| pleia2 | mgagne: if we're required to move to it at some point anyway for puppet 3, makes sense | 00:00 |
|---|---|---|
| fungi | mgagne: pleia2: so i have manually confirmed both != nil and .has_key?() do what we want. if the latter is safer, then i agree we should use it | 00:01 |
| *** vipul is now known as vipul-away | 00:01 | |
| pleia2 | fungi: how does .has_key?() syntax work exactly? | 00:03 |
| dstufft | clarkb: thanks, i'll see what I can do. It won't make it into 1.4.1 but I'll try to get it fixed | 00:03 |
| mgagne | fungi: would you allow a nil value to be injected in a template? I'm not sure how puppet translates undef value to erb template variables. Will it still define the variable but set it to nil? | 00:03 |
| mgagne | fungi: in that case, you might want != nil | 00:04 |
| mgagne | fungi: unless a ruby guru hit me with a stick and tell me it's not the correct way to do it | 00:04 |
| pleia2 | fungi: ok, where did you put != nil ? just replaced "" with nil or...? | 00:06 |
| *** ^d has quit IRC | 00:06 | |
| *** sarob has quit IRC | 00:12 | |
| *** sarob has joined #openstack-infra | 00:12 | |
| pleia2 | hm, with this it seems it's less easy to toggle replication for -dev servers | 00:13 |
| *** arosen1 has quit IRC | 00:13 | |
| *** krtaylor has joined #openstack-infra | 00:14 | |
| fungi | pleia2: sorry, was afk or a bit | 00:14 |
| fungi | <% if replication.has_key?('replicatePermissions') -%> | 00:15 |
| fungi | or alternatively i was also able to make this work | 00:15 |
| *** sarob has quit IRC | 00:15 | |
| fungi | <% if replication['replicatePermissions'] != nil -%> | 00:15 |
| pleia2 | let's go with nil | 00:15 |
| fungi | both of those gave us the desired end result across the entire template | 00:16 |
| * pleia2 nods | 00:16 | |
| jog0 | mordred: looks like we have a fun failure http://logs.openstack.org/54/40754/1/check/gate-nova-pep8/035fcb5/console.html | 00:18 |
| *** woodspa has joined #openstack-infra | 00:18 | |
| *** arosen1 has joined #openstack-infra | 00:19 | |
| *** UtahDave has quit IRC | 00:21 | |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Add replication of git from gerrit to git.o.o https://review.openstack.org/37794 | 00:22 |
| *** woodspa has quit IRC | 00:22 | |
| *** woodspa has joined #openstack-infra | 00:23 | |
| pleia2 | so that fixes the immediate problem, but I am pretty sure this makes it so both review and review-dev are now shipped with the same template and we don't want that (review-dev sending updates to git.o.o would not be so good) | 00:23 |
| *** wu_wenxiang has quit IRC | 00:24 | |
| clarkb | pleia2: I don't think there is overlap in project names, but ya that wouldn't be good | 00:24 |
| fungi | pleia2: seems to solve the config file issue on the last patchset | 00:25 |
| pleia2 | fungi: \o/ | 00:25 |
| fungi | pleia2: why shouldn't review-dev send updates there? we have it seding updates to github | 00:25 |
| pleia2 | fungi: I guess it wouldn't hurt if there is no overlap | 00:26 |
| fungi | we have test projects on review-dev which we use to test out things we wouldn't put in a production repo, but it's nice to have them treated the same | 00:26 |
| clarkb | fungi: does cgit give us good separation at the org level? | 00:26 |
| *** sarob has joined #openstack-infra | 00:26 | |
| pleia2 | cgit will only serve up things in the projects.yaml anyway | 00:26 |
| clarkb | my only concner would be having gtest in the same list as nova | 00:26 |
| pleia2 | git:// will serve up everything in /var/lib/git/ though | 00:26 |
| fungi | clarkb: we have org-level separation | 00:27 |
| fungi | in cgit and on the filesystem | 00:27 |
| clarkb | in that case ignore me | 00:27 |
| fungi | it's a naive implementation, but we don't just dump them all repos into the same parent directory | 00:28 |
| fungi | s/them// | 00:28 |
| jeblair | i would rather not have review-dev push to git.o.o | 00:28 |
| pleia2 | hm, how about replicate_local? I don't think gerrit-dev does that, unless I'm missing something | 00:28 |
| fungi | jeblair: do we want to have review-dev also stop replicating to github in that case? | 00:28 |
| *** nijaba has quit IRC | 00:28 | |
| jeblair | if it wants to push to somewhere, we can have it push to git-dev. | 00:29 |
| jeblair | fungi: i don't think it matters; it may as well. | 00:29 |
| pleia2 | so right now the whole replicate [] thing is in gerrit.pp, perhaps this should instead be in review(_dev).pp ? | 00:29 |
| *** nijaba has joined #openstack-infra | 00:29 | |
| jeblair | pleia2: at least; possibly site.pp | 00:30 |
| jeblair | fungi: there's nothing that ties the repos on review-dev that get published to github to openstack; i don't think our silly test repos should be on git.o.o though. | 00:31 |
| fungi | okay, i'll buy that reasoning | 00:32 |
| pleia2 | seems messy for our clean site.pp :) | 00:32 |
| pleia2 | but either way | 00:32 |
| jeblair | pleia2: it does. bit of a grey area. i'd probably go with review(_dev).pp for now. | 00:32 |
| *** sarob_ has joined #openstack-infra | 00:33 | |
| fungi | reasonable since that's where the original switch was exposed to enable/disable replication prior to this refactoring | 00:33 |
| * pleia2 nods | 00:33 | |
| jeblair | (on the one hand it is "how openstack configures the server", otoh, it makes it a bit difficult for us to spin up a dev gerrit instance without getting github replication thrown in to boot) | 00:34 |
| *** arosen2 has joined #openstack-infra | 00:34 | |
| jeblair | but at least it will probably just error out in that case. | 00:34 |
| clarkb | jeblair: I think that is ok, they will just race each other in the force pushing :) | 00:35 |
| pleia2 | so, is gerrit-dev replicating to anything? | 00:37 |
| *** sarob has quit IRC | 00:37 | |
| pleia2 | s/is/should | 00:37 |
| clarkb | at the very least github and the local replica | 00:37 |
| fungi | it currently replicates locally and to github | 00:37 |
| *** sarob_ has quit IRC | 00:37 | |
| *** arosen1 has quit IRC | 00:37 | |
| fungi | er, review-dev not gerrit-dev | 00:37 |
| pleia2 | yeah, that's what I meant | 00:38 |
| pleia2 | ah, I see review-dev.projects.yaml.erb now | 00:38 |
| pleia2 | no overlap | 00:39 |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Add replication of git from gerrit to git.o.o https://review.openstack.org/37794 | 00:42 |
| pleia2 | ok, hopefully that sorts it | 00:42 |
| *** vipul-away is now known as vipul | 00:43 | |
| fungi | pleia2: actually not. you're going to need to pass that replication hash table down from openstack_project::review{,_dev} through openstack_project::gerrit into ::gerrit | 00:50 |
| pleia2 | fungi: ah right | 00:51 |
| fungi | pleia2: as that patch stands now, you have a dupe declaration of ::gerrit | 00:51 |
| pleia2 | yeah | 00:51 |
| fungi | Duplicate declaration: Class[Gerrit] is already declared in file /opt/config/fungi/modules/openstack_project/manifests/gerrit.pp at line 153; cannot redeclare at /opt/config/fungi/modules/openstack_project/manifests/review_dev.pp:83 | 00:51 |
| * pleia2 thinks | 00:52 | |
| fungi | just treat it like any of the other variables you see passed into openstack_project::gerrit within review{,_dev}.pp | 00:55 |
| fungi | and then set a default of [] for it in modules/openstack_project/manifests/gerrit.pp like you do in modules/gerrit/init.pp | 00:56 |
| pleia2 | right right, thanks | 00:57 |
| *** avtar has quit IRC | 01:03 | |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Add replication of git from gerrit to git.o.o https://review.openstack.org/37794 | 01:04 |
| fungi | closer, but... Invalid parameter replication at /opt/config/fungi/modules/openstack_project/manifests/review_dev.pp:80 on node review-dev.openstack.org | 01:05 |
| pleia2 | aaaah | 01:06 |
| pleia2 | ok, I got this one | 01:06 |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Add replication of git from gerrit to git.o.o https://review.openstack.org/37794 | 01:08 |
| *** Ryan_Lane has joined #openstack-infra | 01:08 | |
| fungi | check your commas | 01:10 |
| fungi | oh, though you also added that default empty [] to the wrong file | 01:10 |
| fungi | let me mark up the review and show you what i'm suggesting | 01:11 |
| pleia2 | sneaky commas | 01:11 |
| pleia2 | heh, thanks | 01:11 |
| *** rcleere has joined #openstack-infra | 01:12 | |
| *** arosen1 has joined #openstack-infra | 01:12 | |
| *** arosen2 has quit IRC | 01:13 | |
| fungi | see if those inline comments make sense | 01:14 |
| fungi | i *think* that's all you'll need | 01:15 |
| *** svarnau has quit IRC | 01:16 | |
| *** zaro0508 has joined #openstack-infra | 01:16 | |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Add replication of git from gerrit to git.o.o https://review.openstack.org/37794 | 01:17 |
| fungi | pleia2: no errors! also the template looks like it's still generating the intended replication config so i think we're good | 01:20 |
| fungi | once jenkins comes back in favor, i'll cram this through since we've now addressed everyone's concerns | 01:20 |
| pleia2 | phew | 01:20 |
| harlowja | anyone got a sec, unsure whats happening with jenkins here :-/ | 01:20 |
| pleia2 | fungi: thanks for your patience, much appreciated | 01:20 |
| fungi | harlowja: got a link> | 01:20 |
| fungi | ? | 01:20 |
| fungi | pleia2: any time! | 01:20 |
| *** xchu has joined #openstack-infra | 01:21 | |
| harlowja | http://logs.openstack.org/62/29862/63/check/gate-cinder-python26/0479b6a/console.html unsure if u guys might have an idea here | 01:21 |
| harlowja | has this been a common problem for others so far :-/ | 01:21 |
| harlowja | ah, nm, i think i might know | 01:21 |
| fungi | ImportError: cannot import name utils | 01:22 |
| harlowja | nope, nm, don't know | 01:22 |
| fungi | can't say i've seen that one before | 01:22 |
| harlowja | k, hmmm, the odd part is if i run the tests locally, all good | 01:22 |
| harlowja | i think the import error is hiding some other problem though | 01:22 |
| fungi | looking closer | 01:23 |
| harlowja | like the tests run but then if i run from python interpreter directly | 01:23 |
| harlowja | >>> from cinder.volume import utils as volume_utils | 01:23 |
| harlowja | Traceback (most recent call last): | 01:23 |
| harlowja | File "<stdin>", line 1, in <module> | 01:23 |
| harlowja | File "cinder/volume/__init__.py", line 23, in <module> | 01:23 |
| harlowja | from cinder.common import config | 01:23 |
| harlowja | File "cinder/common/config.py", line 126, in <module> | 01:23 |
| harlowja | help=_("Deploy v1 of the Cinder API. ")), | 01:23 |
| harlowja | NameError: name '_' is not defined | 01:23 |
| harlowja | so i'm wondering if that is happening when it does the import, but the import error is getting rethrown | 01:23 |
| harlowja | if i use run_tests.sh though all fine | 01:23 |
| fungi | harlowja: looks like it's happening on 2.6 and 2.7 | 01:24 |
| harlowja | ya, sorta weird | 01:25 |
| fungi | harlowja: are you able to run any one of those failing tests in isolation? | 01:25 |
| harlowja | attempting | 01:25 |
| fungi | could be something else is importing utils in an earlier test when you run it from run_tests but the ordering changes when the ci systems run it under tox/testr | 01:26 |
| *** woodspa_ has joined #openstack-infra | 01:26 | |
| harlowja | hmmm | 01:27 |
| harlowja | let me see what i can find out | 01:27 |
| * harlowja begins adding statements in various places, ha | 01:27 | |
| *** woodspa_ has quit IRC | 01:27 | |
| fungi | though my conjecture is unlikely, given it's failing on 'from cinder.volume import utils as volume_utils' | 01:28 |
| harlowja | ya, something that hasn't really changed | 01:29 |
| *** nijaba has quit IRC | 01:29 | |
| fungi | stale .tox virtualenvs maybe? | 01:30 |
| *** nijaba has joined #openstack-infra | 01:30 | |
| *** woodspa has quit IRC | 01:30 | |
| harlowja | recreated a brand new vm just for this test, so don't think so :) | 01:30 |
| fungi | yeah, seems unlikely then | 01:30 |
| harlowja | ya, i'll see if i can figure it out, gotta be something really weird, ha | 01:31 |
| fungi | it's almost like utils is there but lacks an _init_.py or something? | 01:32 |
| *** nati_ueno has quit IRC | 01:34 | |
| harlowja | unsure :-/ | 01:35 |
| harlowja | seems like it has one | 01:35 |
| harlowja | the import error isn't so helpful, i don't think thats the real issue somehow | 01:36 |
| *** Ryan_Lane has quit IRC | 01:41 | |
| harlowja | i have a feeling it might be connected to https://github.com/openstack/cinder/commit/fa479f7b26 | 01:41 |
| harlowja | but not sure yet | 01:41 |
| fungi | that's well beyond my current level of understanding of cinder internals, unfortunately | 01:46 |
| harlowja | np, i'll figure it out :) | 01:46 |
| harlowja | thx fungi | 01:47 |
| fungi | harlowja: got a little further by adding 'from cinder.openstack.common.gettextutils import _' to cinder/common/__init__.py but that's probably not the real solution | 01:57 |
| harlowja | ya, i think i've found out parts of it | 01:57 |
| harlowja | https://bugs.launchpad.net/oslo/+bug/1208734 and i think it was being used by some of this new code, which then caused the import to fail like that | 01:57 |
| uvirtbot | Launchpad bug 1208734 in oslo "Drop openstack.common.exception" [Undecided,In progress] | 01:57 |
| harlowja | although the import failure isn't related to that bug, i think its a side-effect | 01:58 |
| bodepd | I had a patch to gerrit rejected,and I don't understand why: https://gist.github.com/bodepd/6180807 | 01:58 |
| bodepd | any clues would be appreciated :) | 01:58 |
| *** thomasm has joined #openstack-infra | 01:59 | |
| bodepd | I guess it's telling me that I need to merge them separetely, can I override this? Is just a push straight to the code repo without a patch the best way to proceed? | 02:00 |
| clarkb | bodepd: don't circumvent gerrit | 02:01 |
| clarkb | it is claiming there are no new changes | 02:01 |
| bodepd | clarkb: and I can push something with more than 1 commit? | 02:02 |
| clarkb | bodepd: yes, pushing more than one commit will create a change per commit | 02:03 |
| bodepd | clarkb: what if I just need to sync up some branches? | 02:04 |
| clarkb | oh you are pushing merge changes and stuff | 02:04 |
| *** yaguang has joined #openstack-infra | 02:05 | |
| clarkb | yeah so the typical backport procedure we use is cherry pick commit onto stable branch and propose for review | 02:05 |
| clarkb | because you don't get to circumvent review against $branch just because you merged to a different branch | 02:05 |
| clarkb | we have occasionally setup feature branches and made the acls allow merge commits so that you can get away with a single change for a branch of development | 02:05 |
| clarkb | bodepd: in this case the easiest thing to do is probably to propose each of the non merge commit changes to stable/grizzly | 02:06 |
| bodepd | clarkb: I kind of screwed up my branches (cut the branches too early) | 02:06 |
| clarkb | I think the problem is you are trying to push a bunch merge commits | 02:06 |
| bodepd | and now I have 100s of commits | 02:06 |
| mgagne | clarkb: if the branch only needs the be fast-forward, what do you suggest? Can it be done in some way or can the branch be deleted and recreated? | 02:06 |
| clarkb | mgagne: delete and recreate if it only needs to be fast forwarded would be one quick way | 02:07 |
| fungi | mgagne: bodepd: a ff can be simulated by deleting the branch and creating a new | 02:07 |
| fungi | what clarkb said | 02:07 |
| bodepd | clarkb: thanks | 02:07 |
| mgagne | and now we cannot delete branches by ourselves =) | 02:07 |
| fungi | however i'm happy to delete it for you in just a moment | 02:08 |
| mgagne | bodepd: whatever we do, we should make sure it's really a fast-forward and nothing else | 02:08 |
| clarkb | https://github.com/openstack-infra/config/blob/master/modules/openstack_project/files/gerrit/acls/openstack/swift.config is an alternative if you want to allow pushes of merge commits (in this case I think the branch move is more appropriate) | 02:08 |
| fungi | gerrit's best and worst feature is that it forces you into one very specific workflow | 02:09 |
| bodepd | what about github? | 02:12 |
| bodepd | do I need to delete the branches there? | 02:12 |
| mgagne | bodepd: it's synced by gerrit | 02:12 |
| bodepd | so, | 02:12 |
| fungi | gerrit will delete the branch from github and add the new one | 02:13 |
| *** zul has quit IRC | 02:14 | |
| bodepd | could we get help with delete our stable/grizzly branhces? | 02:19 |
| fungi | bodepd: absolutely | 02:19 |
| fungi | let me know which projects | 02:19 |
| *** UtahDave has joined #openstack-infra | 02:19 | |
| bodepd | ssh://bodepd@review.openstack.org:29418/stackforge/puppet-nova.git | 02:19 |
| bodepd | all on stackforge: | 02:19 |
| harlowja | so ya, fungi was the removal of a class which this new code depended on, somehow that exception was the one that ended up geting thrown, more of a side-effect of the problem | 02:20 |
| bodepd | puppet-nova, puppet-keystone, puppet-glance, puppet-swift, puppet-cinder, puppet-horizon, puppet-quantum, puppet-openstack | 02:20 |
| fungi | harlowja: okay, that *does* make sense | 02:20 |
| harlowja | sorta :-P | 02:20 |
| harlowja | not really the real error, which sorta is sad, ha | 02:21 |
| fungi | bodepd: so just delete the "stable/grizzly" branch from each of those? | 02:21 |
| *** sarob has joined #openstack-infra | 02:21 | |
| bodepd | yes. please. | 02:21 |
| bodepd | and I assume I can push a new branch to them? | 02:21 |
| fungi | bodepd: do you need new branches created and if so do you have a list of shas you want to branch from? | 02:21 |
| bodepd | I need to create a new branch based on master | 02:21 |
| *** emagana has quit IRC | 02:21 | |
| bodepd | for all of those | 02:21 |
| fungi | oh, you may be able to push new branches once i've deleted them, but i can just branch them each from the tip of master while i'm in there if you prefer | 02:22 |
| bodepd | mgagne: and then we can just submit a patch for those release commits when they are merged | 02:22 |
| bodepd | branching them from the tip of master woudl be perfect and appreciated | 02:22 |
| fungi | bodepd: mgagne: i did stackforge/puppet-nova | 02:23 |
| fungi | take a look and make sure it's sane before i move on to the others | 02:23 |
| *** Ryan_Lane has joined #openstack-infra | 02:24 | |
| *** erfanian has joined #openstack-infra | 02:24 | |
| *** melwitt has quit IRC | 02:26 | |
| pleia2 | fungi: thanks | 02:26 |
| bodepd | fungi: I can confirm that this looks correct. | 02:27 |
| * pleia2 gets flight delay notice: Your departure is now 2:05 AM | 02:27 | |
| bodepd | mgagne: can we merge those patches before the next ones? | 02:27 |
| fungi | bodepd: great. doing the other 7 on your list now | 02:27 |
| pleia2 | yeah, I definitely won't be around much tomorrow | 02:27 |
| fungi | pleia2: wow | 02:27 |
| mgagne | bodepd: sorry? | 02:27 |
| fungi | pleia2: get a nap | 02:27 |
| pleia2 | fungi: heh, it was already a late flight, just worse now :) | 02:27 |
