| clarkb | jeblair: looks like removing gate-tempest-devstack-vm-full.FAILURE makes it render | 00:01 |
|---|---|---|
| clarkb | jeblair: does that imply there are no failures over the last 24 hours? seems odd it wouldn't render that as a flat line on the bottom of the graph | 00:01 |
| fungi | yeah, https://review.openstack.org/41698 merged at about 18:30 utc, so about 5.5 hours ago? | 00:01 |
| clarkb | hmm it has had failures | 00:03 |
| *** danger_fo is now known as danger_fo_away | 00:03 | |
| clarkb | I must be doing something wrong | 00:03 |
| jeblair | clarkb: no, i think there's something wrong with that metric | 00:04 |
| clarkb | regular job is 65:9 pass:fail over the last 24 hours in the gate. testr job is 12:2 | 00:05 |
| clarkb | 14% vs 16% | 00:05 |
| clarkb | but the samples are small | 00:06 |
| *** fifieldt has joined #openstack-infra | 00:06 | |
| jeblair | clarkb: the whisper file was 0 bytes. i removed it, and it has been replaced, but there's no data for that metric | 00:06 |
| clarkb | jeblair: ok, thanks | 00:06 |
| jeblair | there are 17 other gate job files that are 0 bytes, i will remove them | 00:06 |
| clarkb | the initial trned according to logstash is good, but getting graphite to report the numbers is better | 00:08 |
| jeblair | clarkb: erm, how hard would it be for the zmq plugin to include the node name on all its reports? | 00:09 |
| *** sarob_ has joined #openstack-infra | 00:10 | |
| jeblair | i just ran: git clone git://git.openstack.org/openstack-infra/zmq-event-publisher | 00:10 |
| jeblair | :) | 00:10 |
| jeblair | clarkb: i'll try to hack on that | 00:12 |
| clarkb | jeblair: shouldn't be too hard if that data is provided to the listener somehow | 00:12 |
| clarkb | jeblair: if the Run carries that info or refers to it then you can add it | 00:13 |
| clarkb | and you just need to update the gson model and set the value | 00:13 |
| *** sarob has quit IRC | 00:14 | |
| *** sarob_ has quit IRC | 00:14 | |
| *** pcrews has quit IRC | 00:16 | |
| clarkb | jeblair: you can get the executor of the Run if it is being built | 00:19 |
| jeblair | yeah, i'm hoping that will be enough | 00:20 |
| jeblair | trying now | 00:20 |
| *** sandywalsh has quit IRC | 00:21 | |
| morganfainberg | clarkb: ping | 00:23 |
| clarkb | morganfainberg: hi | 00:23 |
| morganfainberg | clarkb: i figured i'd come in here and ask you, since you might know | 00:23 |
| morganfainberg | clarkb: iirc run_tests is deprecated across all projects (or should be), am i remembering wrong? i.e. tox is the way forward | 00:24 |
| morganfainberg | clarkb: and if you don't know, thats fine, just curious if i remeber it right | 00:24 |
| clarkb | morganfainberg: run_tests is something we would like to deprecate but haven't gotten to that point | 00:25 |
| clarkb | there are a couple things that tox doesn't do well that causes some problems | 00:25 |
| morganfainberg | clarkb: ok, see, that's what i wanted to hear. there is a review in keystone's queue that is advocating changes to run_tests so you can do run_tests <module>.<class> for example | 00:26 |
| morganfainberg | clarkb: i was hesitant to say a hard no if run_tests was … still in reality used. | 00:26 |
| morganfainberg | clarkb: i still prefer tox personally :P | 00:26 |
| clarkb | all of the other run_tests and tox should already support that feature | 00:27 |
| clarkb | tox -epy27 foo.bar | 00:27 |
| clarkb | or run_tests.sh foo.bar | 00:27 |
| morganfainberg | ah, ok. yeah it's due to a lack of __init__ in keystone's test dir i think | 00:28 |
| morganfainberg | i know tox does the right thing | 00:28 |
| jeblair | clarkb: cool, we get it for all 3 phases | 00:28 |
| clarkb | jeblair: awesome | 00:29 |
