartinYeap, I'll start with setting up the server, it seems to be straightforward to me.18:58
mrmartinwhat is missing to me currently, if we'll create a for production, and we are pushing accepted commits to groups-staging.openstack.org19:00
mrmartinhow could we initiate pushing changes from staging to prod site?19:00
fungimrmartin: tagging might be a good fit there19:01
mrmartinMaybe different git branches could be also a solution?19:01
fungimrmartin: basically that way the staging server gets a continuous-deployment watching the tip of the branch, but the production server only updates when it sees a new tag19:01
mrmartinfungi: do you know an example in current infra project for this type of deployment pattern?19:02
fungimrmartin: not off the top of my head, but jeblair might know of one i'm forgetting19:04
vishyhas anyonge gotten pbr + debian packaging working yet?19:04
fungimrmartin: branches could work too though. basically have master and production branches and then cherry-pick backported changes from master into production. it's a bit more work that way but could be more flexible if your development is non-linear19:05
fungimrmartin: the tag solution would be more appropriate if your production version merely lagged behind staging until you were satisfied with the staging version19:05
fungivishy: i think pabelanger and zigo both have19:06
mrmartinfungi: we have a continuous development process, but releasing to prod after every sprint (2week period)19:06
fungimrmartin: right, so probably at the end of the sprint you just tag the tip of master with a new release version number and we could have the production server only update to tag refs rather than following the master branch continuously19:07
mrmartinok, I'll be happy with a working staging system first :)19:08
fungimrmartin: absolutely. i agree that's the right place to start19:08
mrmartinguys, thanks for the answers, I'm going to make some footprint in the infra repo.19:08
fungimrmartin: and we definitely have existing examples of using the pupet vcsrepo module to update software following the tip of a specified branch name (master or whatever) from an arbitrary git remote url19:09
mrmartinthanks, I saw that.19:09
fungimrmartin: git grep vcsrepo in that openstack-infra/config project to see several19:09
fungiahh, perfect19:09
dtroyerdstufft: pip 1.4 question:  did pip used to fix permissions on installed files?  prettytable has this: and now it's borken.  chmod +r fixes it of course, I'm trying to figure out why now since 0.7.2 was released in April19:09
mrmartinfungi: this will be a drupal based project, so upgrade is a bit tricker than checking out from git repo.19:10
fungimrmartin: well, the module can certainly update a local source tree on the server when there's a new commit and then subscribe some arbitrary scripts/commands to that so they run automatically when that happens19:11
dstufftdtroyer: not that I'm aware of. it could also be a setutools19:11
dtroyerah, so it's still precise-era setuptools.   I'm trying not to replace that due to the package implications.19:12
pabelangerfungi, vishy: Not yet, I plan to package something in the next week or so19:12
pabelangerwill have to do it for JJB and zuul shortly19:12
*** sarob_ has quit IRC19:13
fungipabelanger: ahh, do you happen to know if zigo has debianized any of the newer pbrish versions of the official openstack projects?19:13
*** sarob has joined #openstack-infra19:13
* fungi hasn't been keeping up19:13
pabelangerfungi, not sure to be honest.  Would have to look at his packaging repo19:13
pabelangerbut exists19:14
vishypabelanger: pbr seems to be blowing up our debian builds19:14
pabelangerso I would _assume_ yes19:14
fungivishy: zul may also have some suggestions since i think he packages some current openstack stuff for ubuntu/canonical19:14
vishypabelanger: i'm not actually as concerned with bpackaging pbr itself, more about packaging packages that use pbr19:14
pabelangervipul, any logs?19:14
dtroyerdstufft: ok, thanks.  going back to pure precise to test again.   just for my sanity, prettytable should have those files world-readable, right?19:14
*** nayward has quit IRC19:15
dstufftdtroyer: yea I believe so, the directory it's installed into is world readable right?19:15
zulfungi:  i got the new pbr out this week19:15
dtroyeryes, just the contents aren't.  they're in the zip that way19:16