| bodepd | mgagne: I want to just go ahead and merge everything else | 02:28 |
| bodepd | mgagne: any objection? | 02:28 |
| mgagne | bodepd: I can approve | 02:28 |
| mgagne | bodepd: all? | 02:28 |
| openstackgerrit | A change was merged to openstack-infra/config: Add replication of git from gerrit to git.o.o https://review.openstack.org/37794 | 02:28 |
| bodepd | mgagne: the release patches | 02:29 |
| mgagne | bodepd: ok | 02:29 |
| mgagne | bodepd: approving | 02:29 |
| *** nijaba has quit IRC | 02:29 | |
| *** nijaba has joined #openstack-infra | 02:30 | |
| *** nijaba has quit IRC | 02:30 | |
| *** nijaba has joined #openstack-infra | 02:30 | |
| bodepd | fungi: if you can hold off for a minute, it make my life slightly easier | 02:30 |
| bodepd | fungi: if you already started, not problem | 02:30 |
| bodepd | as always, thanks openstack-infra | 02:30 |
| mgagne | yep, always there to help people out =) | 02:31 |
| fungi | bodepd: oops, sorry | 02:32 |
| fungi | just finished them all | 02:32 |
| pleia2 | fungi: ah! I didn't put an ssh key in gerrit to git, so I fear it may hang upon first attempt to replicate | 02:34 |
| bodepd | fungi: no worries. I can sort out which ones still need updates. thanks | 02:35 |
| pleia2 | s/key/known_host | 02:35 |
| fungi | pleia2: i can manually fix it while you work up a patch. what's needed in it? | 02:35 |
| *** sarob has quit IRC | 02:35 | |
| *** sarob has joined #openstack-infra | 02:36 | |
| pleia2 | fungi: just .ssh/known_hosts file for whatever runs the git syncronization, with known_hosts populated with git.o.o's info | 02:36 |
| pleia2 | unless it overrides it somehow, I don't see a similar definition for github | 02:38 |
| fungi | i've stopped puppet on review.o.o for a bit and commented out the cgit section in the replication config | 02:38 |
| *** sarob has quit IRC | 02:40 | |
| pleia2 | fungi: does the gerrit2 user handle replication? | 02:42 |
| fungi | pleia2: it does, but i think it may use a keypair in ~gerrit2/review_site/etc/ | 02:43 |
| *** huangtianhua has joined #openstack-infra | 02:44 | |
| fungi | see the end of https://review.openstack.org/Documentation/config-replication.html | 02:45 |
| fungi | so ~gerrit2/.ssh/known_hosts i guess | 02:45 |
| *** huangtianhua has quit IRC | 02:45 | |
| jerryz | could anybody tell me why i can't see any jobs on jenkins.openstack.org dashboard? | 02:46 |
| pleia2 | fungi: thanks, that's what I was planning on | 02:46 |
| mgagne | jerryz: try jenkins01 and jenkins02 | 02:46 |
| *** zul has joined #openstack-infra | 02:47 | |
| fungi | jerryz: if you're looking for job status, http://status.openstack.org/zuul is a better overview these days | 02:47 |
| clarkb | we are trying to deemphasize jenkins itself | 02:47 |
| fungi | pleia2: so... it looks like we're not managing ~gerrit2/.ssh/known_hosts in puppet | 02:47 |
| jerryz | thanks | 02:48 |
| fungi | there are 8 keys listed in it currently | 02:48 |
| pleia2 | fungi: oh :) | 02:48 |
| pleia2 | fungi: maybe we just add it manually for now and file a bug to get it sorted? | 02:49 |
| pleia2 | everything should be set on the git.o.o side, I've added gerrit's ssh keys to it already | 02:50 |
| fungi | pleia2: i've cached it and can authenticate via 'ssh -i review_site/etc/ssh_host_rsa_key cgit@git.openstack.org' | 02:51 |
| pleia2 | great, we should be ok then :) | 02:51 |
| fungi | so assuming it's the ssh_host_rsa_key it's going to try to use, we should be set | 02:52 |
| fungi | however, that seems somewhat unlikely | 02:52 |
| fungi | it would be odd for it to use .ssh/known_hosts yet authenticate with review_site/etc/ssh_host_rsa_key instead of .ssh/id_rsa | 02:54 |
| fungi | i'll see if i can confirm what public key github is authorizing | 02:54 |
| *** erfanian has quit IRC | 02:59 | |
| fungi | pleia2: yeah, so we're authenticating to github with ~gerrit2/.ssh/id_rsa.pub | 03:02 |
| fungi | that was marginally painful to confirm | 03:02 |
| pleia2 | fungi: ok, installed the pubkey on git.o.o that looked to make the most sense from hiera | 03:03 |
| *** SergeyLukjanov has joined #openstack-infra | 03:03 | |
| pleia2 | 'gerrit_ssh_rsa_pubkey_contents' | 03:03 |
| pleia2 | yeah, so that's what is in review_site/etc/ssh_host_rsa_key - not a clue what ~gerrit2/.ssh/id_rsa.pub is (wouldn't it be nice if it were the same?) | 03:06 |
| fungi | well, it's not the same unfortunately because i couldn't authenticate to git.o.o with it | 03:07 |
| pleia2 | tsk | 03:08 |
| clarkb | the host rsa key is the host key | 03:08 |
| clarkb | so that when we switch gerrit servers everyone doesnt need to update their known hosts | 03:08 |
| clarkb | I think | 03:08 |
| pleia2 | oh | 03:08 |
| clarkb | pub key is the thing actually used to auth with | 03:08 |
| pleia2 | so is gerrit's id_rsa.pub in hiera? | 03:08 |
| clarkb | no clue and I am not on a machine I can check with | 03:09 |
| clarkb | my guess is yes | 03:09 |
| pleia2 | maybe ssh_project_rsa_pubkey_contents ? | 03:09 |
| fungi | pleia2: i find nothing in puppet managing ~gerrit2/.ssh/id_rsa.pub but i'm going to grep hiera for the contents anyway just to see | 03:10 |
| pleia2 | fungi: yeah, it's definitely not set up by puppet | 03:10 |
| fungi | pleia2: and it's definitely not stashed in hiera either | 03:11 |
| fungi | i'll add it in there as a first order of business | 03:12 |
| pleia2 | heh, ok :) | 03:12 |
| *** yaguang has quit IRC | 03:14 | |
| pleia2 | once that's in, I'll update the site.pp to include it on git.o.o so it can properly populate the authorized_keys file, then open a bug about gerrit's ssh stuff not being managed by puppet and we can tidy that up later | 03:14 |
| fungi | pleia2: using gerrit_replication_ssh_rsa_pubkey_contents | 03:15 |
| pleia2 | thank you! | 03:15 |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Put correct gerrit ssh pubkey on git.o.o https://review.openstack.org/40788 | 03:19 |
| fungi | i'll also open a bug suggesting we start managing ~gerrit2/.ssh contents in puppet | 03:19 |
| pleia2 | that's what I meant :) | 03:20 |
| *** HenryG has quit IRC | 03:22 | |
| fungi | or yes, please open the bug if you want. i totally didn't see you say that right there until i reread it just now | 03:22 |
| * pleia2 does so | 03:23 | |
| pleia2 | fungi: did you add secret key to hiera too? | 03:24 |
| fungi | pleia2: yep, i did | 03:24 |
| *** arosen1 has quit IRC | 03:24 | |
| pleia2 | assuming gerrit_replication_ssh_rsa_key_contents | 03:24 |
| fungi | yep! i did not add the known_hosts file contents but we'll likely want to populate it from a list of vars anyway | 03:25 |
| * pleia2 nods | 03:25 | |
| pleia2 | thanks | 03:25 |
| openstackgerrit | A change was merged to openstack-infra/config: Put correct gerrit ssh pubkey on git.o.o https://review.openstack.org/40788 | 03:25 |
| fungi | and clean out some of the cruft currently in there | 03:25 |
| fungi | since there were already 8 different hosts in it before i added git.o.o | 03:25 |
| fungi | and i'm pretty sure no more than a couple were currently used | 03:26 |
| *** yaguang has joined #openstack-infra | 03:26 | |
| pleia2 | no doubt :) | 03:26 |
| fungi | likely just the entry for github and the entry for connecting to its own sshd | 03:26 |
| *** turul_ has joined #openstack-infra | 03:27 | |
| *** turul_ is now known as afazekas | 03:28 | |
| *** nijaba has quit IRC | 03:29 | |
| fungi | in a few minutes once git.o.o picks up that change, i'll retest from review.o.o with the correct id and then turn puppet back on there if it works | 03:29 |
| pleia2 | git.o.o is a bit wonky at the moment from what https://review.openstack.org/#/c/40770/ seeks to fix, not sure if that will prevent this change from going | 03:30 |
| *** nijaba has joined #openstack-infra | 03:30 | |
| pleia2 | sorry, this all went more rabbit-holey than I expected :) | 03:30 |
| fungi | everything does | 03:31 |
| fungi | looking | 03:31 |
| *** sandywalsh has quit IRC | 03:32 | |
| *** zul has quit IRC | 03:33 | |
| openstackgerrit | A change was merged to openstack-infra/config: Fix double-declaration of cgit class https://review.openstack.org/40770 | 03:34 |
| *** Ryan_Lane has quit IRC | 03:38 | |
| *** Ryan_Lane has joined #openstack-infra | 03:39 | |
| *** sarob has joined #openstack-infra | 03:46 | |
| lifeless | ok so run-mirror finally beaten into shape, 12m to mirror into HP cloud | 03:47 |
| lifeless | mordred: would you accept a bindep using patch for pbr's tests? | 03:48 |
| pleia2 | fungi: well, git.o.o is not all well (i'll dig into it and fix later), but the ssh key is updated anyway so you should be able to test that anyway | 03:49 |
| pleia2 | (should fix itself on next puppet run I think) | 03:49 |
| mordred | lifeless: depends | 03:50 |
| fungi | yep, that works ow | 03:50 |
| fungi | now | 03:50 |
| mordred | pbr is in the gate | 03:50 |
| pleia2 | fungi: hooray! | 03:50 |
| lifeless | mordred: every change to nova does a pbr test run? or every change to nova uses pbr ? | 03:50 |
| fungi | puppet restarted on review.o.o now | 03:50 |
| pleia2 | thanks | 03:50 |
| *** sarob has quit IRC | 03:51 | |
| *** cody-somerville has quit IRC | 03:55 | |
| lifeless | mordred: ... ? | 03:55 |
| lifeless | jeblair: for context, a mirror refresh in hpcloud takes 9 minutes | 03:57 |
| lifeless | jeblair: thats about 6 minutes longer than we want image building to be ;) | 03:57 |
| lifeless | jeblair: but we need to do it inline with the builds because we get latest git which may require latest mirror contents. | 03:57 |
| lifeless | jeblair: [I thought some hard data might be useful for you]. I'm going to time doing it from my machine too. | 03:58 |
| clarkb | lifeless: run-mirror is probably iterating over all of the branches | 03:58 |
| mordred | lifeless: yes | 03:58 |
| clarkb | if we make that configurable you can probably make it go much quicker over just master | 03:58 |
| fungi | i think -b master might help | 03:59 |
| lifeless | clarkb: fungi: time run-mirror -b remotes/origin/master --verbose -c mirror.yaml | 03:59 |
| mordred | -b remotes/origin/master | 03:59 |
| lifeless | ... | 03:59 |
| lifeless | real 8m58.749s | 03:59 |
| lifeless | user 5m5.752s | 03:59 |
| lifeless | sys 0m36.284s | 03:59 |
| clarkb | does it already ahve the option? also you want to keep your pip cache around so that you only deal with the difs | 03:59 |
| * fungi added that option a while back | 03:59 | |
| mordred | lifeless: pbr gates nova and nova gates pbr | 03:59 |
| lifeless | clarkb: and that was a mere 3 hours after doing the initial run | 03:59 |
| lifeless | mordred: so? join the dots, I don't see the concern. | 04:00 |
| mordred | lifeless: no specific concern - just saying | 04:00 |
| lifeless | mordred: so I'm proposing to remove the 'apt-get-install' line in favour of something with bindep. | 04:00 |
| lifeless | mordred: something like apt-get install `bindep --missing` | 04:00 |
| lifeless | mordred: and then a other-requirements.txt file | 04:01 |
| clarkb | lifeless: hmm, at that point its just doing a pip install -r iirc | 04:01 |
| lifeless | clarkb: sure, I'm not attacking or defending it. Just - you know - data. | 04:01 |
| clarkb | ya, I am trying to figure out how to make it go faster because you gave us the data :) | 04:01 |
| lifeless | clarkb: :) | 04:01 |
| lifeless | Run: /tmp/tmpNkEUqt/venv/bin/pip install -U --exists-action=w --download-cache=/home/ubuntu/download/pip/openstack --build /tmp/tmpNkEUqt/build -r /tmp/tmpfepvdg | 04:02 |
| lifeless | seems to be a/the big time sink | 04:02 |
| clarkb | ya, I wonder if new pip is quicker? | 04:02 |
| fungi | it has flames and racing stripes | 04:02 |
| clarkb | lifeless: would be interesting to see how latest pip does if you are using an older version | 04:02 |
| lifeless | it's running pip from a venv | 04:03 |
| lifeless | and auto upgrades it itself. | 04:03 |
| lifeless | so nyar nyar | 04:03 |
| lifeless | Using download cache from /home/ubuntu/download/pip/openstack/https%3A%2F%2Fpypi.python.org%2Fpackages%2Fsource%2Fp%2Fpip%2Fpip-1.4.1.tar.gz | 04:03 |
| lifeless | -. 1.4.1 | 04:03 |
| lifeless | mordred: this idea came to me when I saw your comment in the project about getting the deps into the tree (right above the line in my patch [in gerrit] to fix mirroring python-ldap | 04:04 |
| dstufft | it'll go faster if you use --no-allow-external | 04:05 |
| dstufft | and turn on --allow-external Foo for only the things that need it | 04:05 |
| lifeless | it could be CPU bound for all I know | 04:05 |
| lifeless | reinstalling everything | 04:05 |
| dstufft | oh there's also lxml | 04:05 |
| dstufft | which takes a bunch of time | 04:05 |
| lifeless | it might e 'building numpy and pythonldap etc etc takes 5 minutes' | 04:05 |
| *** sandywalsh has joined #openstack-infra | 04:06 | |
| clarkb | lifeless: it should only build those the first time | 04:06 |
| clarkb | right? or is the cahce not a build cache | 04:06 |
| dstufft | it's not a build cache | 04:06 |
| dstufft | it's a download cache | 04:06 |
| lifeless | clarkb: ^ | 04:06 |
| clarkb | :/ | 04:06 |
| lifeless | clarkb: it has to reinstall them to detect changed transitive deps. | 04:07 |
| clarkb | can we have wheels yet? :) | 04:07 |
| fungi | we blow away the build cache because, if memory serves, easy_install will see it and reuse different/older versions than you request | 04:07 |
| dstufft | you can make a build cache | 04:08 |
| pleia2 | hrm, /etc/ssl/private/git.openstack.org.key isn't being created | 04:08 |
| dstufft | filling the cache is an explicit step though | 04:08 |
| dstufft | for right now | 04:08 |
| *** vogxn has joined #openstack-infra | 04:08 | |
| dstufft | once wheels are on by default there's a chance we'll make a global build cache | 04:08 |
| fungi | right. we tried using a build cache early on and it went way faster, but also didn't download new versions of things | 04:08 |
| lifeless | real 8m35.737s | 04:09 |
| lifeless | user 5m1.092s | 04:09 |
| lifeless | sys 0m35.444s | 04:09 |
| *** UtahDave has quit IRC | 04:09 | |
| lifeless | so yeah, 8m. Waaaay to slow. | 04:09 |
| dstufft | this is openstack/requirements/requirements.txt? | 04:09 |
| lifeless | [for inline updates] | 04:09 |
| clarkb | global-requirements.txt but yes | 04:09 |
| lifeless | dstufft: run-mirror -b remotes/origin/master --verbose -c mirror.yaml | 04:12 |
| lifeless | where mirror.yaml is something like | 04:12 |
| lifeless | cache-root: /home/ubuntu/download | 04:12 |
| lifeless | mirrors: | 04:12 |
| lifeless | - name: openstack | 04:12 |
| lifeless | projects: | 04:12 |
| lifeless | - https://github.com/openstack/requirements | 04:12 |
| lifeless | output: /home/ubuntu/mirror | 04:12 |
| *** jerryz has quit IRC | 04:13 | |
| lifeless | jog0: btw what did you mean by pointing at the git docs in my .gitignore patch? | 04:13 |
| fungi | pleia2: looks from the log like you need to create the /etc/ssl/private directory. that's probably somewhat of a debianism | 04:13 |
| pleia2 | fungi: ah, thanks | 04:14 |
| fungi | or else the equivalent of the ca-certificates package for centos isn't installed | 04:14 |
| * pleia2 nods | 04:14 | |
| fungi | regardless, the dir isn't there | 04:14 |
| * fungi sees how far that's going to go | 04:14 | |
| mordred | I want to be a partof this conversation | 04:15 |
| mordred | but I just went to the best caiprinha bar in São Paulo | 04:15 |
| lifeless | mordred: which? | 04:15 |
| mordred | so I will not | 04:15 |
| mordred | any of them | 04:15 |
| lifeless | mordred: which bar | 04:15 |
| mordred | www.velosobar.com.br | 04:15 |
| mordred | Veloso | 04:16 |
| mordred | also | 04:16 |
| clarkb | lifeless: jog0 is suggesting you set those rules locally in your global git config | 04:16 |
| mordred | there are fruits here that we don't have elsewhere | 04:16 |
| lifeless | clarkb: so we should delete the .gitconfig file? | 04:16 |
| mordred | lifeless: such as jabuticaba and caju | 04:16 |
| clarkb | no, keep it for project specific stuff like pyc files | 04:16 |
| fungi | pleia2: /etc/pki/tls/certs and /etc/pki/tls/private on centos | 04:16 |
| lifeless | mordred: looks nice; have fun. Why are you in brazil btw? | 04:16 |
| clarkb | mordred jabuticaba is amazing | 04:17 |
| lifeless | clarkb: I don't understand the difference. | 04:17 |
| fungi | pleia2: if you want to make it more technically correct | 04:17 |
| mordred | lifeless: because | 04:17 |
| clarkb | mordred my grandparents have three trees | 04:17 |
| pleia2 | fungi: should do, I'll put together a patch for later | 04:17 |
| clarkb | so so good and they look funny on the tree too | 04:17 |
| mordred | clarkb: where are your grandparents? | 04:17 |
| clarkb | mordred big island hawaii | 04:17 |
| pleia2 | fungi: how is gerrit? puppet dashboard is suspiciously quiet, want to wrap this up soon though so you can get some rest :) | 04:18 |
| mordred | lifeless: consegi next week - free software in government conference | 04:18 |
| mordred | lifeless: also, I have a lot of friends down here | 04:18 |
| mordred | clarkb: awesome | 04:18 |
| fungi | pleia2: i'll be awake for a while still. about to push up a patch and then confirm we haven't broken github replication | 04:18 |
| clarkb | they grow off the trunk rather than at the end of branches | 04:18 |
| lifeless | mordred: cool | 04:18 |
| mordred | lifeless: yes | 04:19 |
| * mordred will now go to sleep | 04:19 | |
| clarkb | lifeless your editor files are a you thing | 04:19 |
| lifeless | mordred: ciao | 04:19 |
| mordred | clarkb: I am jealous of your grandparents | 04:19 |
| clarkb | python files are a project thing | 04:19 |
| mordred | clarkb: tell them to grow some caju | 04:19 |
| mordred | it's AMAZING | 04:19 |
| lifeless | clarkb: editor backups are a standard understood by lots of tools; in what way are they a 'me' thing? | 04:19 |
| clarkb | I am not familiar with caju /me googles | 04:19 |
| clarkb | lifeless its your editor (I personally wouldn't -1) but I think that is the argument | 04:20 |
| lifeless | clarkb: I think I should have this discussion with jog0 then :) | 04:20 |
| mordred | me just learned thigns | 04:20 |
| mordred | caju == cashew truee | 04:20 |
| mordred | tree | 04:20 |
| *** Ryan_Lane has quit IRC | 04:21 | |
| lifeless | clarkb: as if you don't endorse it, me convincing you won't undo his -1 :) | 04:21 |
| mordred | in addition to the NUT | 04:21 |
| mordred | you can also eat the cashew APPLE | 04:21 |
| mordred | which they make drinks from down here | 04:21 |
| lifeless | mordred: the pronounciation really should have given it away:) | 04:21 |
| clarkb | cacao before it is rotted to be chocolate is aweaome | 04:21 |
| clarkb | caju sounds like that but for cashews | 04:21 |
| lifeless | and yes, caju is most excellent | 04:21 |
| mordred | SO GOOD | 04:22 |
| mordred | when you turn it in to caipirinhas | 04:22 |
| lifeless | well | 04:24 |
| lifeless | anythinh caipirinhas is good | 04:24 |
| lifeless | sugar + ethanol + ice + fruit | 04:24 |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Fix ssl location for git.o.o on CentOS https://review.openstack.org/40792 | 04:24 |