| jeblair | clarkb: James E. Blair proposed a change to openstack-infra/zmq-event-publisher: Include the node name https://review.openstack.org/41814 | 00:29 |
| clarkb | lgtm | 00:31 |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Add high max-repo-count for index page in cgit https://review.openstack.org/41815 | 00:33 |
| pleia2 | there, commit from train | 00:33 |
| mgagne | *cough* trailing whitespaces *cough* | 00:34 |
| * mgagne has an OCD | 00:34 | |
| pleia2 | heh | 00:35 |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack-infra/config: Add high max-repo-count for index page in cgit https://review.openstack.org/41815 | 00:37 |
| *** sandywalsh has joined #openstack-infra | 00:37 | |
| pleia2 | there, no more OCD triggering | 00:37 |
| mgagne | :O | 00:37 |
| pleia2 | there are lots of options for cgit (all documented in the man page), worth a browse | 00:37 |
| mgagne | pleia2: I'm curious about cgit as gitweb has its limitations | 00:38 |
| * mgagne observes how it's being installed/configured by infra | 00:38 | |
| pleia2 | mgagne: feedback on git.openstack.org/cgit welcome :) | 00:38 |
| mgagne | pleia2: it needs an openstack skin :D | 00:39 |
| pleia2 | yes! | 00:39 |
| pleia2 | must find some art-inclined person | 00:39 |
| *** ryanpetrello has quit IRC | 00:41 | |
| mgagne | pleia2: I guess updating logo would be a good start | 00:41 |
| pleia2 | yeah, we don't have a static files directory on that server for apache at the moment, so I should make one and drop things like logo and favicon in there | 00:42 |
| pleia2 | both are cgit config file options, so that helps | 00:42 |
| mgagne | pleia2: OpenStack logo and background look to fit without problem | 00:44 |
| pleia2 | great | 00:45 |
| *** ryanpetrello has joined #openstack-infra | 00:50 | |
| *** dina_belova has joined #openstack-infra | 00:51 | |
| openstackgerrit | lifeless proposed a change to openstack-dev/pbr: Stop checking periods in commit messages https://review.openstack.org/40838 | 00:51 |
| openstackgerrit | lifeless proposed a change to openstack-dev/pbr: Fix python-ldap mirroring. https://review.openstack.org/40732 | 00:51 |
| *** woodspa has joined #openstack-infra | 00:52 | |
| *** dina_belova has quit IRC | 00:57 | |
| *** enikanorov-w has quit IRC | 00:57 | |
| *** blamar has joined #openstack-infra | 01:01 | |
| *** enikanorov-w_ has joined #openstack-infra | 01:03 | |
| *** cody-somerville has quit IRC | 01:03 | |
| *** colinmcnamara1 has joined #openstack-infra | 01:03 | |
| *** cody-somerville has joined #openstack-infra | 01:04 | |
| colinmcnamara1 | Jeremy, are you there. Have a new contributor that has a borked username issue getting this error in git-review - http://pastebin.com/TYXuuWC4 | 01:04 |
| colinmcnamara1 | remote: Permission to openstack/openstack-manuals.git denied to sarahnovotny. | 01:05 |
| colinmcnamara1 | fatal: unable to access 'http://github.com/openstack/openstack-manuals.git/': The requested URL returned error: 403 | 01:05 |
| *** zul has joined #openstack-infra | 01:06 | |
| *** senk has joined #openstack-infra | 01:08 | |
| clarkb | oh cool sarah novotny is going to contribute? | 01:08 |
| colinmcnamara1 | absolutely | 01:08 |
| senk | :) | 01:08 |
| clarkb | colinmcnamara1: the problem there is you cannot push to github | 01:08 |
| mordred | neat | 01:08 |
| colinmcnamara1 | sitting right next to me in a bar in vail | 01:08 |
| colinmcnamara1 | :) | 01:08 |
| colinmcnamara1 | awesome | 01:09 |
| * senk waves | 01:09 | |
| colinmcnamara1 | walking her through her first commit | 01:09 |
| mordred | hey senk ! | 01:09 |
| clarkb | is `git review` spitting that out? | 01:09 |
| mordred | senk: happy to have you aboard! | 01:09 |