vishypabelanger: trying to get them but it was something to do with not being able to find a git repo19:16
fungizul: have you packaged anything yet which uses pbr?19:16
vishygot someone coming along with the exact error19:16
zulfungi:  yeah all of the havana packages in ubuntu use it19:16
fungivishy: ^19:16
vishyzul: did you have to do anything tricky to get it to build19:16
vishyzul: error we are getting is there ^^19:17
zulvishy:  nope just added them as a build depends19:17
*** sarob has quit IRC19:17
pabelangerHmm, looks like an issue with python-cinderclient-1.0.5/setup.cfg19:17
vishyi seem to remember having to manually create a PKG-INFO in the past to get around this19:17
vishy* issues similar to this19:18
zulvishy: whats the issue?19:18
vishysee the gist above19:18
jfriedlyYeah, creating a PKG-INFO was going to be the next thing I tried19:18
zulvishy: oh...that19:19
jeblairvishy: why not use an sdist tarball?19:19
*** mrodden has quit IRC19:19
jeblairjfriedly: ^?19:19
jfriedlyjeblair:  I'm open to the idea, but I don't know what an sdist tarball is.  Googling it now19:20
zulvishy:  you are buidling from git directly? you have to export PBR_VERSION=<your version number> before creating your tarball19:20
vishyjeblair: well we just copied the existing ubuntu control files19:20
zuli ran into this yesterday19:20
vishyzul: we use pbuilder19:20
jeblairvishy, jfriedly:
vishyzul: so we put that into build string?19:20
zulvishy:  doesnt matter its a pbr thing19:21
jeblairthere's the docs that explains what's going on there19:21
zulvishy:  yeah19:21
dstufftdtroyer: if you narrow it down to a versio of pip that works and a version that doesn't can probably figure out what changed19:21
vishyso PBR_VERSION or OSLO_PACKAGE_VERSION will work19:21
vishymordred: any tips on where exactly that goes in the rules file?19:22
mordredvishy: anywhere. probably up top is easiest19:22
vishyah yes i remember this now19:22
fungivishy: you can assign enviroment variables anywhere you like in a makefile, as long as it's before it's used19:22
vishythis is why I added PKG-INFO in the past19:22
vishyseemed easier to maintain then hacking up rules19:22
mordredalso - you if you can consume the tarball we build, you don't have to do anything19:23
mordredbut whichever thing works for you19:23
vishymordred: yeah our packaging infrastructure isn't really set up to pull in external tarballs and buidl packages yet19:23
jeblairjfriedly, mordred: branch19:23
mordredvishy: nod.19:23
zulvishy: is your packaging sources up somewhere i would love to see what you guys are doing19:23
vishyalthough it probably needs a bit of an overhaul19:23
vishyzul: no not really19:24
zulvishy: dang19:24
vishywe mostly just copieed your stuff and then break it :)19:24
mordredvishy: if, at some point in the future, you can separate the python sdist step from the pbuilder step, it'll be more resilient in the long rung19:24
* zul covers up my answers19:24
jeblairjfriedly, vishy: we publish sdist tarballs here:  ; they are made with 'python sdist'19:24
vishymordred: I think we were doing that with git-build-package for some of mour packages19:24
fungivishy: gbp can suck in tarballs (look into the pristine tarball branch feature)19:25
vishybut using git-buildpackage when ubuntu doesn't was making it hard to pull in changes iirc19:25
*** ^d is now known as ^demon|away19:25
mordredvishy: I've been meaning to put together a simple g-b-p recipe for folks19:25
vishyfungi: yeah i remember that19:25
vishyi went through all of this when i upgraded us from oneiric to precise and then again with essex -> folsom and folsom -> grizzly19:26
fungiso if you don't want to download ours, just running the sdist first and then importing that in your tarball branch should theoretically work19:26
vishyall in all I've decided that packaging sucks :(19:26
pabelangervishy, what is the benefit for you to import the source with gbp?19:26
zulvishy: welcome to the club19:26
vishypabelanger: I don't know if we are actually using gbp currently19:27
fungieverything sucks. my goal is to use the things which suck least19:27
pabelangervishy, okay19:27
vishybut a lot of our packages have backported patches19:27
vishyso for consistency we do everything with source19:27
pabelangerquilt works nicely for that :D19:27
vishyoh god19:27
vishyyeah if you don't like dvcs then quilt is perfect19:28
fungiquilt + gbp is a little messy honestly19:28
fungiit's like layering pre-rcs patch handling on top of a good vcs19:28