| pleia2 | I'll get this centos thing eventually | 04:24 |
| lifeless | no, you won't :) | 04:25 |
| pleia2 | haha | 04:26 |
| lifeless | it, like rhel, will get you | 04:26 |
| fungi | it will get you first | 04:26 |
| fungi | ha | 04:26 |
| pleia2 | nice | 04:26 |
| fungi | lifeless beat me to the punchline there | 04:26 |
| lifeless | when it's supported for 10 years, with no change, ease of use is a rare thing, for $various reasons | 04:26 |
| openstackgerrit | A change was merged to openstack-infra/config: Fix ssl location for git.o.o on CentOS https://review.openstack.org/40792 | 04:29 |
| *** Ryan_Lane has joined #openstack-infra | 04:30 | |
| *** nijaba has quit IRC | 04:30 | |
| *** nijaba has joined #openstack-infra | 04:30 | |
| *** Ryan_Lane has quit IRC | 04:31 | |
| *** Ryan_Lane1 has joined #openstack-infra | 04:31 | |
| *** afazekas has quit IRC | 04:31 | |
| *** SergeyLukjanov has quit IRC | 04:33 | |
| *** rcleere has quit IRC | 04:37 | |
| *** rcleere has joined #openstack-infra | 04:39 | |
| *** nicedice_ has quit IRC | 04:40 | |
| *** Ryan_Lane has joined #openstack-infra | 04:43 | |
| *** boris-42 has joined #openstack-infra | 04:43 | |
| dstufft | lifeless: I was bored https://gist.github.com/dstufft/08cfaf16539a56bc5eb2 | 04:44 |
| dstufft | that's download time only | 04:44 |
| lifeless | cool | 04:44 |
| lifeless | dstufft: where's your -U :) | 04:45 |
| dstufft | lifeless: it's an empty venv there's nothing to upgrade | 04:45 |
| *** Ryan_Lane1 has quit IRC | 04:45 | |
| *** nayward has joined #openstack-infra | 04:49 | |
| fungi | pleia2: that worked! | 04:50 |
| pleia2 | fungi: almost, apache didn't start | 04:51 |
| fungi | yeah | 04:51 |
| *** afazekas has joined #openstack-infra | 04:52 | |
| fungi | trying it manually to get more details | 04:52 |
| pleia2 | placing bets on whether it's selinux | 04:54 |
| pleia2 | :) | 04:54 |
| fungi | No such file or directory: httpd: could not open error log file /etc/httpd/${APACHE_LOG_DIR}/git-error.log. | 04:54 |
| pleia2 | well that's no so bad | 04:54 |
| pleia2 | ErrorLog ${APACHE_LOG_DIR}/git-error.log | 04:57 |
| pleia2 | I wonder why it expanded so weirdly | 04:57 |
| pleia2 | or didn't expand, I guess | 04:58 |
| lifeless | eh | 05:04 |
| lifeless | robertc@lifelesshp:~$ du -sh work/jeepyb-test/mirror/ | 05:04 |
| lifeless | 25M work/jeepyb-test/mirror/ | 05:04 |
| lifeless | robertc@lifelesshp:~$ du -sh work/jeepyb-test/download/ | 05:04 |
| lifeless | 56M work/jeepyb-test/download/ | 05:04 |
| lifeless | after 20m | 05:04 |
| lifeless | This is what I mean by 'mirroring slow from NZ' | 05:04 |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/git-review: Pin hacking <0.6 https://review.openstack.org/40799 | 05:05 |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/git-review: Wrap multiple-commit warning text https://review.openstack.org/40800 | 05:05 |
| *** rcleere has quit IRC | 05:07 | |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Update httpd log directory in template for git.o.o https://review.openstack.org/40801 | 05:15 |
| openstackgerrit | A change was merged to openstack-infra/config: Update httpd log directory in template for git.o.o https://review.openstack.org/40801 | 05:20 |
| *** nayward has quit IRC | 05:25 | |
| *** nayward has joined #openstack-infra | 05:25 | |
| *** nijaba has quit IRC | 05:30 | |
| *** nijaba has joined #openstack-infra | 05:31 | |
| *** nijaba has quit IRC | 05:31 | |
| *** nijaba has joined #openstack-infra | 05:31 | |
| *** dina_belova has joined #openstack-infra | 05:36 | |
| *** dina_belova has quit IRC | 05:38 | |
| pleia2 | fungi: git.o.o is ok, but I think when that VM was set up port 443 wasn't opened | 05:39 |
| fungi | pleia2: yay | 05:39 |
| fungi | and yeah, iptables in the puppet manifest | 05:39 |
| pleia2 | I updated puppet | 05:40 |
| pleia2 | added 443 in the list of ports in my initial ssl patch, but the VM itself (at least on hpcloud, I need to give it a list of ports I want open) | 05:40 |
| fungi | pleia2: looks like it's there, yeag | 05:40 |
| fungi | is this on hpcloud and not rackspace? | 05:41 |
| pleia2 | it's rackspace | 05:41 |
| pleia2 | not sure how those VMs work with ports | 05:41 |
| fungi | shouldn't be needed... tcpdump to the rescue | 05:41 |
| fungi | it's getting there and the server's sending tcp/rst | 05:42 |
| pleia2 | huh | 05:43 |
| lifeless | mordred: can you use a run-mirror mirror without the html files? like will just the raw paths do? | 05:43 |
| fungi | yeah, netstat -lnt says nothing's listening on 443 | 05:43 |
| lifeless | mordred: (with a file:// url to it) | 05:43 |
| lifeless | fungi: I so want a -i in there | 05:44 |
| fungi | lifeless: indeed | 05:44 |
| *** jhesketh has quit IRC | 05:46 | |
| fungi | pleia2: no errors in the log, so my guess is IfModule mod_ssl.c evaluates false | 05:46 |
| pleia2 | fungi: ah, I need to finish packing and do last minute close up home things before trip, I'll see about whipping up a patch tomorrow | 05:47 |
| fungi | pleia2: no worries. it's very close now! | 05:48 |
| fungi | have a safe trip | 05:49 |
| pleia2 | thanks :) | 05:49 |
| *** mikal has quit IRC | 05:54 | |
| *** mikal has joined #openstack-infra | 05:55 | |
| lifeless | just saw this in g+ - http://hynek.me/articles/hardening-your-web-servers-ssl-ciphers/ | 05:57 |
| lifeless | might be worth an eyeball from one of you sysadmin types ;) | 05:57 |
| *** jhesketh has joined #openstack-infra | 05:59 | |
| lifeless | clarkb: fungi & pleia2 ^ | 05:59 |
| *** sshturm__ has quit IRC | 06:00 | |
| *** Ryan_Lane has quit IRC | 06:06 | |
| lifeless | and wheee - | 06:06 |
| lifeless | real 39m4.291s | 06:06 |
| lifeless | user 6m33.628s | 06:06 |
| lifeless | sys 0m46.968s | 06:06 |
| lifeless | thats a update on a fully mirrored cache, from NZ using run-mirror -b remotes/origin/master | 06:06 |
| lifeless | mordred: clarkb: fungi: jeblair: ^ :) Again, just FYI. | 06:07 |
| lifeless | dstufft: I haven't tried your tweaked link list yet. | 06:07 |
| lifeless | later.. | 06:07 |
| *** Ryan_Lane has joined #openstack-infra | 06:10 | |
| *** thomasbiege1 has joined #openstack-infra | 06:15 | |
| *** markmcclain has quit IRC | 06:19 | |
| *** nijaba has quit IRC | 06:30 | |
| *** nijaba has joined #openstack-infra | 06:31 | |
| *** pabelanger has quit IRC | 06:50 | |
| *** vogxn has quit IRC | 06:58 | |
| *** thomasbiege1 has quit IRC | 07:05 | |
| *** pabelanger has joined #openstack-infra | 07:13 | |
| *** mestery_ has joined #openstack-infra | 07:17 | |
| *** mestery has quit IRC | 07:19 | |
| *** jpich has joined #openstack-infra | 07:21 | |
| *** SergeyLukjanov has joined #openstack-infra | 07:21 | |
| *** dina_belova has joined #openstack-infra | 07:23 | |
| *** Ryan_Lane has quit IRC | 07:23 | |
| ttx | mgagne: (late answer): you can list bugs and blueprints at the project group level, see for example https://bugs.launchpad.net/openstack/+bugs | 07:25 |
| *** vogxn has joined #openstack-infra | 07:26 | |
| ttx | the problem with LP project group is that a single project can belong to only one project group... and any project can add itself to a group | 07:26 |
| *** nijaba has quit IRC | 07:31 | |
| *** Ryan_Lane has joined #openstack-infra | 07:31 | |
| *** nijaba has joined #openstack-infra | 07:32 | |
| *** CliMz has joined #openstack-infra | 07:36 | |
| *** vogxn has quit IRC | 07:43 | |
| *** emagana has joined #openstack-infra | 07:44 | |
| *** thomasbiege1 has joined #openstack-infra | 07:54 | |
| *** dina_belova has quit IRC | 07:55 | |
| *** yaguang has quit IRC | 08:03 | |
| *** derekh has joined #openstack-infra | 08:07 | |
| *** thomasbiege1 has quit IRC | 08:12 | |
| *** dina_belova has joined #openstack-infra | 08:14 | |
| *** odyssey4me has joined #openstack-infra | 08:14 | |
| *** vogxn has joined #openstack-infra | 08:18 | |
| *** fbo_away is now known as fbo | 08:22 | |
| *** vogxn has quit IRC | 08:22 | |
| *** vogxn has joined #openstack-infra | 08:23 | |
| *** boris-42 has quit IRC | 08:23 | |
| *** yaguang has joined #openstack-infra | 08:28 | |
| *** nijaba has quit IRC | 08:31 | |
| *** psedlak has joined #openstack-infra | 08:31 | |
| *** nijaba has joined #openstack-infra | 08:32 | |
| *** xchu has quit IRC | 08:32 | |
| *** xchu has joined #openstack-infra | 08:44 | |
| *** ruhe has joined #openstack-infra | 08:45 | |
| *** Ryan_Lane has quit IRC | 08:49 | |
| *** dkehn has quit IRC | 09:01 | |
| *** thomasbiege1 has joined #openstack-infra | 09:07 | |
| *** yaguang has quit IRC | 09:13 | |
| *** yaguang has joined #openstack-infra | 09:14 | |
| *** xchu has quit IRC | 09:19 | |
| *** Ryan_Lane has joined #openstack-infra | 09:21 | |
| *** yaguang has quit IRC | 09:22 | |
| *** yaguang has joined #openstack-infra | 09:23 | |
| *** dina_belova has quit IRC | 09:27 | |
| *** dina_belova has joined #openstack-infra | 09:27 | |
| *** dina_belova has quit IRC | 09:28 | |
| *** nijaba has quit IRC | 09:31 | |
| *** xchu has joined #openstack-infra | 09:32 | |
| *** nijaba has joined #openstack-infra | 09:32 | |
| *** nijaba has quit IRC | 09:32 | |
| *** nijaba has joined #openstack-infra | 09:32 | |
| ekarlso- | hmmm | 09:34 |
| ekarlso- | mordred: ping | 09:34 |
| ekarlso- | or anyone else rly | 09:34 |
| ekarlso- | getting a error when doing devstack on precise | 09:35 |
| ekarlso- | http://pastebin.com/ZeLWUq2c | 09:35 |
| *** spiffxp has quit IRC | 09:35 | |
| *** Ryan_Lane has quit IRC | 09:36 | |
| *** spiffxp has joined #openstack-infra | 09:37 | |
| *** yaguang_ has joined #openstack-infra | 09:38 | |
| *** yaguang has quit IRC | 09:38 | |
| *** dkehn has joined #openstack-infra | 09:40 | |
| *** boris-42 has joined #openstack-infra | 09:40 | |
| *** ruhe has quit IRC | 09:50 | |
| *** Mikhail has joined #openstack-infra | 09:56 | |
| *** mdurnosvistov has quit IRC | 09:56 | |
| *** Mikhail is now known as Guest65861 | 09:56 | |
| *** koobs has quit IRC | 10:08 | |
| *** koobs has joined #openstack-infra | 10:08 | |
| *** dina_belova has joined #openstack-infra | 10:10 | |
| *** xchu has quit IRC | 10:11 | |
| *** amotoki has quit IRC | 10:12 | |
| *** emagana_ has joined #openstack-infra | 10:14 | |
| *** mikal has quit IRC | 10:17 | |
| *** emagana has quit IRC | 10:18 | |
| *** mikal has joined #openstack-infra | 10:19 | |
| *** dina_belova has quit IRC | 10:27 | |
| *** dina_belova has joined #openstack-infra | 10:28 | |
| *** nijaba has quit IRC | 10:28 | |
| *** nijaba has joined #openstack-infra | 10:32 | |
| *** nijaba has quit IRC | 10:32 | |
| *** nijaba has joined #openstack-infra | 10:32 | |
| lifeless | argh stab ' | 10:35 |
| lifeless | 2013-08-07 22:40:31.032 | ./setup.py:1:1: H803 git commit title ('Fix python-ldap mirroring.') should not end with period' | 10:35 |
| openstackgerrit | lifeless proposed a change to openstack-dev/pbr: Stop checking periods in commit messages https://review.openstack.org/40838 | 10:37 |
| openstackgerrit | lifeless proposed a change to openstack-dev/pbr: Fix python-ldap mirroring. https://review.openstack.org/40732 | 10:37 |
| *** thomasbiege1 has quit IRC | 10:38 | |
| *** zul has joined #openstack-infra | 10:50 | |
| *** weshay has joined #openstack-infra | 10:57 | |
| *** ruhe has joined #openstack-infra | 10:58 | |
| *** yaguang_ has quit IRC | 10:59 | |
| *** yolanda has joined #openstack-infra | 11:05 | |
| yolanda | hi, i'm trying to setup a generic zuul deployment, and i'm having problems if i don't specify any project. How can i set an empty projects section into layout.yaml? | 11:05 |
| *** ruhe has quit IRC | 11:08 | |
| *** nijaba has quit IRC | 11:09 | |
| *** ruhe has joined #openstack-infra | 11:10 | |
| *** CaptTofu has quit IRC | 11:13 | |
| *** CaptTofu has joined #openstack-infra | 11:14 | |
| *** nijaba has joined #openstack-infra | 11:16 | |
| *** nijaba has quit IRC | 11:16 | |
| *** nijaba has joined #openstack-infra | 11:16 | |
| *** CaptTofu has quit IRC | 11:19 | |
| *** CaptTofu has joined #openstack-infra | 11:19 | |
| yolanda | any idea? i tried something like projects: ~ but it complains with MultipleInvalid: expected a list for dictionary value @ data['projects'] | 11:20 |
| *** dkehn has quit IRC | 11:20 | |
| *** dkehn has joined #openstack-infra | 11:24 | |
| *** woodspa has joined #openstack-infra | 11:25 | |
| *** mikal has quit IRC | 11:28 | |
| *** dina_belova has quit IRC | 11:28 | |
| *** mikal has joined #openstack-infra | 11:30 | |
| *** lcestari has joined #openstack-infra | 11:32 | |
| *** dina_belova has joined #openstack-infra | 11:34 | |
| *** HenryG has joined #openstack-infra | 11:35 | |
| *** emagana_ has quit IRC | 11:36 | |
| *** vogxn has quit IRC | 11:40 | |
| *** thomasbiege1 has joined #openstack-infra | 11:41 | |
| *** rfolco has joined #openstack-infra | 11:44 | |
| *** ruhe has quit IRC | 11:45 | |
| *** dina_belova has quit IRC | 11:49 | |
| *** ArxCruz has joined #openstack-infra | 11:49 | |
| *** ruhe has joined #openstack-infra | 11:51 | |
| *** weshay has quit IRC | 12:03 | |
| *** ruhe has quit IRC | 12:06 | |
| openstackgerrit | Chuck Short proposed a change to openstack-infra/config: python3: Add python3 python-novaclient gate https://review.openstack.org/40848 | 12:07 |
| *** mriedem has joined #openstack-infra | 12:08 | |
| *** dina_belova has joined #openstack-infra | 12:11 | |
| *** chuckieb has joined #openstack-infra | 12:16 | |
| *** dina_belova has quit IRC | 12:18 | |
| *** dprince has joined #openstack-infra | 12:21 | |
| *** sandywalsh has quit IRC | 12:22 | |
| *** mikal has quit IRC | 12:22 | |
| *** mikal has joined #openstack-infra | 12:24 | |
| HenryG | So I pulled the latest devstack, and now when I run stack.sh I get "ImportError: No module named pip" | 12:33 |
| HenryG | Seems to be due to https://github.com/openstack-dev/devstack/commit/408a4a7d1c24322b35f9a8617c7c62adeeee0dbe | 12:34 |
| *** SergeyLukjanov has quit IRC | 12:34 | |
| HenryG | Because if I rewind to before that it works. | 12:35 |
| openstackgerrit | Chuck Short proposed a change to openstack-infra/config: python3: Add non-voting gates for clients https://review.openstack.org/40851 | 12:35 |
| *** sandywalsh has joined #openstack-infra | 12:35 | |
| *** emagana has joined #openstack-infra | 12:37 | |
| *** ruhe has joined #openstack-infra | 12:42 | |
| *** SergeyLukjanov has joined #openstack-infra | 12:43 | |
| *** emagana has quit IRC | 12:45 | |
| *** weshay has joined #openstack-infra | 12:45 | |
| *** dina_belova has joined #openstack-infra | 12:48 | |
| *** ruhe has quit IRC | 12:49 | |
| *** anteaya has joined #openstack-infra | 12:57 | |
| *** ruhe has joined #openstack-infra | 12:59 | |
| *** ruhe has quit IRC | 13:00 | |
| *** rpodolyaka has joined #openstack-infra | 13:03 | |
| rpodolyaka | Hello, guys! I've made a few patches to sqlalchemy-migrate, which we have started to maintain recently. They enable running of the suite on different SQLAlchemy versions (0.7.x and 0.8.x) and different DB backends (MySQL and PostgreSQL in addition to SQLite). Could someone of you take a look at https://review.openstack.org/#/c/38459/ and https://review.openstack.org/#/c/40436/ , please? (the corresponding patch to openstack-infra/config is he | 13:08 |
| mordred | rpodolyaka: your last message cut of a little | 13:15 |
| openstackgerrit | Jiri Stransky proposed a change to openstack-infra/config: Add tuskar and tuskarclient to stackforge https://review.openstack.org/40856 | 13:16 |
| *** Ryan_Lane has joined #openstack-infra | 13:16 | |
| rpodolyaka | modred: oops... here is it: I've made a few patches to sqlalchemy-migrate, which we have started to maintain recently. They enable running of the suite on different SQLAlchemy versions (0.7.x and 0.8.x) and different DB backends (MySQL and PostgreSQL in addition to SQLite). Could someone of you take a look at https://review.openstack.org/#/c/38459/ and https://review.openstack.org/#/c/40436/ , please? (the corresponding patch to openstack-in | 13:16 |
| *** adalbas has joined #openstack-infra | 13:17 | |
| rpodolyaka | mordred: I believe, I've already shown you WIP patches. I've cleaned up them since the last time and enabled testing on MySQL and PostgreSQL | 13:18 |
| mordred | rpodolyaka: k. looking now | 13:18 |
| mordred | rpodolyaka: which is the openstack-infra/config patch? | 13:20 |
| *** Ryan_Lane has quit IRC | 13:20 | |
| mordred | that part keeps getting cut off :) | 13:20 |
| rpodolyaka | modred: https://review.openstack.org/#/c/39304/ | 13:21 |
| rpodolyaka | oops, typo :) | 13:21 |
| HenryG | pip problem solved -- needed to clear pip cache | 13:24 |
| *** prad has joined #openstack-infra | 13:24 | |
| *** vogxn has joined #openstack-infra | 13:28 | |
| openstackgerrit | Jiri Stransky proposed a change to openstack-infra/config: Add tuskar and tuskarclient to stackforge https://review.openstack.org/40856 | 13:29 |
| *** krtaylor has quit IRC | 13:29 | |
| *** changbl has quit IRC | 13:30 | |
| *** wu_wenxiang has joined #openstack-infra | 13:32 | |
| *** _TheDodd_ has joined #openstack-infra | 13:32 | |
| fungi | HenryG: great! i'm still stuck on trying to decipher ekarlso-'s websockify installation exception from earlier | 13:34 |
| *** avtar has joined #openstack-infra | 13:36 | |
| fungi | yolanda: i strongly suspect zuul hasn't been tested with an empty projects list in its layout, though jeblair may have more insights once he's around | 13:37 |
| *** cody-somerville has joined #openstack-infra | 13:41 | |
| *** vijendar has joined #openstack-infra | 13:42 | |
| ttx | fungi, mordred: we should apply capping to stable/folsom before it starts failing again -- any idea how to proceed ? | 13:44 |
| *** ruhe has joined #openstack-infra | 13:45 | |
| *** ruhe has quit IRC | 13:45 | |
| *** afazekas has quit IRC | 13:47 | |
| *** vogxn has quit IRC | 13:48 | |
| *** ruhe has joined #openstack-infra | 13:50 | |
| *** pentameter has joined #openstack-infra | 13:56 | |
| *** dkranz has quit IRC | 14:00 | |
| *** mrodden has joined #openstack-infra | 14:06 | |
| fungi | ttx: good question, there's probably a lot of research to be done to determine release versioning schemes used by different dependencies, and also there's a bit of a hole in that... do we hunt down all the transitive dependencies and cap them too? often that's where we wind up seeing breakages from new releases of things | 14:07 |
| *** ruhe has quit IRC | 14:09 | |
| mordred | sdague: are you on vacation already? | 14:09 |
| mordred | fungi: I say we start by grabbing a pip-freeze from a successful devstack run | 14:10 |
| fungi | mordred: and just populate the global-requirements.txt with <= each entry in that? | 14:11 |
| mordred | fungi: yeah. | 14:11 |
| fungi | merged with the minimum bounds we already have in it anyway | 14:11 |
| mordred | fungi: yup | 14:11 |
| fungi | and not worry about what if a dependency makes a security-oriented point release i guess? | 14:12 |
| mordred | fungi: I think we'll have to deal with that as it happens | 14:12 |