| colinmcnamara1 | and when she is submitting to git review is getting a 403 error | 01:09 |
| colinmcnamara1 | yes | 01:09 |
| jeblair | that's slightly confusing because it has 'github.com' in the url | 01:10 |
| clarkb | ya https://github.com/openstack/openstack-manuals/blob/master/.gitreview should force it to talk to review.openstack.org | 01:10 |
| colinmcnamara1 | her lauchpad and githib id's weren't the same | 01:10 |
| colinmcnamara1 | there is a entry in the mailing lists where jeremy had to tweek something in gerrit for an earlier user with a username change | 01:10 |
| mordred | pleia2: \o/ | 01:10 |
| clarkb | unless you have a gerrit remote in the repo that points to github | 01:10 |
| anteaya | if she is changing her username in gerrit, yes the db must be changed manually | 01:10 |
| clarkb | colinmcnamara1: I don't think this is related to the username though | 01:10 |
| colinmcnamara1 | ok | 01:10 |
| colinmcnamara1 | you are the expert, not em | 01:11 |
| jeblair | senk, colinmcnamara1: can you run 'git remote -v' and paste the output? | 01:11 |
| colinmcnamara1 | it is openstack-manauls | 01:11 |
| colinmcnamara1 | err openstack-manuals | 01:11 |
| colinmcnamara1 | having sarah run it | 01:11 |
| senk | pixie:openstack-training sarah.novotny$ git remote -v | 01:11 |
| senk | gerritssh://sarahnovotny@review.openstack.org:29418/openstack/openstack-manuals.git (fetch) | 01:11 |
| senk | gerritssh://sarahnovotny@review.openstack.org:29418/openstack/openstack-manuals.git (push) | 01:11 |
| senk | originhttp://github.com/openstack/openstack-manuals.git (fetch) | 01:11 |
| senk | originhttp://github.com/openstack/openstack-manuals.git (push) | 01:11 |
| senk | pixie:openstack-training sarah.novotny$ | 01:11 |
| senk | ew. sorry. i'll pastebin | 01:11 |
| jeblair | that looks right... :/ | 01:11 |
| senk | http://pastebin.com/A7D7ueem | 01:12 |
| clarkb | can you also paste what `git review -v` outputs? | 01:12 |
| jeblair | senk: can you pastebin 'git review -v' ? | 01:12 |
| senk | http://pastebin.com/2BbFqT7s | 01:13 |
| jeblair | neat, why would git-review chose origin? | 01:14 |
| senk | that's me. | 01:14 |
| clarkb | senk: what version of git review? `git review --version` | 01:14 |
| senk | kill it? | 01:14 |
| clarkb | ya I think that last pastebin captures the relavent info | 01:15 |
| colinmcnamara1 | so, she did put a weird setting in .gitreview that specifies origin. that may be it | 01:15 |
| senk | git-review version 1.22 | 01:15 |
| jeblair | colinmcnamara1: in ~/.gitreview ? | 01:15 |
| clarkb | ya, new versions have also allowed you to set things in a global config too (to support wikimedia) but they may have been too helpful | 01:15 |
| clarkb | 1.22 is new enough to have that goodness | 01:16 |
| pleia2 | for reference, which may clear up some of the confusion, we don't actually use github directly, for us it's simply a discoverable mirror | 01:16 |
| jeblair | senk: can you pastebin ~/.gitreview ? | 01:16 |
| senk | it was the origin. | 01:16 |
| colinmcnamara1 | WOOOT!!! | 01:16 |
| jeblair | senk: cool :) | 01:16 |
| colinmcnamara1 | as always, you guys are awesome | 01:16 |
| senk | <3 | 01:16 |
| pleia2 | welcome to the project senk :) | 01:17 |
| senk | medianici lies about using the "defaultremote = origin" that if you don't know what you're doing. | 01:17 |
| senk | s/medianici/mediawiki/ | 01:17 |
| jeblair | https://review.openstack.org/#/c/41820/ | 01:17 |
| clarkb | senk: ya their workflow is slightly more special than ours | 01:17 |
| clarkb | senk: and they end up doing weirder stuff | 01:17 |
| jeblair | which is impressive, i mean, i thought we were pretty special. | 01:17 |
| clarkb | well they don't mirror to github or other places so they expect the origin to me gerrit | 01:18 |