vishymanually updating quilt patches and rolling out changes is really hard19:28
*** rscottcoyle has quit IRC19:28
vishyI like my git cherry-pick personally :)19:28
pabelangerI don't mind it, might you I am only backporting a few patches for my needs19:29
fungiworth noting, however, that v3 packages do support a single-patch workflow where you can squash your branch when constructing the quilt set19:29
vishyin our folsom nova we had roughly 40 patches at one point19:29
vishya lot of those got pulled in to the stable release but rebasing onto stable with 40 patches using quilt sounds like a recipe to go insane19:29
fungiso the resulting package only carries one patch inside the deb which is the diff between the tarball and your modifications that way19:30
fungiconstructed from a squash during the build process19:30
zulheh try using dpatch19:32
*** sarob has joined #openstack-infra19:33
*** emagana has quit IRC19:33
fungiyeah, deb src 3.0(quilt) is on my list of things that suck less than dpatch19:34
*** mrodden has joined #openstack-infra19:34
dtroyerdstufft: heh, pip 1.0 (via precise python-pip) fixes the permissions.  appended to the paste:
openstackgerritMonty Taylor proposed a change to openstack-dev/pbr: Added documentation for packagers
openstackgerritMonty Taylor proposed a change to openstack-dev/pbr: Update requirements in integration test
*** emagana has joined #openstack-infra19:35
*** mrodden1 has joined #openstack-infra19:35
dstufftdtroyer: it'd be great if you could try 1.1, 1.2, 1.3, 1.4 and see where it starts failing19:35
dtroyernow you're cutting into my quiet time on the airplane… :)19:36
mordreddtroyer, dstufft aroo?19:36
dtroyermordred: aroo!19:36
*** mrodden has quit IRC19:38
*** melwitt has joined #openstack-infra19:39
*** reed has joined #openstack-infra19:43
mordredjeblair: ? so that when I land the pbr doc patch, the docs will actually publish?19:44
jeblairmordred: lgtm (did not aprv)19:49
mordredjeblair: great - thanks -mind if I do?19:50
dtroyerdstufft: it stars with 1.4, 1.4.1 still does it.19:53
fungimordred: went ahead and approved since i +2'd it previously19:53
mordredfungi: yay!19:53
mordredthank yous19:53
*** sarob has quit IRC19:53
dstufftdtroyer: ok 1.3.1 does not?19:53
dtroyerlet me try19:53
*** sarob has joined #openstack-infra19:53
fungimordred: when you get a moment, can you confirm you're okay with the tweak i had to make to to work around that jjb parser issue?19:54
openstackgerritA change was merged to openstack-infra/config: Fix pbr doc jobs
dtroyerdstufft: 1.3.1 is good19:55
dstufftdtroyer: ok thanks19:55
*** sarob has quit IRC19:58
harlowjamordred qq, is it recommended to start using the pbr addition (i think) of python  test and python  flake8 yet20:01
harlowjaor is that more experimental20:01
mordredharlowja: it should work - and I welcome any bugs pointing to the contrary!20:01
harlowjawas thinking about making anvil which can act as the way to call out to all the components test just use that20:02
harlowjainstead of via
harlowjadoes it transparently handle like some projects using nose still, or mostly just testr?20:02
*** nayward has joined #openstack-infra20:03
*** mrmartin has quit IRC20:03
*** odyssey4me has quit IRC20:05
*** psedlak has quit IRC20:06
*** zul has quit IRC20:06
dstufftdtroyer: nothing stands out at me in the diff betwen 1.3.1 ad 1.4, can you file a ticet?20:08
dtroyerdstufft: sure…I'll try on fedora too in case that sheds any light.20:10
*** weshay has quit IRC20:13
*** odyssey4me has joined #openstack-infra20:13
openstackgerritA change was merged to openstack-infra/config: Gate jeepyb and openstack/requirements on pbr
*** vipul is now known as vipul-away20:17
openstackgerritMonty Taylor proposed a change to openstack-infra/devstack-gate: Enable flag to update requirements on run
mordreddtroyer: I just updated the devstack change around requirements updates as you suggested20:19
mordreddtroyer: but I'd like to land that patch ^^ first20:19
dtroyermordred: roger.  you'll need a rebase because I stuck in your only-if-clean test with the change, go ahead and nuke that bit20:21
mordreddtroyer: you know what - I think I like sdague's solution better20:24
sdaguedtroyer: yeh, see -qa, we're talking about it over there20:25
*** mjblack_ has joined #openstack-infra20:26
fungijeblair: mordred: clarkb: over ram quota for several days now on hpcloud-az2 apparently?