| mordred | I think the idea that we'll magically get goodness for our stable-1 release is basically continually undercut by people doing crazy things | 14:12 |
| fungi | probably so. it's a pretty large list to keep tabs on though | 14:12 |
| mordred | fungi: we should probably take the current requirements.txt and test-requirements.txt in requirements:stable/folsom | 14:13 |
| mordred | make copies of them to the side | 14:13 |
| mordred | then re-branch current master as stable/folsom | 14:13 |
| mordred | and then replace the contents of global-requirements with the old contents | 14:14 |
| *** datsun180b has joined #openstack-infra | 14:14 | |
| *** afazekas has joined #openstack-infra | 14:14 | |
| fungi | seems feasible | 14:14 |
| *** krtaylor has joined #openstack-infra | 14:15 | |
| mordred | then backport the devstack update-requirements first stuff | 14:15 |
| *** markmcclain has joined #openstack-infra | 14:15 | |
| *** markmcclain has quit IRC | 14:15 | |
| mordred | then rinse and repeat with stable/grizzly | 14:15 |
| zul | can someone +1 https://review.openstack.org/#/c/40851/ please | 14:16 |
| koolhead17 | zul, o.0 | 14:17 |
| fungi | zul: how is that different from https://review.openstack.org/40323 | 14:19 |
| zul | fungi: damn it! i totally missed that one | 14:20 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Fix pep8 error https://review.openstack.org/40869 | 14:21 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Sync requirements with global requirements https://review.openstack.org/40870 | 14:21 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Stop checking periods in commit messages https://review.openstack.org/40838 | 14:23 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Fix python-ldap mirroring. https://review.openstack.org/40732 | 14:23 |
| anteaya | where did pleia2 head off to? | 14:24 |
| *** burt has joined #openstack-infra | 14:24 | |
| *** vogxn has joined #openstack-infra | 14:24 | |
| fungi | anteaya: pennsylvania (i think) | 14:25 |
| anteaya | cool | 14:25 |
| fungi | given her departure time, i'm going to guess that she might still be in transit | 14:26 |
| *** changbl has joined #openstack-infra | 14:26 | |
| anteaya | fungi: a fair assumption | 14:27 |
| *** ^d has joined #openstack-infra | 14:27 | |
| *** ^d has joined #openstack-infra | 14:27 | |
| fungi | mordred: to what extent is the requirements proposal tooling dependent on recent packaging assumptions for the projects? (pbr, oslo, et cetera) | 14:28 |
| *** adalbas has quit IRC | 14:28 | |
| fungi | just trying to think through where the backport pain points are going to crop up | 14:28 |
| sdague | mordred: no | 14:28 |
| sdague | I'm on irc through tuesday (though tuesday all meetings) | 14:29 |
| fungi | sdague: that's not my idea of a vacation | 14:29 |
| sdague | fungi: it's not vacation until wed :) | 14:29 |
| fungi | quite clearly | 14:29 |
| mordred | sdague: yay! because ... syncing things and dev-requirements == not doing the right thing | 14:30 |
| mordred | and I wanted to talk to you about it | 14:30 |
| sdague | mordred: ok, I got a meeting right now | 14:30 |
| sdague | let's talk after | 14:30 |
| mordred | sdague: awesome | 14:31 |
| *** xchu has joined #openstack-infra | 14:34 | |
| *** markmcclain has joined #openstack-infra | 14:40 | |
| openstackgerrit | A change was merged to openstack-infra/config: Add python33 tests to all openstack python clients https://review.openstack.org/40323 | 14:40 |
| sdague | mordred: ok, I can not pay attention to this meeting for a bit. What's the issue? | 14:42 |
| *** adalbas has joined #openstack-infra | 14:44 | |
| *** pabelanger_ has joined #openstack-infra | 14:45 | |
| *** pabelanger_ has quit IRC | 14:45 | |
| *** pabelanger_ has joined #openstack-infra | 14:45 | |
| *** pabelanger has quit IRC | 14:45 | |
| *** pabelanger_ is now known as pabelanger | 14:45 | |
| *** pabelanger_ has joined #openstack-infra | 14:46 | |
| mordred | sdague: https://review.openstack.org/#/c/40755/1/requirements.txt | 14:46 |
| *** cppcabrera has joined #openstack-infra | 14:46 | |
| *** zyluo has joined #openstack-infra | 14:46 | |
| mordred | sdague: the merge is downgrading oslo.config in the target repos | 14:46 |
| sdague | right, so that's expected.... yeh the pre release problem is an odd one | 14:47 |
| *** pycabrera has joined #openstack-infra | 14:47 | |
| *** pycabrera has quit IRC | 14:48 | |
| sdague | ok, what's the desired behavior? | 14:48 |
| mordred | is it time to start exploring the "upload pre-release files only as wheels to pypi?" approach? | 14:48 |
| sdague | well... maybe | 14:48 |
| *** rcleere has joined #openstack-infra | 14:48 | |
| *** markmc has joined #openstack-infra | 14:48 | |
| sdague | honestly, I think we need to think about getting other project git trees onto the unit test nodes | 14:49 |
| mordred | there's two issues | 14:49 |
| sdague | because we'll have to figure out that for neutron vs. nova | 14:49 |
| mordred | one is testing | 14:49 |
| mordred | one is being able to use the repos not on our test nodes | 14:49 |
| *** pycabrera has joined #openstack-infra | 14:49 | |
| mordred | because oslo.config prereleases are not on pypi, we get into the weird problem | 14:50 |
| sdague | right | 14:50 |
| mordred | but we didn't put them there because we don't want grizzly to get an update all of a sudden - and until pip 1.4 that's not possible | 14:50 |
| *** pabelanger has quit IRC | 14:50 | |
| mordred | BUT - we pretty much require pip 1.4 for havana now | 14:50 |
| sdague | so short term, we can take into account dev requirements | 14:50 |
| *** pabelanger has joined #openstack-infra | 14:50 | |
| *** cppcabrera has quit IRC | 14:50 | |
| *** pycabrera is now known as cppcabrera | 14:51 | |
| mordred | so the idea was that we upload pre-releases of osloto pypi - but only in wheel form | 14:51 |
| mordred | pip less than 1.3 will not find them | 14:51 |
| mordred | because it doesn't know what wheels are | 14:51 |
| markmc | what happens if you use 1.4 with grizzly? | 14:51 |
| mordred | it still won't find them | 14:51 |
| *** apevec has joined #openstack-infra | 14:51 | |
| mordred | because grizzly does not have a pre-release version in its requirements | 14:51 |
| markmc | ah, yes | 14:52 |
| mordred | so it won't upgrade oslo.config until oslo.config releases its next real release | 14:52 |
| markmc | yes | 14:52 |
| markmc | have we figured out how to do wheel-only uploads? | 14:52 |
| mordred | that part should be pretty simple - slightly different upload job - but simple and easy to do | 14:52 |
| sdague | mordred... that makes me realize we need to get oslo.config into stable/grizzly devstack before the oslo release | 14:53 |
| mordred | sdague: yes | 14:53 |
| sdague | that should be easy enough though | 14:53 |
| mordred | we have some requirements and devstack backportin to do | 14:53 |
| mordred | so I think steps are a) short term account for dev-reqs in update.py better | 14:54 |
| mordred | b) backport devstack and requirements changes to stable branches | 14:54 |
| sdague | mordred: yeh, a I can probably do today | 14:54 |
| mordred | c) get wheel support and wheel uploads properly into havana | 14:54 |
| mordred | I believe fungi may have started on b | 14:54 |
| *** dina_belova has quit IRC | 14:54 | |
| sdague | it will be later this afternoon, after the qa meeting | 14:54 |
| mordred | I'll start hacking on c | 14:54 |
| fungi | starting to dig into b today, yes | 14:55 |
| mordred | I think first step is I'm going to add wheel building to run-mirror | 14:55 |
| sdague | a) will also let me play with my update.py min versions thing | 14:56 |
| mordred | fungi: did we land the gate-geepyb-on-pbr-integration yet? | 14:56 |
| mordred | sdague: markmc will be veryhappy about that | 14:56 |
| mordred | he's wanted that for AGES | 14:56 |
| sdague | which I think we need to do as well, because our min versions is very bad | 14:56 |
| fungi | mordred: don't remember if it's approved yet... checking | 14:56 |
| sdague | :) | 14:56 |
| sdague | mordred: it makes me realize that crazy idea I had about random walking the version space to figure out what deps actually work.... | 14:57 |
| *** yaguang has joined #openstack-infra | 14:57 | |
| sdague | will have most of the infrastructure in place after that | 14:57 |
| fungi | mordred: still not passing tests... https://review.openstack.org/38176 | 14:57 |
| mordred | sdague: ooh. I like that crazy idea! | 14:57 |
| sdague | yeh, you did when I put it on the ML as well :) | 14:57 |
| mordred | fungi: oh sod it | 14:57 |
| fungi | mordred: just the layout validator. i can fix it up if you need | 14:58 |
| mordred | fungi: can you please? I think I've hit a mental block on that change | 14:58 |
| fungi | you bet | 14:58 |
| sdague | mordred: your pep8 tox add to requirements is failing pep8? | 15:01 |
| sdague | any chance you can get that fixed soon, so we can get your update.py patches in? | 15:01 |
| mordred | sdague: on it | 15:02 |
| fungi | "irony" | 15:02 |
| sdague | I think it's just a tox config issue | 15:02 |
| mordred | sdague: yup. fixed | 15:04 |
| openstackgerrit | Monty Taylor proposed a change to openstack/requirements: Remove unneeded tests dir reference https://review.openstack.org/40002 | 15:04 |
| openstackgerrit | Monty Taylor proposed a change to openstack/requirements: Update ourselves to ourselves https://review.openstack.org/39963 | 15:04 |
| openstackgerrit | Monty Taylor proposed a change to openstack/requirements: Add pep8 checks https://review.openstack.org/39967 | 15:04 |
| dansmith | mordred: linkedin just told me you went to NCSU.. | 15:05 |
| fungi | lies, all lies | 15:05 |
| * fungi lives 10 minutes from ncsu main campus | 15:06 | |
| mordred | dstufft: that is, in fact, a true statement | 15:06 |
| mordred | I was in CATT at the time too, in case that means anything to you | 15:06 |
| fungi | tab completion fails you again ;) | 15:06 |
| mordred | GAH | 15:07 |
| mordred | dansmith: ^^ | 15:07 |
| dansmith | fungi: oh? | 15:07 |
| dansmith | mordred: actually, it doesn't | 15:07 |
| dansmith | mordred: but from the date linkedin says you were there, I think the bricks must have still been fresh :P | 15:07 |
| *** rscottcoyle has joined #openstack-infra | 15:07 | |
| mordred | dansmith: it was the computer geek theme dorm back in the day | 15:07 |
| dansmith | mordred: oh, nice | 15:07 |
| mordred | dansmith: we had ethernet IN OUR DORM ROOMS | 15:08 |
| dansmith | heh | 15:08 |
| mordred | and by ethernet - I mean twisted-pair and 10base2 | 15:08 |
| dansmith | the real stuff | 15:08 |
| mordred | yup | 15:08 |
| mordred | terminators and whatnot | 15:08 |
| fungi | 10base5 was the real stuff ;) | 15:08 |
| mordred | it was very fancy | 15:08 |
| mordred | haha | 15:08 |
| *** boris-42 has quit IRC | 15:08 | |
| dansmith | fungi: well, I meant the stuff with actual ether, 10base5 included :) | 15:09 |
| * fungi still has nightmares about routing thicknet lines | 15:09 | |
| *** mrodden has quit IRC | 15:10 | |
| fungi | dansmith: are you local? | 15:10 |
| dansmith | fungi: no, I'm in PDX now, but I grew up in Chapel Hill and lived in Raleigh through school before I moved out here | 15:10 |
| *** dkranz has joined #openstack-infra | 15:11 | |
| *** rnirmal has joined #openstack-infra | 15:11 | |
| fungi | ahh, cool. i've lived in chapel hill and (eww) cary over the past decade, but am in raleigh now | 15:11 |
| *** mestery_ is now known as mestery | 15:12 | |
| clarkb | The number of north carolinians around here is scary ;) | 15:12 |
| fungi | north carolina is scary in general | 15:12 |
| fungi | though it's a little less scary here "in the city" (heh) than it was in hillbilly country where i grew up | 15:13 |
| mordred | fungi is a fillbilly | 15:14 |
| mordred | gah | 15:15 |
| mordred | I CANNOT TYPE THIS WEEK | 15:15 |
| *** krtaylor has quit IRC | 15:15 | |
| jeblair | i reckon we all ought to git together and have us a pig pickin sometime ;) | 15:16 |
| * jeblair made barbecue this past weekend | 15:16 | |
| *** cppcabrera is now known as cppcabrera_afk | 15:17 | |
| *** dhellmann_ has quit IRC | 15:18 | |
| fungi | ooh, that does sound wonderful | 15:18 |
| *** vogxn has quit IRC | 15:18 | |
| fungi | also, i appreciate your colorful use of the term "reckon" | 15:19 |
| ttx | jeblair: the new zuul status view with time estimates is really useful, thx | 15:19 |
| *** mrodden has joined #openstack-infra | 15:19 | |
| *** dhellmann has joined #openstack-infra | 15:19 | |
| ttx | jeblair: could benefit from having all progress bars right-aligned though | 15:19 |
| *** SergeyLukjanov has quit IRC | 15:20 | |
| *** dina_belova has joined #openstack-infra | 15:20 | |
| ttx | would look a lot less like a brick breaker game though | 15:20 |
| *** dina_belova has quit IRC | 15:20 | |
| * fungi plays super breakout with the zuul status screen | 15:21 | |
| * ttx spent too much time playing Arkanoid when he was younger. | 15:21 | |
| clarkb | ttx: sdague wrote that change iirc | 15:21 |
| clarkb | its in review somewhere | 15:22 |
| ttx | clarkb: heh | 15:22 |
| ttx | https://review.openstack.org/#/c/39097/ | 15:23 |
| fungi | arkanoid was neat, but i didn't play it as much as its breakout predecessors | 15:24 |
| *** SergeyLukjanov has joined #openstack-infra | 15:25 | |
| *** dina_belova has joined #openstack-infra | 15:27 | |
| *** pabelanger has quit IRC | 15:30 | |
| *** nicedice_ has joined #openstack-infra | 15:30 | |
| *** vijendar has quit IRC | 15:31 | |
| *** ruhe has joined #openstack-infra | 15:32 | |
| ttx | fungi: missed your anwser earlier. Transitive deps ? Hadn't thought of that. Looks like we almost need to run a pristine install, observe what gets installed, and generate the list of pinned stuff | 15:33 |
| fungi | ttx: that was basically mordred's suggestion (pip freeze after a full devstack run in stable) | 15:34 |
| * fungi runs away to lunch... bbl | 15:34 | |
| ttx | fungi: ack | 15:35 |
| *** CliMz has quit IRC | 15:35 | |
| ttx | fungi, mordred: agree that pinning deps is not about freezing them. We can still manually bump for corner cases. | 15:35 |
| *** yaguang has quit IRC | 15:36 | |
| *** yolanda has quit IRC | 15:36 | |
| zaro | morning | 15:39 |
| anteaya | morning zaro | 15:40 |
| *** markmcclain has quit IRC | 15:41 | |
| *** pabelanger has joined #openstack-infra | 15:44 | |
| *** dina_belova has quit IRC | 15:45 | |
| *** krtaylor has joined #openstack-infra | 15:45 | |
| *** dina_belova has joined #openstack-infra | 15:47 | |
| *** yaguang has joined #openstack-infra | 15:49 | |
| *** cppcabrera_afk is now known as cppcabrera | 15:49 | |
| cppcabrera | n/close | 15:49 |
| *** cppcabrera has left #openstack-infra | 15:49 | |
| *** odyssey4me has quit IRC | 15:52 | |
| *** jpich has quit IRC | 15:57 | |
| *** derekh has quit IRC | 15:57 | |
| *** markmc has quit IRC | 15:58 | |
| zaro | fungi: tried to assign bug 1172419 to you but i could not. is this because i'm lacking permissions? | 15:58 |
| uvirtbot | Launchpad bug 1172419 in openstack-ci "build bitrot jobs after mirror update" [Low,In progress] https://launchpad.net/bugs/1172419 | 15:58 |
| *** rpodolyaka has left #openstack-infra | 15:59 | |
| *** dina_belova has quit IRC | 15:59 | |
| clarkb | `sudo chmod 777` :/ | 15:59 |
| *** xchu has quit IRC | 16:00 | |
| *** anteaya has quit IRC | 16:01 | |
| *** vijendar has joined #openstack-infra | 16:02 | |
| *** dina_belova has joined #openstack-infra | 16:03 | |
| *** zaro0508 has quit IRC | 16:04 | |
| mriedem | i see that oslo doesn't list a version constraint on pbr, does anyone know what it really requires at a minimum for building with pbr? | 16:04 |
| *** datsun180b has quit IRC | 16:04 | |
| *** dina_belova has quit IRC | 16:04 | |
| clarkb | mriedem: probably the latest version of pbr | 16:04 |
| clarkb | pbr has had a ton of bug fixes recently and I am sure mordred would want everyone to use the latest version | 16:05 |
| mriedem | clarkb: ok, yeah, i saw the note go out recently about the overhaul | 16:05 |
| mordred | yes. seriously, nothing in openstack should ever use less than 0.5.21 | 16:05 |
| clarkb | https://review.openstack.org/#/c/36294/8 looks like the change that is `sudo chmod 777`ing | 16:05 |
| mordred | :) | 16:05 |
| *** vogxn has joined #openstack-infra | 16:06 | |
| clarkb | but it appears to have newer draft changes | 16:06 |
| *** thomasbiege2 has joined #openstack-infra | 16:07 | |
| *** nijaba has quit IRC | 16:07 | |
| *** thomasbiege1 has quit IRC | 16:07 | |
| *** emagana has joined #openstack-infra | 16:07 | |
| clarkb | so I can't see if that is the case | 16:07 |
| *** ruhe has quit IRC | 16:09 | |
| *** SergeyLukjanov has quit IRC | 16:11 | |
| koolhead17 | pleia2, ping | 16:13 |
| * jeblair runs errands, biab | 16:13 | |
| *** fbo is now known as fbo_away | 16:14 | |
| *** wu_wenxiang has quit IRC | 16:16 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/nose-html-output: Remove nose from setup_requires https://review.openstack.org/40903 | 16:20 |
| mordred | clarkb: ^^ I would very much like to land that and relase | 16:20 |
| mordred | clarkb: do tags trigger releases of that, or are we still doing it manually? | 16:21 |
| *** reed has joined #openstack-infra | 16:21 | |
| openstackgerrit | Sascha Peilicke proposed a change to openstack/requirements: Require lesscpy>=0.9j https://review.openstack.org/40910 | 16:27 |
| *** svarnau has joined #openstack-infra | 16:27 | |
| *** yaguang has quit IRC | 16:31 | |
| *** yaguang has joined #openstack-infra | 16:31 | |
| *** moted has quit IRC | 16:31 | |
| *** moted has joined #openstack-infra | 16:32 | |
| *** Ryan_Lane has joined #openstack-infra | 16:35 | |
| *** yaguang has quit IRC | 16:38 | |
| clarkb | mordred manual iirc | 16:40 |
| *** markmcclain has joined #openstack-infra | 16:42 | |
| *** anteaya has joined #openstack-infra | 16:47 | |
| *** emagana has quit IRC | 16:48 | |
| clarkb | I can sort it out when I am not ferrying tourists but dont let me stop you | 16:50 |
| *** ruhe has joined #openstack-infra | 16:51 | |
| mordred | clarkb: I can just do it if you're ok with that | 16:51 |
| *** mrodden has quit IRC | 16:55 | |
| *** mrodden has joined #openstack-infra | 16:56 | |
| fungi | zaro: without being on the bug team for a project, you can only assign and unassign a ticket to yourself but can't assign to anyone else. i'll update and address it though | 16:56 |
| *** jgriffith has quit IRC | 16:57 | |
| *** pabelanger has quit IRC | 16:58 | |
| jog0 | lifeless: keeping editor files in the localgitignore seems like the hard way of fixing the issue, IMHO setting a global gittignore for editor related files is easier. also keeps the repo gitignore more susinct and project related | 17:00 |
| *** pabelanger has joined #openstack-infra | 17:01 | |
| *** dina_belova has joined #openstack-infra | 17:05 | |
| * fungi lists his editor files in ~/.gitignore and adds excludesfile '~/.gitignore' | 17:05 | |
| zaro | fungi: thx. | 17:05 |
| fungi | ...in the [core] section of ~/.gitconfig | 17:05 |