| senk | \o/ | 01:18 |
| pleia2 | https://wiki.openstack.org/wiki/Gerrit_Workflow is what we point folks at | 01:18 |
| senk | aaah. | 01:19 |
| senk | i'll read the gerrit workflow. | 01:20 |
| anteaya | BobBall_Away: if you are still awake, the new patch also works - yay! | 01:20 |
| senk | squeeze! @colinmcnamara1 just approved my first commit! thanks for all your help, all. | 01:22 |
| colinmcnamara1 | WOOOTT!!!!! | 01:22 |
| pleia2 | \o/ | 01:22 |
| senk | i hate autocorrect. that was squee. | 01:22 |
| mordred | senk: I kinda read it as "squee" anyway | 01:23 |
| *** colinmcnamara1 has quit IRC | 01:23 | |
| anteaya | congratulations senk | 01:23 |
| mordred | morganfainberg: (reaching WAY back in to scrllback) - keystone's test suite needs work to get it into alignment with project goodness | 01:25 |
| fungi | squee ideed! | 01:25 |
| fungi | sorry i'd stepped away for a bit | 01:26 |
| mordred | morganfainberg: most specifically, it needs to a) stop fetching client libs from the internets b) start using testr and c) move tests into keystone | 01:26 |
| mordred | morganfainberg: unti lit does that, it's going to be a special snowflake | 01:26 |
| lifeless | spethial | 01:26 |
| mordred | and not able to take advantage of global work in the areas of testing | 01:26 |
| *** senk has quit IRC | 01:27 | |
| morganfainberg | mordred: that much i figured. | 01:27 |
| mordred | morganfainberg: :) just trying to catch up on scrollback. I think I'm going to lose | 01:28 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/reviewstats: Add nodepool to infra projects https://review.openstack.org/41823 | 01:28 |
| morganfainberg | mordred: hehe no worries | 01:28 |
| fungi | lifeless: if you get a bit, your input on my comment at https://review.openstack.org/#/c/41731/2/modules/openstack_project/files/gerrit/acls/stackforge/diskimage-builder.config would be most welcome | 01:30 |
| lifeless | what is milestone-proposed ? | 01:32 |
| fungi | lifeless: that more or less answers my question then ;) | 01:32 |
| fungi | lifeless: it's the pre-release freeze branch the openstack server projects use | 01:32 |
| fungi | lifeless: just wanted to confirm it was in that acl by mistake | 01:32 |
| *** jerryz has joined #openstack-infra | 01:33 | |
| lifeless | right, this is more like a client | 01:33 |
| lifeless | will release when it needs to | 01:33 |
| fungi | great. approving then | 01:33 |
| jeblair | clarkb: do you want to approve 41814 and release? exercise the publish machinery, and tomorrow i can try upgrading from the update center? | 01:34 |
| clarkb | jeblair: sure. | 01:34 |
| openstackgerrit | A change was merged to openstack-infra/reviewstats: Add nodepool to infra projects https://review.openstack.org/41823 | 01:35 |
| *** emagana has quit IRC | 01:35 | |
| *** ^d has joined #openstack-infra | 01:36 | |
| *** ^d has joined #openstack-infra | 01:36 | |
| openstackgerrit | A change was merged to openstack-infra/config: Set tag uploading permissions for TripleO https://review.openstack.org/41731 | 01:36 |
| *** gyee has quit IRC | 01:37 | |
| morganfainberg | mordred: i think we have a patchset pending that moves the tests (at the very least). | 01:38 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Use cgit server instead of github for everything https://review.openstack.org/38177 | 01:40 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Rename github-org to repo-group https://review.openstack.org/38178 | 01:40 |
| clarkb | jeblair: do you know if https://review.openstack.org/#/c/40455/ has been tested since it merged? | 01:40 |
| mordred | morganfainberg: there is. termie -2'd it and then disappeared again | 01:40 |