* fungi is checking how many machines we have running there20:31
fungi96 running in az220:33
jeblairfungi: i think that's tops20:34
*** vipul-away is now known as vipul20:34
fungi53 in az1 and 57 in az3 by comparison20:35
fungiso if there's not cruft in az2 then it's just way, way, WAY imbalanced20:35
jeblairfungi: the devstack launchers think that there are 0 machines in az220:36
fungiso i should nova delete them all?20:36
*** mjblack_ has left #openstack-infra20:36
jeblairlemme check jenkins-dev20:36
jeblairsame situation20:37
jeblairfungi: yeah, go for it.20:38
jeblairfungi: i'm going to skip figuring out why this happened in favor of continuing to hack on the thing that will replace it20:40
fungijeblair: please don't let me stop you20:40
fungii'm still reviewing your wip version, but needed to reeducate myself about python decorators to do so20:40
*** ^demon|away is now known as ^demon20:42
*** avtar has joined #openstack-infra20:42
jeblairfungi: fyi the nodes/images i'm creating while testing all have 'nodepool' in their names.  anything prefixed with dev- should be associated with jenkins-dev, and the rest, production.20:43
fungijeblair: oh, awesome20:43
vipulhey you guys aware of devstack failures?20:44
vipul19:55:05 IOError: [Errno 13] Permission denied: '/usr/local/lib/python2.7/dist-packages/prettytable-0.7.2.egg-info/top_level.txt'20:44
vipulkeystone doesn't start20:44
*** lcestari has quit IRC20:45
fungivipul: i believe sdague and dtroyer were digging into it20:45
vipulawesome, thanks fungi20:46
fungivipul: something about newer versions of pip deploying prettytable with non-world-readable contents20:46
sdaguevipul: yeh, the ghetto work around20:47
sdaguechmod -R a+r /usr/local/lib/python2.7/20:47
vipulhey as long as it work :D20:47
sdaguedtroyer is working on a real fix, but it probably won't be in tree until monday20:47
*** woodspa has quit IRC20:55
*** changbl has quit IRC20:58
*** thomasem has quit IRC20:58
*** burt has quit IRC21:00
*** datsun180b has joined #openstack-infra21:01
*** datsun180b has quit IRC21:02
*** leifmadsen has quit IRC21:02
*** datsun180b has joined #openstack-infra21:03
openstackgerritJordan OMara proposed a change to openstack-infra/config: Add tuskar-ui to stackforge
*** erfanian has joined #openstack-infra21:07
mgagneIn doc/ of openstack-infra/config project, where is openstack/layout.html located? This file is mentioned in layout.html21:07
*** dprince has quit IRC21:07
*** changbl has joined #openstack-infra21:07
mgagneanswering myself:
mordredmgagne: :)21:09
*** jpeeler has quit IRC21:09
*** ewindisch has quit IRC21:10
*** vijendar has quit IRC21:10
*** pcm_ has quit IRC21:17
*** nayward has quit IRC21:17
*** rnirmal has quit IRC21:24
mordredfungi: I think our new integration gate patch broke something21:26
sdagueok... so while we were passing tests for 4 hours, neutron totally broke itself again21:27
mordredthat's great21:27
*** nati_ueno has quit IRC21:29
mordredfungi: but the script on jenkins looks right21:29
mordredif [ "mirror" == "no-mirror" ]; then21:29
openstackgerritMonty Taylor proposed a change to openstack-infra/config: Shell functions need {}'s not ()'s
mordredfungi: ^^21:35
mordredfungi: I have no idea how either of us missed that21:35
flashgordonfYI I got a ssl warnign for
jeblairflashgordon: yep, it's a self-signed cert21:35
flashgordonjeblair: just making sure ta ht was expected21:36
jeblairflashgordon: yeah, not sure how many or how important jenkins masters will be in the long run; didn't seem worth it atm.21:37
fungimordred: we'll need to revert the original change or fast-track my jjb fix then because that was the workaround for it21:40
mordredfungi: can you point me at your jjb fix?21:40
*** mriedem has quit IRC21:41
*** rfolco has quit IRC21:41
mordredfungi: let's revert the patch for now -I agree with your comment, I think the jjb fix needs more eyes21:42
fungimordred: okay, revert incoming21:42
openstackgerritJeremy Stanley proposed a change to openstack-infra/config: Revert "Gate jeepyb and openstack/requirements on pbr"
fungimordred: ^21:44