| *** avtar has quit IRC | 17:06 | |
| apevec | adam_g, ttx - I'll start tagging 2013.1.3 | 17:07 |
| *** jgriffith has joined #openstack-infra | 17:08 | |
| *** dina_belova has quit IRC | 17:08 | |
| clarkb | mordred yup fine by me | 17:08 |
| fungi | pleia2: if you're around, i played with git.o.o a bit and confirmed the missing https bits are that it's not set with listen 443 and also does't loadmodule ssl_module mod_ssl.so | 17:09 |
| openstackgerrit | A change was merged to openstack-infra/nose-html-output: Remove nose from setup_requires https://review.openstack.org/40903 | 17:09 |
| fungi | pleia2: also nothing's getting replicated to it yet, but i suspect we may need to restart gerrit to pick that change up | 17:10 |
| mordred | fungi: I, for one, favor making gerrit pick up that config setting | 17:11 |
| fungi | mordred: you welcome our new gerrit restart overlords, eh? | 17:12 |
| mordred | fungi: yes | 17:12 |
| fungi | i can float a change to subscribe the gerrit service to that file, i suppose | 17:12 |
| fungi | still trying to sort out your zuul layout validation fail on the jeepyb gating change. it's choking on an envvar in a shell snippet, looks like. maybe { getting interpreted by the yaml parser? | 17:13 |
| bodepd | I still cannot submit patches to the branches of the stackforge projects | 17:15 |
| *** dina_belova has joined #openstack-infra | 17:15 | |
| fungi | bodepd: i'll check your acls got applied | 17:16 |
| *** fbo_away is now known as fbo | 17:17 | |
| bodepd | fungi: is that in the acl? | 17:17 |
| fungi | bodepd: stackforge/puppet-nova for example definitely shows in the webui that puppet-manager-core has exclusive approval and -2/+2 review rights on refs/heads/stable/* | 17:17 |
| bodepd | fungi: let me try that... | 17:18 |
| fungi | bodepd: what error do you get? | 17:18 |
| bodepd | it says there are no changes | 17:18 |
| bodepd | (no new changes) | 17:19 |
| fungi | bodepd: following a cherry-pick and review procedure similar to https://wiki.openstack.org/wiki/StableBranch#Workflow | 17:19 |
| *** nati_ueno has joined #openstack-infra | 17:19 | |
| *** emagana has joined #openstack-infra | 17:20 | |
| *** dina_belova has quit IRC | 17:20 | |
| fungi | bodepd: or if all the commits in your current branch are destined for e.g. stable/grizzly then just try 'git review stable/grizzly' and see what it says | 17:20 |
| fungi | bodepd: similarly, (i continue to assume you're using the git-review tool here) adding a defaultbranch parameter to the .gitreview on your stable branches will allow you to not need to specify the review target branch if you're on a remote tracking branch | 17:22 |
| fungi | a number of our projects make use of that convenience | 17:22 |
| mgagne | fungi: http://paste.openstack.org/show/43610/ | 17:23 |
| fungi | what's the sha of the tip of your current local branch you're trying to push? | 17:24 |
| mgagne | 0ce791d | 17:24 |
| fungi | ahh, and gerrit says 87edf77 is the current tip of stable/grizzly there, so there definitely should be differences | 17:25 |
| *** HenryG has quit IRC | 17:25 | |
| clarkb | are you pushing any unreviewed merge commits? | 17:26 |
| *** pcrews has quit IRC | 17:26 | |
| clarkb | gerrit doesnt want you to push merge commits | 17:26 |
| *** Ryan_Lane has quit IRC | 17:26 | |
| fungi | ahh, yeah if 0ce791d is a merge commit it might not think there is a relevant change | 17:27 |
| mgagne | a merge? it's a normal commit with an empty file added | 17:27 |
| *** ruhe has quit IRC | 17:28 | |
| bodepd | fungi: I was missing the cherry-pick command. Sorry, I did not find that document before. thanks for the link | 17:31 |
| fungi | mgagne: does 87edf77 show up in your current patch history on your local branch? how many commits back? are there any merge commits appearing between that sha and the tip of your local branch? | 17:32 |
| mgagne | fungi: http://paste.openstack.org/show/43612/ | 17:33 |
| fungi | but yes, generally gerrit's workflow best accommodates cherry-picking from your development branch into your stable branch and submitting those for review | 17:33 |
| pleia2 | fungi: thanks! I'm going to get food and possibly a nap, then I'll get some patches in :) | 17:35 |
| pleia2 | fungi: and I think you're right about gerrit, it's part of config files so reload/restart makes sense | 17:35 |
| *** datsun180b has joined #openstack-infra | 17:36 | |
| fungi | okay, so 0ce791d has 87edf77 as its direct parent. and git diff 87edf77..0ce791d has some change of content? (i can't remember if committing just an empty file will be sane on gerrit but i'll test that myself in a sec) | 17:36 |
| apevec | adam_g, ttx - tags pushed, how long should it take for tarballs to show up? | 17:36 |
| *** mrodden1 has joined #openstack-infra | 17:36 | |
| mgagne | fungi: ok, I can add content to the file | 17:36 |
| ttx | apevec: you got that last security patch for cinder/grizzly in ? | 17:36 |
| *** mrodden has quit IRC | 17:36 | |
| apevec | ttx, yep | 17:36 |
| ttx | tarballs take up to 10min | 17:37 |
| mgagne | fungi: ok, it seems it's a bad idea to commit empty file | 17:37 |
| ttx | apevec: I have a script that does everything from tagging to uploading tarballs and comparing md5s | 17:37 |
| ttx | we should adapt that for stable branch releases | 17:37 |
| mgagne | bodepd: you should be able to submit changes to stable/grizzly... | 17:37 |
| apevec | ttx, we are using your script for stable | 17:37 |
| apevec | it just works | 17:38 |
| ttx | apevec: the upload_release one... not the mp_delivery.sh one | 17:38 |
| apevec | https://github.com/ttx/openstack-releasing/blob/master/upload_release.py | 17:38 |
| apevec | yep | 17:38 |
| ttx | apevec: we should push that under gerrit somewhere too, now that we are a PROGRAM | 17:38 |
| *** dprince has quit IRC | 17:38 | |
| ttx | apevec: mp_delivery does tagging + wait + upload | 17:38 |
| apevec | ah, I see | 17:39 |
| ttx | still a bit flaky atm | 17:39 |
| apevec | no hands! | 17:39 |
| *** SergeyLukjanov has joined #openstack-infra | 17:39 | |
| fungi | mgagne: though interestingly, i seem to be able to commit a single empty file for review... http://paste.openstack.org/show/43613 | 17:39 |
| nati_ueno | Hi who's core for review days? I wanna reviewer for https://review.openstack.org/#/c/40480/ | 17:39 |
| ttx | apevec: you can track the jobs @ https://jenkins.openstack.org/job/glance-tarball/? for example | 17:40 |
| ttx | also under http://status.openstack.org/zuul/ in the release pipe | 17:41 |
| apevec | ttx, thanks! | 17:41 |
| ttx | (lower right corner of the zuul status view) | 17:41 |
| mgagne | fungi: could it be impossible to submit empty file to stable? | 17:41 |
| *** pcrews has joined #openstack-infra | 17:42 | |
| *** sarob has joined #openstack-infra | 17:42 | |
| fungi | mgagne: maybe, though that seems like a side effect if so and not something gerrit would enforce directly | 17:42 |
| ttx | apevec: I can add you to the auto-accepted people on openstack-announce | 17:44 |
| ttx | apevec: which email address are you using for openstack-announce ? | 17:44 |
| apevec | ttx, apevec@gmail.com | 17:44 |
| ttx | apevec: oh, you're cleared already | 17:45 |
| ttx | so it should just work | 17:45 |
| *** thomasbiege has joined #openstack-infra | 17:45 | |
| *** thomasbiege2 has quit IRC | 17:45 | |
| * ttx calls it a day | 17:46 | |
| fungi | ttx: have a great evening! | 17:46 |
| ttx | apevec: you seem to be on the right track now :) | 17:46 |
| openstackgerrit | gordon chung proposed a change to openstack/requirements: add pycadf library https://review.openstack.org/40926 | 17:46 |
| *** jerryz has joined #openstack-infra | 17:46 | |
| apevec | ttx, yep. have a nice evening! | 17:47 |
| * ttx is a bit lost with the new jenkins setup | 17:47 | |
| ttx | jeblair: used to check https://jenkins.openstack.org/job/nova-tarball/ but now it presents an outdated view | 17:48 |
| ttx | https://jenkins01.openstack.org/job/nova-tarball/ has the right thing | 17:48 |
| ttx | and jenkins02 has an empty list | 17:48 |
| ttx | not sure how we can get a useful list now | 17:49 |
| ttx | or pertinent weather report for jobs | 17:49 |
| ttx | I guess you'll have to educate me. Tomorrow. | 17:50 |
| *** vogxn has quit IRC | 17:50 | |
| fungi | ttx: logs.openstack.org will probably be a better centralized place to pull them from | 17:53 |
| *** annegentle is now known as annegentle_vacay | 17:53 | |
| *** _TheDodd_ has quit IRC | 17:55 | |
| mordred | fungi, jeblair: dkranz has offered to help get fedora in the devstack gate | 17:56 |
| fungi | oh! | 17:57 |
| fungi | mordred: for d-g, do we need puppet to still work? | 17:57 |
| mordred | dkranz: I believe you want jeblair's d-g patch re-arranging config to land | 17:57 |
| mordred | fungi: yes | 17:57 |
| fungi | i think at the moment we need newer puppet and hiera to support fedora18 | 17:57 |
| *** mrodden has joined #openstack-infra | 17:57 | |
| mordred | although - I believe what we decided the other day is that we thought centos in the devstack gate would make more sense | 17:57 |
| *** ^demon has joined #openstack-infra | 17:57 | |
| mordred | sdague: ^^ | 17:57 |
| fungi | or an older ruby installed on the fedora machines would theoretically work | 17:58 |
| mordred | for the same reason that we use precise | 17:58 |
| mordred | right? | 17:58 |
| clarkb | mordred the rootwrap discussion has basically become "replace openstack with a bunch of gearman workers" | 17:58 |
| mordred | awesome | 17:58 |
| clarkb | fedora or centos? | 17:58 |
| mordred | clarkb: right. what did we decide the other day? | 17:59 |
| *** ^d has quit IRC | 17:59 | |
| *** thomasm has quit IRC | 17:59 | |
| clarkb | I thibk we need something that csn run tests for 18 months | 17:59 |
| clarkb | 6 for dev cycle 12 for stable | 17:59 |
| *** ^demon is now known as ^d | 17:59 | |
| sdague | mordred: so centos is really a problem for devstack as a primary | 18:00 |
| mordred | sdague: whyfore? | 18:01 |
| sdague | because it's still not really functional | 18:01 |
| *** mrodden1 has quit IRC | 18:01 | |
| sdague | centos is *sooooooo* old | 18:01 |
| dkranz | mordred: I think fedora is what a lot of people are using for dev so that made sense to me. | 18:02 |
| mordred | so - with the gate... | 18:02 |
| mordred | once something goes into stable | 18:02 |
| mordred | we continue testing it ont he release it was tested on when it was in dev | 18:02 |
| mordred | because of moving targets and whatnot | 18:02 |
| *** _TheDodd_ has joined #openstack-infra | 18:02 | |
| sdague | mordred: so like, for instance, you want real neutron testing | 18:03 |
| sdague | you can't use centos, kernel doesn't do it | 18:03 |
| mordred | wow. stellar | 18:03 |
| mordred | any thought of how to handle the short lifespan of fedora on our 18 month world? | 18:03 |
| sdague | mordred: so fedora is really about 14 months of updates | 18:03 |
| dkranz | Have to step into meeting | 18:03 |
| sdague | which isn't 18, but it's close | 18:04 |
| *** krtaylor has quit IRC | 18:04 | |
| *** pcrews has quit IRC | 18:04 | |
| * mordred really hates supporting stable -1 | 18:04 | |
| sdague | yeh, what if we just didn't with fedora | 18:05 |
| sdague | only do master and stable | 18:05 |
| clarkb | fedora would sit under an existing test right? | 18:06 |
| sdague | clarkb: yeh, move the pg job to fedora | 18:06 |
| fungi | so, another wrinkle... last i looked at it there were challenges getting newer puppet to run on precise, getting older puppet to run on fedora, and getting old and new puppet to use a common master and hiera version... which takes us to two separate puppet masters or something other than puppet | 18:06 |
| sdague | probably change the queue to qpid in the process as well | 18:06 |
| clarkb | as long as not testing pg is ok for stable -1 | 18:06 |
| clarkb | fungi new puppet on precise should work | 18:07 |
| sdague | clarkb: yeh, it's a risk, but it would leave us in a much better state than not testing fedora in master | 18:07 |
| mordred | I think that's not unreasonable | 18:07 |
| bodepd | fungi: I run new puppet on precise | 18:07 |
| mordred | should we at some point work out migrating to new puppet so that we don't have to keep worrying the the ruby/puppet version problems? | 18:08 |
| fungi | clarkb: new puppet needs new ruby and, at least the bug reports i was looking at, the ruby backports to precise weren't new enough or had some serious bugs making them unsuitable for new puppet | 18:08 |
| fungi | bodepd: ^ i meant | 18:08 |
| clarkb | mordred yes | 18:08 |
| fungi | of course i didn't try, so the bugs may have been out of date | 18:08 |
| fungi | or bodepd... maybe you don't run canonical's ruby packages on precise? | 18:08 |
| clarkb | mordred we can use two masters and migrate in steps | 18:08 |
| bodepd | fungi: that is possible. I installed puppet form apt.puppetlabs.com | 18:09 |
| mordred | clarkb: because puppet3 only talks to puppet3 master? | 18:09 |
| bodepd | which I believe pulls in ruby 1.9.3 | 18:09 |
| clarkb | mordred yes | 18:09 |
| mordred | nod | 18:09 |
| fungi | bodepd: ahh, 1.9.3 from somewhere besides canonical's precise-backports i guess. i think 1.9.2.1 or something is their latest | 18:10 |
| adam_g | apevec, cool. sorry i haven't been more of a help this time around, i've been traveling this week | 18:10 |
| apevec | adam_g, np, you get next Oct 10 release :) | 18:10 |
| adam_g | apevec, i have been running many test runs on the tarball since monday and everything looks good | 18:10 |
| apevec | 2013.1.3 tarballs are ready, I'm uploading to LP | 18:11 |
| * clarkb plays devils advocate. | 18:13 | |
| *** pcrews has joined #openstack-infra | 18:13 | |
| clarkb | if distros arent going to invest in supporting their users for usable amounts of time should we be putting extra effort into supporting them? | 18:13 |
| clarkb | we stopped testing new ubuntu because 9 months is far too short | 18:14 |
| clarkb | 14 is better but not significatly | 18:14 |
| *** jpich has joined #openstack-infra | 18:15 | |
| dtroyer | but 3-5 years is a bit long given the pace of Python's development churn | 18:15 |
| *** dina_belova has joined #openstack-infra | 18:15 | |
| fungi | bodepd: actually, looking back at the versions, precise-updates has ruby 1.9.3.0 but i think i read something about puppet 3.x needing 1.9.3.1 or newer. any guess whether that's still the case? | 18:16 |
| mordred | let me say that in a different way - how about we stop caring about stable -1 | 18:18 |
| mordred | it truly seems like we only care about it in theory | 18:18 |
| mordred | but we still let its existence drive other design decisions | 18:19 |
| clarkb | mordred our users care about it | 18:19 |
| mordred | who? | 18:19 |
| dtroyer | at least for Fedora, I'm fine with that. Fedora deployments have already chosen a non-long-term-stable OS. | 18:19 |
| clarkb | people running old stuff | 18:19 |
| mordred | clarkb: these are theoretical people | 18:19 |
| clarkb | you still se a lot of folsom discussions on the list | 18:19 |
| mordred | but I don't see a lot of folsom patches | 18:19 |
| * fungi still sees essex discussions | 18:19 | |
| fungi | thankfully not so many diablo discussions though | 18:20 |
| mordred | and I expect that _we_ are not the source of essex or folsom deployments right now | 18:20 |
| *** dina_belova has quit IRC | 18:20 | |
| fungi | not unless i'm doing it in my sleep | 18:20 |
| mordred | grumble. grumble. | 18:21 |
| jeblair | i'm back | 18:21 |
| mordred | yay1 | 18:21 |
| mordred | we missed you | 18:21 |
| sdague | clarkb: we can care less, by running a subset of precise jobs only | 18:21 |
| jeblair | i'm not keen on the infrastructure team deciding openstack's support policy... | 18:21 |
| clarkb | ++ | 18:22 |
| *** mikal has quit IRC | 18:22 | |
| mordred | jeblair: totally | 18:22 |
| clarkb | to jeblairs comment | 18:22 |
| * mordred takes off infrastructure hat, puts on normal openstack dev hat "why do we care about stable -1?" | 18:22 | |
| * mordred puts infrastructure hat back on | 18:22 | |
| clarkb | I don't want to make the decision. but do want to talk about the considerations that need to be made surrounding support | 18:22 |
| *** mikal has joined #openstack-infra | 18:23 | |
| fungi | bodepd: clarkb: found it... "Ruby Bug Warning: Ruby 1.9.3-p0 has bugs that cause a number of known issues with Puppet 3.2.0 and later, and you should use a different release. To the best of our knowledge, these issues were fixed in the second public release of Ruby 1.9.3 (p125), and we are positive they are resolved in p392 (which ships with Fedora 18). Unfortunately, Ubuntu Precise ships with p0 for | 18:23 |
| fungi | some reason, and there’s not a lot we can do about it. If you’re using Precise, we recommend using Puppet Enterprise or installing a third-party Ruby package." | 18:23 |
| fungi | from http://docs.puppetlabs.com/puppet/3/reference/release_notes.html#puppet-320 | 18:23 |
| jeblair | mordred: at the risk of running a circular argument, because there is a team of people who (claim they) care about stable -1. | 18:24 |
| mordred | jeblair: yeah. nod | 18:24 |
| jeblair | mordred: but yeah that's the not-infrastructure-hat question | 18:25 |
| mordred | jeblair: more immediatley imporant - what do you think of sdague's suggestion that we run the pg/qpid tests on fedora and stop running them for stable -1 | 18:25 |
| *** pcrews has quit IRC | 18:25 | |
| jeblair | mordred: the idea of running fewer tests on stable-1 over time bothers me. | 18:26 |
| mordred | premise being the being broken on fedora is more costly thatn stable -l1 possibly being untested in the postgres/qpid combo | 18:26 |
| jeblair | mordred: being broken on latest fedora, which is also something we 'support' also bothers me. to be fair. | 18:26 |
| mordred | yeah | 18:27 |
| mordred | also, the _way_ we're broken on fedora is definitely something that running d-g there would have caught | 18:27 |
| *** zul has quit IRC | 18:28 | |
| jeblair | mordred: if openstack is impossible to run on centos6, then, ah, it seems like someone has broken our support policy. | 18:28 |
| mordred | jeblair: :) | 18:28 |
| mordred | jeblair: I believe it's impossible to test neutron for real on centos6 | 18:29 |
| *** pcrews has joined #openstack-infra | 18:29 | |
| mordred | jeblair: due to too old kernel | 18:29 |
| *** sarob has quit IRC | 18:29 | |
| *** sarob has joined #openstack-infra | 18:29 | |
| jeblair | mordred: when you say stable-1 which of the current stable releases do you mean? | 18:31 |
| sdague | jeblair: folsom | 18:31 |
| mordred | folsom | 18:31 |
| jeblair | it's one thing to test fedora on master and the most recent stable but not older than that; but it's another thing to stop testing other things, eg postgres, on older releases. | 18:33 |
| *** sarob has quit IRC | 18:34 | |
| sdague | jeblair: well we could do that with precise | 18:34 |
| sdague | jeblair: or equiv | 18:34 |
| jeblair | so neutron really doesn't work on rhel at all? | 18:35 |