| clarkb | jeblair: pushing a new version of zmq-event-publisher will exercise that | 01:40 |
| clarkb | mordred: thats not very nice | 01:40 |
| clarkb | zaro: ^ | 01:40 |
| mordred | morganfainberg: I've been thinking about proposing a new version of that patch | 01:40 |
| jeblair | clarkb: i think this is the first one | 01:41 |
| mordred | morganfainberg: maybe you'll get lucky and I'llwrite hte patches to migrate _everything_ on my plane flight on thursday night :) | 01:41 |
| clarkb | jeblair: I am semi inclined to not do the first release with the new script at almost 7pm | 01:41 |
| morganfainberg | mordred: hehe ok | 01:41 |
| *** changbl has joined #openstack-infra | 01:42 | |
| mordred | morganfainberg: oh - sasha already did that | 01:43 |
| morganfainberg | mordred: yeah i was looking at that one specifically | 01:43 |
| morganfainberg | that sascha provided. | 01:43 |
| clarkb | jeblair: if you are going to around to help sort through any potential issues I am ahppy to do it though | 01:43 |
| mordred | morganfainberg: reviewed | 01:44 |
| morganfainberg | nod. | 01:44 |
| jeblair | clarkb: oh, ok, er not really, but i'm not anticipating anything beyond "it worked" or "it did not work". | 01:44 |
| clarkb | but if it did not work it will need correcting :) | 01:44 |
| jeblair | clarkb: if it does not work, we are no worse off than now. | 01:44 |
| jeblair | clarkb: sure, but that can wait, right? | 01:44 |
| clarkb | thats true, I will go ahead and give it a shot | 01:44 |
| clarkb | jeblair: tag pushed | 01:50 |
| *** dina_belova has joined #openstack-infra | 01:51 | |
| jeblair | https://jenkins.openstack.org/job/zmq-event-publisher-jenkinsci-upload/2/console | 01:52 |
| clarkb | http://repo.jenkins-ci.org/list/releases/org/jenkins-ci/plugins/zmq-event-publisher/0.0.2/ seems to have worked | 01:55 |
| clarkb | we'll know when the updates site updates | 01:55 |
| *** dina_belova has quit IRC | 01:56 | |
| SpamapS | did I see the tripleo-ptl thing land in my email jenkins updates, or was that just a merged (but something has to be HUP'd?) | 02:05 |
| SpamapS | ! [remote rejected] 0.1.0 -> 0.1.0 (can not create new references) | 02:06 |
| SpamapS | for os-apply-config | 02:06 |
| mordred | SpamapS: it takes a potential double cron cycle | 02:06 |
| mordred | lemme see if the grou pis there | 02:06 |
| SpamapS | Grou's? | 02:07 |
| * SpamapS hides | 02:07 | |
| mordred | SpamapS: try again | 02:07 |
| clarkb | the group is there but os-apply-config does not have it in its acl | 02:08 |
| mordred | oh. piddle | 02:08 |
| SpamapS | cron: keeping the coffee machine brewing since 1983 | 02:08 |
| clarkb | probably needs to be run by hand (this is the race condition in puppet that I am talking about) | 02:09 |
| * mordred goes to re-run manage_projects (man, I REALLY want to figure out what the heck is up with the race condition) | 02:09 | |
| clarkb | I don't think all of the files are updated befure the subscription can run | 02:09 |
| mordred | ah. | 02:09 |
| * mordred running | 02:09 | |
| clarkb | some other project added the ptl group, runs manage_projects then os-apply-configs acl file is updated | 02:09 |
| jgriffith | anybody want to educate me on sitepackages=True/False in tox.ini | 02:09 |
| jgriffith | ie https://review.openstack.org/#/c/39501/1 | 02:09 |
| jgriffith | My first thought was "sure, sounds good" | 02:10 |
| clarkb | jgriffith: sure | 02:10 |
| clarkb | sitepackages are evil | 02:10 |
| jgriffith | but I'm wondering if there's a caveat I'm missing | 02:10 |
| *** yaguang has joined #openstack-infra | 02:10 | |
| jgriffith | clarkb: well, then that was easy :) | 02:10 |
| jgriffith | clarkb: My thought was if you're using venv, use venv | 02:10 |