sdaguehey, how easy would it be to flush the gate?21:45
sdaguebecause we really need to turn back on tempest actually voting things down21:46
openstackgerritA change was merged to openstack-infra/config: Enable ssl module in httpd on
jeblairsdague: what did i miss?21:46
sdaguejeblair: the tox rule for testr was botched21:46
jeblairsdague: and not too hard; i'm starting work on it while you fill me in21:46
sdagueso it always passed21:46
sdagueneutron already landed a break21:47
jeblairsdague: ok, and is fixed now?21:47
sdaguewhich is preventing us from turning it around21:47
sdaguejeblair: this will
jeblairsdague: which are the changes that will fix things?21:47
sdaguejeblair: so we don't know how to fix neutron now21:47
sdagueso I just skipped the test that's preventing us from landing21:48
sdaguebut it ran up earlier and bounced off the gate21:48
openstackgerritA change was merged to openstack-infra/config: Revert "Gate jeepyb and openstack/requirements on pbr"
sdaguejeblair: but basically if any other breaks land... we're in a race21:50
sdagueso I'd rather flush the entire gate, let us land that change, then ensure people can't land more breaks21:50
jeblairok, i'm ready to restart zuul now, and will reverify that change first21:50
jeblairsdague: er, you need to aprv it21:52
sdaguejeblair: cool21:52
sdaguejeblair: yes, I'm waiting for mtreinish to make sure I didn't screw it up21:52
*** prad has quit IRC21:52
sdaguesorry, one sec21:53
*** avtar is now known as idililbuddy21:53
*** idililbuddy is now known as idilogger21:53
*** ^demon has quit IRC21:53
*** ewindisch has joined #openstack-infra21:53
*** idilogger has left #openstack-infra21:53
*** pentameter has quit IRC21:54
jeblairi'll cancel jobs while we're waiting21:54
sdaguejeblair: mtreinish is making sure we got the skip tracker bits right21:54
mgagnegit.o.o is down. Is it related to recently merged change for ssl support?21:55
mtreinishsdague: I pushed it through it skips fine21:56
mtreinishI'll fix the skip tracker pick up later21:56
jeblairmgagne: i'm not sure git.o.o is quite in production yet21:56
sdaguejeblair: ok, we're approved on the patch21:56
*** datsun180b has quit IRC21:57
sdagueso let zuul run any time21:57
mordredmgagne: actually, the recently merged patch is intended to fix that21:57
mordredbut replication from gerrit to git.o.o hasn't turned on yet (we're waiting on restarting gerrit)21:57
mgagnemordred: ok. just wanted to let you know. Used to partially work before21:57
mordredmgagne: yup. thanks! it should be back up soonish21:57
jeblairsdague: you're at the head of the queue21:58
notmynamemordred: quick follow-up to this morning discussion; what do the other projects do with pbr? how does it work for them?21:58
mordrednotmyname: the client projects use it the same way as swiftclient - which is closer to swift usage21:58
mordredthe server projects all do pre-release versioning21:59
notmynamemordred: ah ok :-)21:59
mordredso they have a version field in setup.cfg21:59
notmynamemordred: so we're somewhat figuring this out fist ;-)21:59
mordredalso, we've basically said that the generated version number made by pbr for the pre-release stuff canot be counted on to always be increasing21:59
mordrednotmyname: and fist :)21:59
mordredfor post-release versions it should always increase22:00
mordredbut yeah - you'll be the first to do post-release _and_ milestone-proposed22:00
dtroyersdague: is the oslo fix ready to go?22:02
flashgordonsdague: unless you object I will cut hacking 0.7 today22:02
flashgordonand after gets merged22:03
dtroyeralso, I just pushed up for the prettytable permissions bits.  it works for me on precise, I'll try to test fedora on the plane...22:03
flashgordonsdague: hacking 0.7 has a H202 fix and custom ignore import rules22:03
sdaguedtroyer: the oslo fix should be good22:07
openstackgerritA change was merged to openstack-infra/gitdm: Add to foundation group file
sdaguejeblair: cool22:07
sdagueflashgordon: sounds good to me22:08
*** changbl has quit IRC22:08
flashgordonsdague: can you review mordred's patch above ^^22:08