| apevec | netns doesn't work w/ vanilla rhel kernel | 18:35 |
| sdague | jeblair: not in the ways people want to use it, without an updated kernel | 18:35 |
| sdague | right, exactly | 18:35 |
| sdague | jeblair: rhel6 kernel is 2.6.32 | 18:36 |
| jeblair | sdague: yes, though it's more complicated than that because they backport ALL KINDS of stuff in their kernels | 18:36 |
| apevec | there's netns patches rhel6 kernel provided in RDO | 18:36 |
| sdague | in 6.4 they backported openvswitch into it (which didn't exist in that kernel), but network namespaces is still not in the main kernel | 18:36 |
| apevec | s/patches/patched/ | 18:36 |
| jeblair | so can we use an rdo kernel? | 18:37 |
| sdague | jeblair: so I think we'd be tripping over distro issues way more if we went down that path | 18:37 |
| sdague | jeblair: ok, so then what are we testing? nothing any developer is using :) | 18:37 |
| jeblair | sdague: what do you mean? | 18:37 |
| *** mrmartin has joined #openstack-infra | 18:38 | |
| mrmartin | re | 18:38 |
| jeblair | sdague: presumably something someone is deploying, since it's what the tc decided we support, presumably with reasons. | 18:38 |
| sdague | jeblair: we said we would try not to break it | 18:38 |
| jeblair | i don't think we tried very hard | 18:39 |
| sdague | jeblair: well when you can't use openvswitch namespaces | 18:39 |
| sdague | because the distro doesn't support it | 18:39 |
| sdague | you are kind of done | 18:39 |
| apevec | BTW, how come I got quantum-2013.1.3.tar.gz while snaps were neutron-stable-grizzly.tar.gz ? http://tarballs.openstack.org/neutron/?C=M;O=D | 18:39 |
| *** pcrews has quit IRC | 18:39 | |
| sdague | jeblair: and, honestly it doesn't solve the problem I actually want to solve, which is whether devstack works on fedora | 18:40 |
| sdague | which we've got 4 people a day popping into channels asking about | 18:40 |
| sdague | because right now it doesn't | 18:40 |
| sdague | because we broke it | 18:41 |
| sdague | because we don't run it in the gate | 18:41 |
| jeblair | sdague: i see the value of testing master on fedora. | 18:41 |
| sdague | and fixing it on centos 6 wouldn't mean it's fixed on fedora, the drift is too far | 18:41 |
| sdague | jeblair: ok, so that's actually all I really want to fix, having master on fedora. That should give us other options down the road as well | 18:42 |
| jeblair | sdague, mordred: so the last time we talked about this, i think we all acknowledged it is a really hard problem, and we did a lot of work and it took a design summit session to hash it out. | 18:42 |
| sdague | jeblair: so it would clearly be dependent on a volunteer from the redhat side to get us through the hard work | 18:42 |
| jeblair | sdague, mordred: i think it's worth exploring how we can test on fedora, but i also don't want to throw out the extensively considered position that we came to. | 18:43 |
| jeblair | sdague, mordred: so maybe hashing out long-term support plans is premature at this point | 18:43 |
| sdague | however, I wanted to at least pave the way to make sure there is nothing preventing them from making progress on it | 18:43 |
| *** pabelanger has quit IRC | 18:43 | |
| jeblair | sdague, mordred: and we just say that the low-hanging fruit of just getting one devstack run on fedora for master running is a good short/medium term goal. | 18:44 |
| sdague | jeblair: I'm happy with that | 18:44 |
| jeblair | i mean, it's not really low-hanging, but it's the lowest of what we've got | 18:44 |
| sdague | yep | 18:44 |
| sdague | it would give us options though | 18:44 |
| sdague | to explore from there | 18:44 |
| jeblair | because there's a huge puppet issue to deal with; one that we weren't expecting to work on this cycle | 18:44 |
| sdague | jeblair: agreed | 18:44 |
| jeblair | cool | 18:45 |
| jeblair | lunchtime now | 18:45 |
| sdague | it's just we, as a project, have moved from canonical being dominant distro and doing lots of work, to red hat doing a ton of work | 18:45 |
| sdague | and us breaking fedora by accident is becoming an uncool thing | 18:45 |
| sdague | with all of them trying to contribute | 18:45 |
| jeblair | sdague: yep | 18:46 |
| fungi | so, from a fedora 18 standpoint, i was able to get a basic jenkins slave working with our current puppet infrastructure with the exceptions of some supposedly benign warnings from puppet agent runs and an inability to generate puppet reports into the dashboard. i think neither of those is a concern for devstack slaves (we don't repeatedly run the puppet agent from cron and we don't report from them) | 18:46 |
| fungi | so at this moment, we can spin up fedora 18 machines with those characteristics/minor drawbacks | 18:47 |
| *** pabelanger has joined #openstack-infra | 18:47 | |
| jeblair | fungi: we might be able to live with that | 18:48 |
| *** psedlak has quit IRC | 18:48 | |
| fungi | as to devstack-gate and devstack work needed on top of that, i don't know the answer at the moment | 18:48 |
| fungi | and of course, fedora 19 being just around the corner (?) i don't know what new hurdles that is likely to bring | 18:49 |
| fungi | or i guess it's out now? 2013-07-02 according to http://fedoraproject.org/wiki/Releases/19/Schedule | 18:50 |
| mordred | fungi: git log --color=auto --decorate oneline pyghmi --not remotes/gerrit/master | 18:50 |
| mordred | fungi: yeah - 19 is out | 18:50 |
| mordred | fungi: that line above was posted by someone in #openstack-ironic | 18:50 |
| mordred | which seems ... oh - --decorate oneline | 18:50 |
| mordred | I see it | 18:50 |
| mordred | nevermind | 18:50 |
| mordred | nope. mind again | 18:51 |
| mordred | fungi: git log --color=auto --decorate oneline | 18:51 |
| mordred | fatal: ambiguous argument 'oneline': unknown revision or path not in the working tree. | 18:51 |
| fungi | there is a --oneline option to log | 18:52 |
| fungi | same as --pretty=oneline | 18:53 |
| mordred | yes | 18:53 |
| mordred | but above it's missing a -- | 18:53 |
| mordred | I see the -- in trunk and in the versoin installed in my machine | 18:53 |
| *** pcrews has joined #openstack-infra | 18:54 | |
| fungi | no clue what pyghmi is | 18:54 |
| fungi | mord trunk of and version of what installed on your machine? | 18:55 |
| fungi | mordred: ^ | 18:55 |
| fungi | ahh, you're talking about git-review | 18:55 |
| fungi | i see a similar command line in there, yeah | 18:56 |
| sdague | fungi: f19 is out | 18:56 |
| mrmartin | jeblair: do you have a few minutes to move forward the community portal project? | 18:56 |
| mordred | mrmartin: he's at lunch | 18:57 |
| mrmartin | oh really, lunch time | 18:57 |
| mrmartin | the sun just gone down here | 18:58 |
| *** zyluo has quit IRC | 18:59 | |
| sdague | oh man, finally done with meetings. so mordred the behavior we should have in update.py is to take dev-requirements.txt into account, and allow those as optionals? | 19:00 |
| *** thomasm has joined #openstack-infra | 19:01 | |
| *** melwitt has joined #openstack-infra | 19:02 | |
| mordred | sdague: yes. I think so - if the line matches dev-req, we shoudl not replace it | 19:02 |
| *** ruhe_ has joined #openstack-infra | 19:02 | |
| mordred | sdague: however, here's teh real kicker - how do we deal with updates to dev-req, which would necessarily mean that everyone would not match it, and then probably get repalced with the old value? | 19:03 |
| *** _TheDodd_ has quit IRC | 19:04 | |
| *** dina_belova has joined #openstack-infra | 19:04 | |
| *** sarob has joined #openstack-infra | 19:06 | |
| *** ^d is now known as ^demon|lunch | 19:07 | |
| *** sarob has quit IRC | 19:07 | |
| *** sarob has joined #openstack-infra | 19:08 | |
| *** thomasbiege has quit IRC | 19:10 | |
| *** _TheDodd_ has joined #openstack-infra | 19:11 | |
| *** sarob has quit IRC | 19:12 | |
| *** ruhe_ has quit IRC | 19:14 | |
| sdague | mordred: hmmmmmm | 19:16 |
| sdague | how good is our version parsing? | 19:16 |
| sdague | let me go play with code for a bit | 19:16 |
| sdague | unrelated question | 19:17 |
| *** pcrews has quit IRC | 19:17 | |
| sdague | to get on stackforge does the project need to sign up for the CLA and be Apache2? | 19:17 |
| lifeless | jog0: mmm, so I understand the mechanics you're proposing but not the motivation | 19:19 |
| jog0 | lifeless: because every few weeks someone tries to modify .gitignore for yet another editor | 19:20 |
| lifeless | jog0: the only reasons I can imagine not wanting a particular file type in the project ignore is: a) makes git slow or b) will inappropriately ignore something someone would want to see in 'git status' | 19:20 |
| jog0 | lifeless: it also makes the .gitignore files big enough that people don't really think if something should be there or not | 19:20 |
| lifeless | jog0: there's a limited set of editors, we could just bulk add them all. And - whats the issue, just say yes, if it makes the life of your contributors easier ? | 19:20 |
| lifeless | jog0: that comes back to b),yes? | 19:21 |
| jog0 | for example should 'keeper' be in nova gitignore? | 19:21 |
| clarkb | sdague: no, I dont think we have solid rulea but I think open source is sufficient | 19:22 |
| clarkb | jeblair ^ | 19:22 |
| *** ruhe has joined #openstack-infra | 19:22 | |
| jog0 | also it just keeps things cleaner and prevents the issue of having to update gitignore for every project all the time | 19:23 |
| *** dina_belova has quit IRC | 19:23 | |
| lifeless | jog0: I simply don't understand the issue | 19:23 |
| lifeless | jog0: someone *else* proposes the update, 2 cores agree it makes sense for .gitignore, done. | 19:23 |
| mordred | sdague: no. it does not | 19:23 |
| mordred | open source | 19:23 |
| lifeless | jog0: maybe 60 seconds of thought from them both, and move on. | 19:24 |
| *** pcrews has joined #openstack-infra | 19:24 | |
| lifeless | jog0: nova's .gitignore has changed 24 times. | 19:24 |
| lifeless | jog0: nova has > 20K commits | 19:25 |
| lifeless | jog0: 1 in 1000 are .gitignore : how is that an issue ? | 19:26 |
| *** adalbas has quit IRC | 19:29 | |
| *** pcrews has quit IRC | 19:32 | |
| *** mrodden1 has joined #openstack-infra | 19:35 | |
| *** mrodden has quit IRC | 19:36 | |
| *** sarob has joined #openstack-infra | 19:36 | |
| jog0 | lifeless: it comes down to it seems like a non-optimal way to do it | 19:37 |
| jog0 | I am not going to -1 the patch, its not worth that | 19:37 |
| *** dina_belova has joined #openstack-infra | 19:37 | |
| jog0 | lifeless: so if you get to other cores to +2 it I am fine with it | 19:38 |
| lifeless | ok | 19:38 |
| lifeless | thanks | 19:38 |
| jog0 | lifeless: I was worried that I -1ed it for a second. | 19:38 |
| jeblair | mrmartin: hi there | 19:38 |
| *** sarob has quit IRC | 19:39 | |
| *** _TheDodd_ has quit IRC | 19:39 | |
| jeblair | mrmartin: how can i help? would you like one of us to make the initial commit that sets up the server class in puppet, then you can work on it from there? | 19:39 |
| *** sarob has joined #openstack-infra | 19:39 | |
| lifeless | jog0: I thought you had | 19:39 |
| lifeless | jog0: :) | 19:39 |
| *** nayward has quit IRC | 19:40 | |
| jog0 | lifeless: so did I | 19:40 |
| jog0 | from the way you were asking | 19:40 |
| lifeless | jog0: sorry for the stress! | 19:40 |
| *** dina_belova has quit IRC | 19:41 | |
| * fungi should have popped some popcorn for that one | 19:43 | |
| *** sarob has quit IRC | 19:43 | |
| *** adalbas has joined #openstack-infra | 19:44 | |
| jog0 | lifeless: haha it keeps me on my toes | 19:44 |
| fungi | clarkb: if you get a moment to glance at https://review.openstack.org/38176 (not sure if you're near a big enough screen) can you tell me whether i'm making sense? | 19:45 |
| *** vipul is now known as vipul-away | 19:46 | |
| *** gyee has joined #openstack-infra | 19:47 | |
| *** changbl has quit IRC | 19:48 | |
| *** _TheDodd_ has joined #openstack-infra | 19:48 | |
| *** changbl has joined #openstack-infra | 19:50 | |
| clarkb | fungi yeah jjb should treat that as a string literal I think | 19:50 |
| openstackgerrit | Sean Dague proposed a change to openstack/requirements: use dev-requirements for global sync https://review.openstack.org/40947 | 19:51 |
| clarkb | fungi but a subshell should work too | 19:52 |
| sdague | mordred: ok, that solves the neutron problem | 19:53 |
| sdague | however, there is a more general problem there | 19:53 |
| *** annegentle_vacay has quit IRC | 19:53 | |
| mordred | sdague: yeah? | 19:53 |
| sdague | mordred: the 1.2.0b5 | 19:53 |
| fungi | clarkb: i wonder how deep the rabbit hole goes if i try to fix the parser to not error on that during the comparison tests (since it still seems to be able to parse it under normal production circustances) | 19:54 |
| sdague | so what I did is just say if you are exactly in dev-requires, you can use it | 19:54 |
| sdague | otherwise we set to g-r | 19:54 |
| mordred | well - so maybe it's not a problem | 19:54 |
| sdague | see what you think | 19:54 |
| mordred | if you aren't matching, devstack is going to make you match | 19:55 |
| mordred | for devstack | 19:55 |
| sdague | yeh, it actually means we reintroduce skew, in theory in devstack | 19:55 |
| mordred | then requirements update will generate a commit that will reset you back down to the normal | 19:55 |
| sdague | however, in practice, we don't | 19:55 |
| sdague | because it's olso | 19:55 |
| sdague | oslo | 19:55 |
| clarkb | fungi idunno. may need to pdb again | 19:55 |
| sdague | which we're using git on anyway | 19:55 |
| mordred | exactly | 19:55 |
| sdague | mordred: but this was actually a unit test problem | 19:55 |
| mordred | right | 19:56 |
| mordred | the unittests will then fail on the proposal to bump you back down | 19:56 |
| sdague | so this will let them go through | 19:56 |
| mordred | but then a human can amend that change to set it to the new g-r version | 19:56 |
| sdague | yep | 19:56 |
| sdague | so we can navigate with these changes | 19:56 |
| mordred | yah. and eventually we'll get wheels up | 19:56 |
| sdague | it's just a little more manual when someone bumps dev-req | 19:56 |
| mordred | how abou t... | 19:57 |
| sdague | but it won't ruin anyone | 19:57 |
| mordred | we don't do automated requirements proposal jobs | 19:57 |
| mordred | until we get rid of d-r? | 19:57 |
| mordred | by publishing pre-release wheels | 19:57 |
| sdague | mordred: yep, sounds fine to me | 19:57 |
| mordred | cool. it's a plan | 19:57 |
| lifeless | mordred: dunno if you saw my run-mirror timings :) | 19:58 |
| mordred | lifeless: I did not - did you do one with wheels? | 19:58 |
| lifeless | mordred: no. But - having done a full mirror, an incremental run was 40m | 19:59 |
| lifeless | real 39m4.291s | 19:59 |
| lifeless | user 6m33.628s | 19:59 |
| lifeless | sys 0m46.968s | 19:59 |
| *** sandywalsh has quit IRC | 19:59 | |
| mordred | lifeless: wait- it took you 40m to run run-mirror? | 19:59 |
| lifeless | mordred: yes. | 19:59 |
| lifeless | mordred: latency. NZ. HTTPS spidering. | 20:00 |
| mordred | wow | 20:00 |
| mordred | I thnk it takes us 8 | 20:00 |
| lifeless | yes | 20:00 |
| lifeless | This is why I was telling you it was going to be slow. | 20:00 |
| lifeless | Because, well, it is. | 20:00 |
| lifeless | Like you know Mysql innards, I know replication and HTTP :) | 20:00 |
| lifeless | dstufft has a recipe to avoid the external links except for specific packages | 20:01 |
| lifeless | which I'll play around with shortly | 20:01 |
| lifeless | but I don't expect it to 'fix' it to be fast enough to be inline with a build (but rsync will be) | 20:01 |
| *** vipul-away is now known as vipul | 20:02 | |
| *** mrmartin has quit IRC | 20:02 | |
| sdague | dtroyer: you about? | 20:05 |
| *** pabelanger has quit IRC | 20:06 | |
| sdague | wanted to get to the bottom of the devstack fedora fix, and if we have one that everyone can like | 20:06 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/devstack-gate: WIP: new nodepool https://review.openstack.org/40952 | 20:09 |
| jeblair | mordred, clarkb, fungi: do (some of) you have a minute to take a look at https://review.openstack.org/40952 ? | 20:10 |
| jeblair | it's a skeleton for the new nodepool code | 20:10 |
| *** emagana has quit IRC | 20:10 | |
| jeblair | and i want to double check the approach | 20:10 |
| mordred | jeblair: looking | 20:10 |
| *** sandywalsh has joined #openstack-infra | 20:10 | |
| *** emagana has joined #openstack-infra | 20:11 | |
| jeblair | (though it basically works without actually touching any servers or images at this point; that code is fairly well established, so next up is basically plugging the old stuff into the new) | 20:11 |
| fungi | i could stand a break from trying to wrap my head around jjb's parameter replacement recursion, sure | 20:12 |
| *** locke105 has quit IRC | 20:13 | |
| jeblair | NodePool is the interesting class; it has a gearman worker (NodeUpdateWorker) that responds to gearman events that i expect zuul to send when a job starts or stops on a devstack (or any) node. | 20:14 |
| jeblair | when the worker gets a complete event, it spawns a thread to cleanup (currently: delete) the server | 20:14 |
| fungi | oh, 40952 reminds me that i meant to look into why hold status didn't work for me the last time i tried it (since some of the recent refactoring), but i guess this will obviate that anyway | 20:14 |
| sdague | so anyone else want to +2 this - https://review.openstack.org/#/c/39967/ | 20:15 |
| jeblair | fungi: yeah, in this, hold should hold forever | 20:15 |
| sdague | it's blocking checks for people, as they don't pass pep8 | 20:15 |
| fungi | i *think* currently one of the update scripts isn't checking for hold and is resetting it to delete | 20:15 |
| jeblair | the nodepool's main thread runs in a continous loop that currently does what the launch job does (calculate deficit and launch nodes). but it spawns a thread for every machine it launches | 20:16 |
| jeblair | then there are two functions that get called periodically; one to update images, and then one to delete anything that we might have missed somehow (and also delete old images). | 20:17 |
| *** mrodden has joined #openstack-infra | 20:17 | |
| jeblair | both of those correspond pretty closely to the update image and reap jobs, except this reap function should only need to run every few hours, since most deletions will be handled by the gearman worker | 20:17 |
| fungi | sdague: that checks out, so approved | 20:18 |
| jeblair | a possible problem is that if there are a lot of threads running at once, we might hit rate limit errors with openstack; but i think the best approach is to just play the odds there and retry (with sleeps) if needed. | 20:18 |
| *** locke105 has joined #openstack-infra | 20:19 | |
| fungi | jeblair: so the per-launch threads only stick around until the machine is into a ready state, or persist until machine deletion? | 20:19 |
| jeblair | fungi: until it's ready, then they exit | 20:19 |
| fungi | k | 20:19 |