flashgordonits the globalreqs sync22:08
*** ewindisch has quit IRC22:08
mordredsdague: sob. "Requirement oslo.config>=1.1.0 does not match openstack/requirements value oslo.config>=1.2.0a3"22:09
sdaguemordred: hmph22:09
mordredthat's in the check job22:09
mordredso, did the right thing, and did not change the valuebecause it matched the previous valie22:10
sdaguemordred: oh, the check job22:10
sdagueyeh, that's a different fix22:10
* mordred goes to work on the wheel stuff - this is chasing rainbows22:10
sdagueunicorns and rainbows, that's what openstack is all about22:10
flashgordonmordred sdague:
sdaguewell, my wife is about to show up to pick me up, so I'm about done for the night. Maybe I'll look tomorrow morning if it's still bust22:11
sdaguewell at least our slip up on the testr change over I think is inspiring us to unit test everything in tools in tempest :)22:12
sdaguemodeled on the testing in requirements22:12
jeblairtest all the things! :)22:12
fungii think i need to go do something about dinner22:13
* fungi wil bbiaw22:13
sdaguemore importantly, any time you make a mistake, make sure you figure out a way to never make it again22:15
sdagueok, well the problematic test passed22:17
sdagueso I'm hoping it's good22:17
sdaguejeblair: if you could keep an eye on that top job22:18
sdagueif it goes bad we're in trouble22:18
jeblairsdague: i love how if i miss a parenthesis 200 lines above where i am in emacs, i know because it refuses to indent properly22:18
sdaguebut I have to run22:18
jeblairsdague: will do22:18
sdaguejeblair: I'm with you :)22:18
openstackgerritA change was merged to openstack-dev/hacking: Synced with global requirements
*** ianw has quit IRC22:22
*** ianw has joined #openstack-infra22:23
openstackgerritArnaud Fabre proposed a change to openstack-infra/jenkins-job-builder: RFC: Use a more consistent style for booleans values.
*** rcleere has quit IRC22:29
*** dkliban has quit IRC22:30
*** sarob has joined #openstack-infra22:31
*** vipul is now known as vipul-away22:32
*** vipul-away is now known as vipul22:34
openstackgerritMonty Taylor proposed a change to openstack-infra/jeepyb: Start building wheels in the mirror
mordredfungi, sdague, dstufft ^^22:39
*** ArxCruz has quit IRC22:40
openstackgerritMonty Taylor proposed a change to openstack/requirements: Bump the pip version and add wheel
openstackgerritArnaud Fabre proposed a change to openstack-infra/jenkins-job-builder: Allow defaults to inherit each other.
*** pcrews has quit IRC22:46
*** jaypipes has quit IRC22:47
*** retr0h has quit IRC22:50
openstackgerritArnaud Fabre proposed a change to openstack-infra/jenkins-job-builder: Use a more consistent style for booleans values.
sdaguejeblair: that tempest patch looks like almost landed... which is goodness22:51
sdaguemordred: is that why this failed - ?22:52
sdaguethe mirror issue?22:52
sdaguejeblair: and lastly.... css testing - ?22:53
jeblairsdague: sorry, haven't gotten to it yet :(23:03
*** dkranz has quit IRC23:05
*** vipul is now known as vipul-away23:06
*** mjblack has joined #openstack-infra23:10
*** ewindisch has joined #openstack-infra23:10
*** mgagne has quit IRC23:12
fungimordred: wheels! (with flames?)23:13
*** vipul-away is now known as vipul23:13
*** mjblack_ has joined #openstack-infra23:17
*** mjblack has quit IRC23:17
*** retr0h has joined #openstack-infra23:25
reedhappy weekend23:27
*** reed has quit IRC23:28
*** sarob has quit IRC23:29
openstackgerritJoe Gordon proposed a change to openstack/requirements: Bump hacking to 0.7
*** changbl has joined #openstack-infra23:29
*** vipul is now known as vipul-away23:29
flashgordonzuul is just finishing up the 'post-mirror-py*' jobs for ^23:30
flashgordonmordred: this will be the first new hacking release under the new global reqs system23:30
*** sarob has joined #openstack-infra23:30
*** sarob has quit IRC23:31
*** gyee has quit IRC23:33
*** sarob has joined #openstack-infra23:33
*** sarob_ has joined #openstack-infra23:33
*** vipul-away is now known as vipul23:35
*** sarob has quit IRC23:37

Generated by 2.14.0 by Marius Gedminas - find it at!