| jeblair | fungi: then new per-delete threads are spawned by the gearman worker when it gets notified a job is finished | 20:20 |
| jeblair | oh, and of course, this presupposes the new gearman-plugin disable-on-complete feature is being used to protect against duplicate runs | 20:20 |
| fungi | okay, so the thread persists from the time the task is initiated until it gets a completion response? | 20:20 |
| *** mrodden1 has quit IRC | 20:20 | |
| sdague | fungi: you want to tackle the one of top of it as well | 20:21 |
| fungi | sdague: sure, why not | 20:21 |
| jeblair | fungi: yeah, it'll poll continuously. we can probably give it, say, a 30 minute insane-cloud-provider timeout too. | 20:21 |
| openstackgerrit | Sergey Lukjanov proposed a change to openstack-infra/reviewstats: Add savanna to projects https://review.openstack.org/40953 | 20:21 |
| *** vipul is now known as vipul-away | 20:22 | |
| fungi | jeblair: ahh, makes sense. i was figuring we'd need an escape hatch there because, well, clouds | 20:22 |
| *** dkranz has quit IRC | 20:23 | |
| fungi | but still, too many insane cloud provider incidents and we could see a huge pile of threads | 20:24 |
| mordred | jeblair: generally makes sense | 20:24 |
| mordred | jeblair: also, I like that it is modular enough that I can imagine where a dib workflow would plug in | 20:24 |
| jeblair | mordred: excellent | 20:24 |
| fungi | or i guess the pool limit will keep the thread count from getting too high, i guess we just might wind up with no machines because none are actually launching. and that's already happened before so not exactly a regression | 20:25 |
| apevec | interesting, how did that happen? change-id hash collision https://review.openstack.org/#/q/I130528ae946e8888d35c25e468b4ea6ac29db0cf,n,z | 20:25 |
| apevec | exactly in the example I have in https://wiki.openstack.org/wiki/StableBranchRelease#Rinse.2C_Repeat | 20:25 |
| jeblair | apevec: i wonder where Dirk got his change-id from | 20:27 |
| jeblair | perhaps he copied it from the other change. or perhaps he should buy a lottery ticket. | 20:27 |
| apevec | yep :) | 20:27 |
| fungi | yeah, 31377 came later and very well may have been copied from 28741 (the milestone bump on the same project) | 20:28 |
| apevec | but usually you just let hook generate it | 20:28 |
| apevec | why would you copy it | 20:28 |
| fungi | it would be faaar from the first incident of commit message copy/paste induced errors i've seen involving change-id headers | 20:28 |
| sdague | dtroyer: how are you feeling about landing your devstack fix for fedora, and doing the other cleanups later? | 20:29 |
| apevec | fungi, bump was on stable branch, this commit is on master | 20:29 |
| jeblair | mordred, fungi: ok, i'm going to continue to hack in that general direction | 20:29 |
| fungi | apevec: the why doesn't matter so much. i've seen it from poorly edited squashes, backports/forwardports or anything which might include old commit message data. also simply possible gerrit uses a very bad prng | 20:30 |
| apevec | yeah, whatever | 20:30 |
| fungi | jeblair: definitely. still reading through but looks good so far | 20:31 |
| apevec | I'll change example in wiki :) | 20:31 |
| fungi | apevec: oh! you did say it was in the wiki | 20:31 |
| apevec | heh, copying from wiki would be even stranger | 20:31 |
| fungi | apevec: he almost certainly copied it from the wiki when he tried to do something without having a commit hook installed, gerrit told him he needed a change id, and so he set out to look for one ;) | 20:31 |
| jeblair | Ahhh | 20:32 |
| dtroyer | sdague: [back from lunch] I'm ready for the install_pip.sh patch to go in. it isn't complete but unblocks the next level of stuff | 20:32 |
| sdague | cool, I'm good with it especially given how many folks it's blocking | 20:32 |
| *** ^demon|lunch is now known as ^d | 20:33 | |
| sdague | dtroyer: however, I realized there is actually a problem with mordred's requirements update method | 20:33 |
| sdague | which is reclone on requirements repo | 20:33 |
| sdague | and those updates won't take in the trees | 20:33 |
| sdague | because they are changed | 20:33 |
| dtroyer | I think there are other problems too but am stuck on something else at the moment to chase it down. that's one reason I wanted a flag | 20:34 |
| mordred | sdague: gah | 20:34 |
| sdague | dtroyer: yeh | 20:34 |
| sdague | I'm wondering if what we actually want to do is have update.py take a flag to update to another place | 20:34 |
| sdague | then install that file | 20:34 |
| sdague | then we aren't changing git repo files | 20:35 |
| dtroyer | i think that would solve what I saw too | 20:36 |
| sdague | ok, I'll try to tackle that tomorrow | 20:36 |
| *** sarob has joined #openstack-infra | 20:39 | |
| *** ruhe has quit IRC | 20:41 | |
| *** dina_belova has joined #openstack-infra | 20:42 | |
| *** vipul-away is now known as vipul | 20:43 | |
| *** dina_belova has quit IRC | 20:46 | |
| *** pcrews has joined #openstack-infra | 20:47 | |
| *** woodspa has quit IRC | 20:47 | |
| *** dina_belova has joined #openstack-infra | 20:52 | |
| *** jpich has quit IRC | 20:52 | |
| *** SergeyLukjanov has quit IRC | 20:54 | |
| *** SergeyLukjanov has joined #openstack-infra | 20:54 | |
| *** dina_belova has quit IRC | 20:56 | |
| *** thomasm has quit IRC | 20:58 | |
| *** ArxCruz has quit IRC | 20:58 | |
| *** svarnau has quit IRC | 21:01 | |
| *** sarob has quit IRC | 21:01 | |
| *** sarob has joined #openstack-infra | 21:02 | |
| *** rfolco has quit IRC | 21:06 | |
| *** sarob has quit IRC | 21:07 | |
| *** apevec has quit IRC | 21:09 | |
| openstackgerrit | A change was merged to openstack/requirements: Add pep8 checks https://review.openstack.org/39967 | 21:11 |
| *** krtaylor has joined #openstack-infra | 21:13 | |
| openstackgerrit | A change was merged to openstack/requirements: Update ourselves to ourselves https://review.openstack.org/39963 | 21:16 |
| openstackgerrit | A change was merged to openstack/requirements: Remove unneeded tests dir reference https://review.openstack.org/40002 | 21:16 |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/jenkins-job-builder: Handle KeyError in deep_format() https://review.openstack.org/40973 | 21:20 |
| *** cody-somerville has quit IRC | 21:20 | |
| *** cody-somerville has joined #openstack-infra | 21:21 | |
| bodepd | jeblair: you have a sec? | 21:22 |
| bodepd | or anyone can answer: I have some questions about what will happen when: https://review.openstack.org/#/c/37715 gets merged | 21:22 |
| fungi | bodepd: there will be a coordinated gerrit outage to actually perform the rename during a low-activity period | 21:24 |
| fungi | bodepd: http://ci.openstack.org/gerrit.html#renaming-a-project | 21:24 |
| bodepd | fungi: ok. what about the old project in github? it will still exist with redirects? | 21:24 |
| fungi | that's the general process we follow | 21:25 |
| bodepd | fungi: thanks! | 21:25 |
| *** sarob has joined #openstack-infra | 21:25 | |
| fungi | developers consuming the old remote names should be urged to update them | 21:25 |
| fungi | github doesn't provide project redirects afaik | 21:25 |
| fungi | we usually rename the project directly in github | 21:26 |
| bodepd | that is what I am trying to figure out. | 21:26 |
| fungi | and then we tell everyone to use the new name | 21:26 |
| bodepd | I just merged quantum->neutron | 21:27 |
| bodepd | but now,all tests will be broken | 21:27 |
| fungi | usually at least some subset of users are taken by surprise when it happens, don't subscribe to relevant mailing lists, whatever | 21:27 |
| bodepd | b/c puppet can't load modules from teh wrong project | 21:27 |
| bodepd | so, I'm in an inconsistent broken state | 21:27 |
| bodepd | as of 1 minute ago (or at least I'm pretty sure I am) | 21:28 |
| fungi | bodepd: mmm, so tests passed on the change itself, but the change breaks future test runs once merged? that seems like a missing test | 21:28 |
| *** dansmith has quit IRC | 21:28 | |
| fungi | one exit strategy there is to have a gerrit admin manually merge a revert commit for that change | 21:29 |
| fungi | and then come up with additional tests to prevent similar breakage in the future | 21:29 |
| *** vipul is now known as vipul-away | 21:31 | |
| jgriffith | anybody else encountering funny things trying to install test-requirements? | 21:31 |
| jgriffith | ie: http://paste.openstack.org/show/43630/ | 21:31 |
| fungi | did this just suddenly start? is that timestamp on line #2 utc or mountain time? | 21:32 |
| jgriffith | fungi: mountain | 21:32 |
| bodepd | fungi: the problem is that puppet-quantum is actually broken atm | 21:32 |
| jgriffith | fungi: I've been having weirdness last night and today | 21:33 |
| fungi | so just a minute ago then | 21:33 |
| bodepd | fungi: b/c all of the class names are nuetron::x, so they cannot be loaded | 21:33 |
| jgriffith | fungi: correct | 21:33 |
| jgriffith | oddly latest devstack attempt seemed to come up | 21:33 |
| bodepd | fungi: https://github.com/blog/1508-repository-redirects-are-here | 21:33 |
| fungi | ooh! | 21:33 |
| jgriffith | however there I'm seeing a bunch of failures tryign unit tests | 21:33 |
| jgriffith | due to subunit parser FAIL | 21:33 |
| jgriffith | ??? | 21:33 |
| fungi | the ooh was for github project redirects finally existing | 21:34 |
| jgriffith | and I tried copying the *new* requriements files that devstack mods over as well | 21:34 |
| *** dansmith_ has joined #openstack-infra | 21:34 | |
| jgriffith | darn :) | 21:34 |
| fungi | jgriffith: i think ekarlso- pasted something similar this morning... looking back in my scrollback | 21:35 |
| bodepd | fungi: can we rename the repo by transfering? | 21:35 |
| bodepd | fungi: so that it will rediret? | 21:35 |
| *** dansmith_ is now known as dansmith | 21:35 | |
| bodepd | fungi: that would make my life pretty simple today :) | 21:35 |
| fungi | jgriffith: http://pastebin.com/ZeLWUq2c | 21:35 |
| bodepd | fungi: just reviewed that doc. that is the way you already do it :) | 21:36 |
| fungi | bodepd: well, we still need to schedule an outage for gerrit to do the renaming | 21:36 |
| bodepd | fungi: can we just rename it in github first? | 21:36 |
| *** nayward has joined #openstack-infra | 21:36 | |
| jgriffith | fungi: interesting... so that might be similar to what I get when trying to pull in a review and run_tests.sh on it | 21:36 |
| bodepd | (this shoudl work if the redirect works) | 21:36 |
| fungi | bodepd: we will continue replicating to the old remote name | 21:37 |
| bodepd | which is redirected to the new one | 21:37 |
| bodepd | fungi: ^ | 21:37 |
| fungi | bodepd: also i think anyone trying to commit changes to gerrit will get rejections if they try to use the new name | 21:37 |
| bodepd | fungi: we will update the gitreview config file last | 21:38 |
| bodepd | and time that with the outage | 21:38 |
| fungi | it's worth trying... | 21:38 |
| jgriffith | fungi: hmmm... nova has the same problem | 21:38 |
| bodepd | fungi: I'm just spitballing here,so that I can get tests passing | 21:38 |
| bodepd | fungi: and give you guys time to prepare for the outage | 21:38 |
| jgriffith | ok, I'll poke a bit, see if all the tests fail in Nova as well like they do in Cinder | 21:38 |
| bodepd | fungi: and even if it breaks, everything is broken anyways | 21:38 |
| fungi | jeblair: mordred: clarkb: thoughts on whether a github project rename (given the new redirect functionality mentioned in the https://github.com/blog/1508-repository-redirects-are-here announcement) would present any immediate issues? | 21:39 |
| *** annegentle has joined #openstack-infra | 21:40 | |
| *** nayward has quit IRC | 21:40 | |
| bodepd | fungi: let me know. I'm about to send out the 'everything may explode' email | 21:43 |
| jeblair | fungi: not immediately | 21:44 |
| fungi | okay, wanting to get a little consensus before i go potentially making things harder on us because i didn't think something through | 21:44 |
| bodepd | fungi: I agree. I'm pretty sure it's going to work, but I also just put a cowboy hat on | 21:44 |
| fungi | jeblair: any concerns with me renaming stackforge/puppet-quantum to stackforge/puppet-neutron on github now, in advance of a later outage to rename it on gerrit at some convenient date and time? | 21:45 |
| fungi | the way i read that, github should continue to see our mirror updates to -quantum and reroute them to -neutron accordingly | 21:46 |
| clarkb | fungi: we don't use pages anymore and everything else should just work (tm) | 21:46 |
| fungi | seems to be the theory at any rate | 21:47 |
| *** beagles is now known as beagles_biab | 21:48 | |
| *** dina_belova has joined #openstack-infra | 21:52 | |
| *** markmcclain has quit IRC | 21:54 | |
| *** dina_belova has quit IRC | 21:57 | |
| jeblair | fungi: oh, erm, i'd rather not do that | 21:58 |
| jeblair | (a) confusion (b) manage_projects uses the github api; i'd rather not test that in production | 21:59 |
| jeblair | also, i don't think we've even scheduled the gerrit rename yet? | 21:59 |
| fungi | afaik we have not, no | 21:59 |
| *** SergeyLukjanov has quit IRC | 21:59 | |
| openstackgerrit | gordon chung proposed a change to openstack/requirements: add pycadf library https://review.openstack.org/40926 | 21:59 |
| jeblair | fungi: i thought you meant doing the rename during the gerrit downtime | 22:00 |
| fungi | bodepd was hoping this would get him out of a scrape, but maybe the easier path forward is to help him revert changes or turn his tests non-voting long enough to revert commits himself | 22:00 |
| jeblair | fungi: yeah, if something goes wrong with manage projects, we could have a _big_ cleanup on our hands. i'd recommend those other ideas. | 22:01 |
| *** vipul-away is now known as vipul | 22:01 | |
| bodepd | jeblair: an even crazier question is could I have differnet branches point at different repos | 22:03 |
| bodepd | jeblair: fungi you guys must be doing this over and over again for neutron? | 22:03 |
| jeblair | bodepd: we did it once | 22:04 |
| fungi | bodepd: for the official repositories we took one giant outage to rename everything everywhere it touched ci/infra | 22:04 |
| fungi | and then let the individual projects referencing quantum clean things up internally after the fact | 22:04 |
| bodepd | are there not a ton of peripheral projects that still ahve to do the same thing? | 22:05 |
| fungi | there were transitional bootstrapping packages with a new release of the old name and other things to help that along | 22:05 |
| fungi | i think the only other stackforge project with "quantum" in its name besides yours may be the chef cookbook for quantum and it hasn't renamed yet that i've noticed | 22:06 |
| *** burt has quit IRC | 22:07 | |
| jeblair | fungi: it's 'networking' | 22:07 |
| fungi | oh, so it is | 22:08 |
| jeblair | they dodged that one i think. | 22:08 |
| fungi | indeed | 22:08 |
| clarkb | ya, which is why one of my suggestions when the puppet folk were looking at this was to switch to service name and not projects names | 22:09 |
| bodepd | yep | 22:09 |
| clarkb | dprince didn't like that iirc | 22:10 |
| clarkb | I think consistency may have been part of that decisions, it would be inconsistent or you ahve to do that same to all of the other projects | 22:10 |
| bodepd | is it the project name that determines what repo to clone to? | 22:12 |
| *** mriedem has quit IRC | 22:12 | |
| bodepd | how is openstack core handling the fact that stable/grizzly branches need to be pushed to quantum and not neutron? | 22:12 |
| clarkb | they are pushed to neutron | 22:12 |
| bodepd | that is the last piece of the puzzle that is confusing me | 22:12 |
| bodepd | I thought the whole point of stable branhces is not to break backwards compat | 22:13 |
| bodepd | that seems like a massive break | 22:13 |
| bodepd | (as in, your code is gone) | 22:13 |
| clarkb | iirc it is backward compat because the repo says neutron but hte code says quantum | 22:13 |
| clarkb | its just a name | 22:13 |
| bodepd | well, it's a location. | 22:14 |
| bodepd | so,any script would expect to download code (potentially) and then use it | 22:14 |
| bodepd | I guess it's way more problematic with puppet (or is it...) | 22:14 |
| clarkb | anyone dealing with source should expect to have these problems | 22:14 |
| clarkb | if you deal with releases nothing changes | 22:15 |
| bodepd | then why do you need branhces? | 22:15 |
| bodepd | if everyone can reliably deal with released versions for everything | 22:15 |
| bodepd | anyways,I'm not being productive, just arguing circular logic | 22:15 |
| *** markmcclain has joined #openstack-infra | 22:16 | |
| bodepd | fungi: I don't mind if the build/test machine is down for a little while, I just need to know when it can get back up. | 22:18 |
| *** changbl has quit IRC | 22:18 | |
| bodepd | jeblair: what about just merging this: https://review.openstack.org/#/c/37715 | 22:19 |
| bodepd | jeblair: and I will just fix everything when it blows up? | 22:19 |
| bodepd | jeblair: and the gerrit/github names will be out of sync until you guys have time to clean it up | 22:20 |
| clarkb | I am not arguing that everyone can use released versions and the branches are there to support released versions | 22:21 |
| clarkb | but source repos move | 22:21 |
| clarkb | either from bzr to git or from name/foo to name/bar | 22:21 |
| fungi | bodepd: some of that will tell manage-projects to create a new empty project called puppet-neutron, with everything that entails (local repo, github mirror, et cetera) | 22:22 |
| fungi | and then we won't be able to easily perform the rename itself | 22:22 |
| bodepd | fungi: and the verdict is no on a manual rename? | 22:22 |
| bodepd | fungi: without downtime/associated changes in gerrit | 22:23 |
| fungi | bodepd: the steps we normally follow involve ninja-coordinating the merger of that change with offlining gerrit to run a bunch of update queries on its database amd move things around on the filesystem | 22:23 |
| fungi | doing those tasks at different times wouldn't really work. you'd end up with divergence | 22:24 |
| bodepd | fungi: let me know when you guys could do it, so I can send a notice that the project is going to stop until then | 22:24 |
| mgagne | are you saying that we should have better planned the rename and sync with infra? =) | 22:24 |
| *** weshay has quit IRC | 22:24 | |
| fungi | with a proper move, the review history and open reviews get repointed to the new name, peoples watch lists in gerrit get updated, et cetera | 22:24 |
| bodepd | mgagne: yep. for some reason, I thought it had all been thought through | 22:25 |
| fungi | and we move/rename the remote clones at the same time, all before briging gerrit back online | 22:26 |
| fungi | it really doesn't handle having those sorts of things changed out from under it live | 22:26 |
| fungi | bodepd: i take it you can't just revert the commits which did the name change inside the puppet module until the project rename happens? | 22:27 |
| *** fbo is now known as fbo_away | 22:29 | |
| bodepd | fungi: I think we have a workaround to take the pressure off | 22:32 |
| *** ^d has quit IRC | 22:32 | |
| fungi | that's excellent news! | 22:33 |
| *** vijendar has quit IRC | 22:33 | |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/config: Gate jeepyb and openstack/requirements on pbr https://review.openstack.org/38176 | 22:35 |
| jeremyb | fungi: pbr? | 22:37 |
| fungi | i think that ^ should work around the validation false positive | 22:37 |
| fungi | until the jjb change can be groomed to fix the parser | 22:37 |
| fungi | jeremyb: what about pbr? | 22:38 |
| jeremyb | fungi: is that a beer or? | 22:39 |
| fungi | http://github.com/openstack-dev/pbr (python build reasonableness) | 22:40 |
| jeremyb | ok, notbeer | 22:40 |
| fungi | pbr is definitely not beer | 22:40 |
| jeremyb | :) | 22:40 |
| fungi | i've drank it before | 22:40 |
| fungi | definitely not beer | 22:40 |
| fungi | "yeast soda" is perhaps the best term i have for that class of beverage | 22:40 |
| jeremyb | fakebeer? :) | 22:40 |
| jeremyb | i've had it maybe twice. don't really remember it | 22:41 |
| *** afazekas has quit IRC | 22:41 | |
| clarkb | hey now, pbr and rainier are an important part of the NW beer market | 22:41 |
| jeremyb | i like natty bo | 22:41 |
| clarkb | oh no | 22:41 |
| jeremyb | natty boh* | 22:42 |
| *** adalbas has quit IRC | 22:43 | |
| *** afazekas has joined #openstack-infra | 22:44 | |
| *** sarob has quit IRC | 22:45 | |
| *** datsun180b has quit IRC | 22:45 | |
| *** sarob has joined #openstack-infra | 22:46 | |
| *** _TheDodd_ has quit IRC | 22:47 | |
| *** prad has quit IRC | 22:47 | |
| jeblair | gah | 22:48 |
| jeblair | clarkb, mordred: hpcloud has still not fixed the "can't have dots in a keypair name" bug | 22:48 |
| jeblair | so, uh, | 22:48 |
| jeblair | openstackjenkins2 now has a keypair we can't delete | 22:48 |
| *** sarob has quit IRC | 22:49 | |
| clarkb | :/ | 22:49 |
| jeblair | i'll open a support ticket to delete it | 22:49 |
| clarkb | jeblair: are you doing that with a CLI client or the web UI or both?. | 22:49 |
| fungi | ick | 22:49 |
| jeblair | clarkb: both | 22:50 |
| jeblair | case opened | 22:51 |
| * fungi assumes they rolled their own access key management there | 22:51 | |
| clarkb | iirc HP keystone is not keystone | 22:51 |
| jeblair | btw, i opened a support ticket loooong ago about the problem in general | 22:51 |
| openstackgerrit | Sean Dague proposed a change to openstack/requirements: use dev-requirements for global sync https://review.openstack.org/40947 | 22:53 |
| fungi | i love opening a ticket with a provider about "your system breaks when i try to do this" and it's promptly closed with "i went ahead and took care of it for you" | 22:53 |
| *** dina_belova has joined #openstack-infra | 22:53 | |
| jeblair | no kidding. there should be a support tracker _and_ a bug tracker. | 22:54 |
| *** rcleere has quit IRC | 22:54 | |
| *** vipul is now known as vipul-away | 22:55 | |
| *** vipul-away is now known as vipul | 22:55 | |
| *** dina_belova has quit IRC | 22:58 | |
| *** jrex_laptop has joined #openstack-infra | 23:00 | |
| fungi | that split tends to not work particularly well in a service provider portal because most customers don't know the difference and pick one at random, but it is definitely possible for the people triaging tickets to be able to identify the difference between a bug report and a support request and route them to the correct places | 23:03 |
| *** sarob has joined #openstack-infra | 23:07 | |
| *** pentameter has quit IRC | 23:08 | |
| lifeless | why does glanceclient depend on 'setuptools-git' ? | 23:09 |
| lifeless | it's breaking pbr - http://logs.openstack.org/69/40869/1/check/gate-pbr-devstack-vm-rawinstall/d79baff/console.html.gz | 23:09 |
| lifeless | jeblair: fungi: agreed that the split doesn't work well - in fact I think its artificial and driven by the poor capabilities of most? all? bug trackers. | 23:10 |
| lifeless | there's a continuum and any single report should be able to segue between | 23:10 |
| fungi | i'm willing to bet mordred already has a change molding in review to remove setuptools-git dependencies remaining in any projects | 23:10 |
| clarkb | lifeless: looks like latest released glanceclient needs setuptools_git but tip of master does not | 23:11 |
| clarkb | and vice versa for requirements on pbr | 23:11 |
| fungi | sounds like they should just release earlier and oftener | 23:11 |
| lifeless | we can't land changes in pbr at all until this is resolved.... | 23:12 |
| bodepd | just let me know when you guys are going to do that maintainance. It will require lots of code changes that need to be coordinated | 23:12 |
| lifeless | but I don't grok all the moving parts nor long term direction here yet, so I'm going to just go waaaaaah | 23:12 |
| lifeless | and point at mordred | 23:12 |
| *** rnirmal has quit IRC | 23:13 | |
| clarkb | lifeless: mordred is a good person to point at. My understanding is that glanceclient should release a new version, once thay have been updated to stop depending on d2to1 | 23:15 |
| lifeless | Is there an asymmetric gate thing going on here? | 23:16 |
| lifeless | Presumably this worked. How did it stop working ? | 23:16 |
| clarkb | lifeless: I don't think it is asymmetric gate, I think it is related to setuptools madness where external forces are breaking us | 23:16 |
| fungi | or possibly due to tests using released client libraries instead of vcs | 23:18 |
| clarkb | fungi: I think we flip flopped on that recently | 23:19 |
| clarkb | we now test with vcs | 23:19 |
| fungi | oh? | 23:19 |
| clarkb | yeah because sdague wants to test a known state (++) | 23:19 |
| fungi | i knew we wanted to, just didn't know if it had all gotten working yet | 23:20 |
| sdague | we're only testing with vcs in devstack gate | 23:21 |
| clarkb | which is what the pbr rawinstall should use | 23:21 |
| sdague | ok | 23:21 |
| sdague | I think we're 100% covered there... except which services does pbr turn on? | 23:21 |
| fungi | and presumably also want vcs dependencies in unit tests instead of our releases? | 23:21 |
| clarkb | it uses the d-g hooks to run its tests after d-g sets everything up | 23:21 |
| clarkb | sdague: I believe it isn't spinning up any services, just relying on the copying of source repos then testing that all of them install | 23:22 |
| *** pabelanger_ has quit IRC | 23:23 | |
| *** pabelanger has joined #openstack-infra | 23:23 | |
| fungi | oh, and if project a declares a requirement on project b, we're installing project a from source and project b gets satisfied from released packages i guess | 23:24 |
| clarkb | oh ya | 23:25 |
| clarkb | so order is important... | 23:25 |
| fungi | which means we need to bootstrap skynet into pbr so it can divine the entire dependency graph before running any of the install tests | 23:25 |
| sdague | or just trust dtroyer got it right :) | 23:26 |
| sdague | fortunately... no circ-deps yet | 23:26 |
| fungi | until the next time one of the projects decides to declare a new dependency on another of the projects again | 23:26 |
| sdague | nobody ever changes requirements... :) | 23:27 |
| sdague | http://openstackreactions.enovance.com/2013/08/when-someone-requests-a-feature-freeze-exception/ | 23:28 |
| notmyname | mordred: are ya bored? need something exciting to talk about? really just in a good mood and want it to end? I want to talk about swift and pbr https://review.openstack.org/#/c/28892/ | 23:28 |
| * sdague glares at mordred for turning on E125 on requirements repo..... | 23:29 | |
| lifeless | whats E125? | 23:29 |
| *** sdake_ has quit IRC | 23:29 | |
| clarkb | lifeless: alignment of line continuations in the if condition | 23:29 |
| mrodden | is it possible to get someone to nuke this from lodgeit? http://paste.openstack.org/show/43635/ | 23:29 |
| mrodden | has some host name info and such that i would rather not be available out there forever | 23:30 |
| sdague | it is my least favorite pep8 rule. Because it means you can't use emacs indenting. | 23:30 |
| clarkb | important to note it is purely a pep8 the tool rule and not a pep 8 the pep rule | 23:31 |
| mrodden | is that the one where you have to have 4 space indent instead of 8? | 23:31 |
| jeblair | i think it should be disabled in the requirements repo | 23:31 |
| sdague | mrodden: no it's the one that disallows | 23:31 |
| clarkb | pep 8 the pep has examples that contradict pep8 the tool in the pep 8 pep itself | 23:31 |
| sdague | if (foo | 23:31 |
| sdague | and bar): | 23:31 |
| sdague | do_it() | 23:32 |
| mrodden | oh | 23:32 |
| jeblair | mrodden: https://github.com/jcrocholl/pep8/issues/126 for some more light reading | 23:32 |
| dstufft | space indents in a if condition is bad | 23:32 |
| mrodden | jeblair: lol was just reading that | 23:32 |
| dstufft | 4 space* | 23:32 |
| notmyname | it's always so hard to script "when appropriate.." rules. maybe we should switch to coding in TeX and let Knuth handle our formatting | 23:32 |
| clarkb | mrodden: re deleting a paste, I can't do it currently but that data isn't secret imo | 23:33 |
| clarkb | there is an uber generic hostname with now domain | 23:33 |
| clarkb | and a path... | 23:33 |
| mrodden | yeah... i can deal with it | 23:33 |
| *** vipul is now known as vipul-away | 23:33 | |
| mrodden | its just internal CI shit i'm working on | 23:34 |
| jeblair | notmyname: that would {\it rock}. | 23:34 |
| clarkb | I think the pep8 rule the tool rule exists because the pep8 author must be a vim user. vim's python handling does that by default | 23:36 |
| *** changbl has joined #openstack-infra | 23:36 | |
| jeblair | clarkb: i have no idea what you just said | 23:36 |
| *** nayward has joined #openstack-infra | 23:37 | |
| clarkb | E125 exists because vim does that to your code | 23:37 |
| jeblair | dstufft: so i don't actually care. when i bring this up, people seem under the misapprehension that i actually have an opinion on this; i don't. | 23:37 |
| mrodden | clarkb is correct... | 23:37 |
| jeblair | dstufft: what i _do_ care about is that there is a standard so that different tools can follow it. | 23:37 |
| jog0 | ohh the E125 discussion | 23:38 |
| fungi | see, it's actually a cleverly disguised resurgence of the old vi vs emacs wars | 23:38 |
| jog0 | so looking for some reviews on https://review.openstack.org/#/q/status:open+project:openstack-infra/config+branch:master+topic:nomergealerts,n,z | 23:38 |
| dstufft | jeblair: :) | 23:38 |
| jeblair | dstufft: so i am not happy that the pep8 tool does something that contradicts the pep8 spec, and half the tools do one thing, and half the other. | 23:38 |
| dstufft | I don't use any tools tbh | 23:38 |
| dstufft | typically | 23:38 |
| jog0 | jeblairi like how the standard keeps changing, every so often I notice a update to pep8 itself | 23:38 |
| jeblair | dstufft: specs and standards are good. we should be clear in them and follow them. | 23:38 |
| lifeless | pep8 is a moving target | 23:39 |
| jeblair | lifeless: slowly moving | 23:39 |
| lifeless | jeblair: yes | 23:39 |
| lifeless | jeblair: but moving | 23:39 |
| lifeless | just blacklisted 'import *' for instance | 23:39 |
| jeblair | also, i'm okay with standards changing. otherwise, they'd be irrelevant over time. | 23:39 |
| dstufft | also not universally good | 23:39 |
| lifeless | which is weird, as thats a damn language feature. | 23:39 |
| * fungi is pretty good at using glaciers for target practice | 23:39 | |
| dstufft | :V | 23:39 |
| lifeless | jeblair: yes, things need updating; I wasn't whinging about pep8 evolving. | 23:40 |
| jeblair | one recent change to the pep8 standard was from barry, who made something clear that should have been obvious because all the python code ever did it, but the standard wasn't explicit, so the pep8 tool authors went and did their own thing again. | 23:40 |
| clarkb | dstufft: that is what we have # noqa for :) | 23:40 |
| mrodden | i'm so confused... i have nothing in pip or apt-get but this system thinks i have requests 1.2.3 in my site-packages.... | 23:40 |
| jog0 | so I thought pep was a python enhanchemnt proposal and I asusmed propsals don't change | 23:40 |
| *** fifieldt has joined #openstack-infra | 23:40 | |
| jog0 | after being accepted | 23:40 |
| dstufft | mrodden: lol | 23:40 |
| dstufft | jog0: depends on he type of PEP | 23:41 |
| dstufft | mrodden: there's cases where that can happen | 23:41 |
| openstackgerrit | Sean Dague proposed a change to openstack/requirements: add option for adding a suffix to output files https://review.openstack.org/40995 | 23:41 |
| *** nayward has quit IRC | 23:41 | |
| jog0 | dstufft: I know that now, ... | 23:41 |
| jeblair | dstufft: esp with the recent big changes to pep8, i think either versioning or superceding (rfc-style) might be worth considering | 23:41 |
| dstufft | PEPs have superceding built in | 23:42 |
| dstufft | or rather a mechanism for doing so | 23:42 |
| dstufft | I suspect the PEP8 canges wern't made a new PEP just because of the cultural uh, ca't think of the word, but everyone knows what "PEP8" is, it's the style guide | 23:42 |
| jog0 | so we have hijacked 'pep8' ourselves | 23:43 |
| jog0 | as our style guide tests are all called 'pep8' | 23:43 |
| jog0 | even thought that is a tiny part | 23:43 |
| clarkb | jog0: thats an implementation detail | 23:43 |
| dstufft | and strictly speaking PEP8 is only for the stdlib, but the community took it up and used it elsewhere, but very few projects that do use all of the rules | 23:44 |
| clarkb | jog0: which is easy to fix now that zuul doesn't need to stop everything to reload its config | 23:44 |
| clarkb | jog0: but previously applying that across the board coule have been quite painful | 23:44 |
| dstufft | so the term "pep8" is already super overloaded | 23:44 |
| *** vipul-away is now known as vipul | 23:44 | |
| jeblair | jog0: we can (and probably should/will) rename the jobs easily enough; the pep8 tool is more guilty of hijacking, i think. | 23:44 |
| jog0 | yeah i figured there is no rush to rename | 23:44 |
| sdague | mordred: https://review.openstack.org/40995 is the infrastructure to not overwriting the files in g-r updates | 23:44 |
| sdague | let me know what you think about it | 23:45 |
| jog0 | clarkb: got a question about the x-vrif-minus-2 stuff | 23:45 |
| clarkb | sure | 23:45 |
| jog0 | should I just remove the entire openstack-dev section | 23:46 |
| jog0 | or the events section | 23:46 |
| jeblair | jog0: (btw, i don't think that's worked in >1 year) | 23:46 |
| jgriffith | mordred: ping | 23:46 |
| clarkb | jog0: the entire openstack-dev section as removing the merge events basically noops that section | 23:46 |
| jog0 | clarkb: thanks | 23:46 |
| clarkb | jog0: as an alternative we could look into fixing the x-vrif-minus-2 stuff | 23:46 |
| openstackgerrit | Joe Gordon proposed a change to openstack-infra/config: Don't post merges in openstack-dev and openstack-nova https://review.openstack.org/40685 | 23:47 |
| jog0 | clarkb: should of said that a minute ago ... | 23:47 |
| jog0 | I don't think the -2 is very useful in the main room though | 23:47 |
| clarkb | jog0: agreed | 23:48 |
| clarkb | must better in the project specific rooms if we get it working again | 23:48 |
| jog0 | clarkb: yeah | 23:48 |
| mrodden | so why would stevedore importing an API plugin be finding conflicting requirements other than requirements.txt | 23:48 |
| jog0 | anyway thanks, time to find some food | 23:48 |
| fungi | i find it amusing (though understand) that the software devs are less interested in seeing merge notifications. in here that's the main thing i want notifications about from a systems dev perspective, because i want to know where to look first when things begin breaking | 23:49 |
| sdague | mrodden: because stevedore freezes requirements | 23:49 |
| sdague | and some other package probably installed a different version later | 23:49 |
| mrodden | so stevedore is the one with the requirement? | 23:49 |
| jog0 | fungi: the time between merge and deploy for dev is longer then in here Ithink | 23:50 |
| sdague | mrodden: no, entry points capture the requirements at setup.py time | 23:50 |
| mrodden | its complaining because something brought in requests==1.2.3 and something has a >=1.1,<1.2.3 | 23:50 |
| sdague | mrodden: right | 23:50 |
| jog0 | but yeah its funny how that works | 23:50 |
| sdague | this is actually why I've been getting us to global-requirements in the gate for the last 3 weeks :) | 23:50 |
| mrodden | so... | 23:51 |
| mrodden | i'm lost... | 23:51 |
| *** Ryan_Lane has joined #openstack-infra | 23:51 | |
| sdague | so before we started using entry points.... | 23:51 |
| lifeless | in the mists of time | 23:51 |
| sdague | we were installing and uninstalling libraries left right and center | 23:51 |
| mrodden | yep, get that part | 23:51 |
| sdague | we actually used to install / uninstall python-keystone client 4 - 6 times every devstack build | 23:51 |
| sdague | flipping versions | 23:52 |
| mrodden | ew | 23:52 |
| sdague | before entry points, if the code loaded and worked, all was good | 23:52 |
| *** jpeeler has quit IRC | 23:52 | |
| sdague | after entry points / stevedore, we encode metadata at setup.py time for each package | 23:52 |
| sdague | so if you install an incompatible version later, you'll get those kinds of fails to load | 23:52 |
| sdague | we were always doing it wrong | 23:53 |
| sdague | now when we doing it wrong, it hurts a lot more | 23:53 |
| mrodden | where did it get encoded? | 23:53 |
| *** zul has joined #openstack-infra | 23:53 | |
| mrodden | requirements.txt? | 23:53 |
| sdague | in the egg info I think | 23:53 |
| mrodden | ooo | 23:53 |
| *** dina_belova has joined #openstack-infra | 23:53 | |
| sdague | honestly the entrypoint / stevedore thing is a little fuzzy for me on mechanics | 23:54 |
| sdague | but the effects are clear :) | 23:54 |
| sdague | and the solution is to brute force everyone to use one list | 23:54 |
| sdague | which is what we've been doing in the gate for a week | 23:54 |
| sdague | and still polishing the solution | 23:54 |
| mrodden | ok... | 23:54 |
| mrodden | is there a way i can unscrew this up for now... | 23:55 |
| mrodden | everytime i tox --recreate i hit the same thing | 23:55 |
| sdague | how old is your devstack? | 23:55 |
| mrodden | its not a devstack | 23:55 |
| mrodden | just neutron | 23:55 |
| clarkb | pkg_resources is doing the check iirc | 23:56 |
| clarkb | which the entrypoint imports and makes use of | 23:56 |
| fungi | does any of this pain ease up with the new entrypointless pbr implementation? | 23:57 |
| *** dina_belova has quit IRC | 23:57 | |
| clarkb | fungi: yes, entrypointless pbr does not check aiui | 23:57 |
| clarkb | fungi: one of the things slowing entrypoints down was all of the checking. pbr wanted more speed so ditched that | 23:57 |
| fungi | also, racing stripes | 23:57 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!