| clarkb | they are evil because if you have package foo at version X.Y installed globally but you depend on foo version X.Y+1 tox and pip will not install the proper version in the virtualenv | 02:10 |
| jgriffith | but I'm not the most familiar with that workflow so I wanted to make sure there wasn't something I need to be considering here | 02:10 |
| jgriffith | clarkb: got ya... which I tend to run in to a bit lately :( | 02:11 |
| clarkb | there is a work around for that where you can add -U to the deps on line 11 and 12 | 02:11 |
| jgriffith | oooo... goody! | 02:11 |
| clarkb | which will force pip to update the existing packages as necessary | 02:11 |
| jgriffith | clarkb: nice | 02:11 |
| clarkb | but typically we just disable site packages if they aren't necessary for something like libvirty | 02:12 |
| clarkb | keeps things simple | 02:12 |
| jgriffith | clarkb: ok, so I'm not missing something that I should be thinking of with respect to this change | 02:12 |
| * mordred looing | 02:12 | |
| mordred | hey jgriffith | 02:12 |
| jgriffith | mordred: hola | 02:12 |
| * jgriffith ducks | 02:12 | |
| clarkb | jgriffith: I don't think so. The exception to the rule is when you have a python dep that cannot be satisfied by pip | 02:12 |
| clarkb | libvirt being the big one | 02:12 |
| jgriffith | hmmm... psql-dev perhaps? | 02:12 |
| mordred | jgriffith: ah! I believe you inherited that line from nova | 02:12 |
| jgriffith | :) | 02:13 |
| mordred | jgriffith: the default is off, so technically you could just delete that line | 02:13 |
| mordred | jgriffith: commit 930f5891b0815e1b49b9b2cc840e0c24b2796e84 | 02:13 |
| mordred | is where it came into your tree | 02:13 |
| jgriffith | mordred: ohhh interesting | 02:13 |
| *** woodspa has quit IRC | 02:13 | |
| mordred | I FULLY support removing it from your tree | 02:13 |
| *** nati_ueno has quit IRC | 02:13 | |
| mordred | nova is the only project we expect it to be on in, and that's because we cannot do the other thing | 02:14 |
| jgriffith | I noticed that in the tox docs | 02:14 |
| jgriffith | got ya | 02:14 |
| jgriffith | excellent'e | 02:15 |
| jgriffith | thanks guys! | 02:15 |
| mordred | jgriffith: thank you for coming in and asking! :) | 02:15 |
| jgriffith | :) | 02:15 |
| jgriffith | mordred: nothing worse than *thinking* I know the answer and finding out my assumptions are wrong ;) | 02:15 |
| mordred | jgriffith: now you're describing every moment of my life | 02:17 |
| jgriffith | mordred: hehe... truly, that holds for all of us. Just some admit it, others pretend | 02:17 |
| jgriffith | :) | 02:17 |
| *** mriedem has quit IRC | 02:17 | |
| *** ryanpetrello has quit IRC | 02:19 | |
| openstackgerrit | benley proposed a change to openstack-infra/jenkins-job-builder: Add support for parameter filters in copyartifact https://review.openstack.org/41582 | 02:23 |
| openstackgerrit | benley proposed a change to openstack-infra/jenkins-job-builder: Add support for parameter filters in copyartifact https://review.openstack.org/41582 | 02:24 |
| openstackgerrit | benley proposed a change to openstack-infra/jenkins-job-builder: Add support for parameter filters in copyartifact https://review.openstack.org/41582 | 02:24 |
| *** cody-somerville has quit IRC | 02:29 | |
| *** dguitarbite has joined #openstack-infra | 02:31 | |
| SpamapS | mordred: push worked this time | 02:32 |
| openstackgerrit | benley proposed a change to openstack-infra/jenkins-job-builder: Add display-name job property. https://review.openstack.org/41828 | 02:33 |
| openstackgerrit | benley proposed a change to openstack-infra/jenkins-job-builder: Add display-name job property. https://review.openstack.org/41828 | 02:34 |
| *** anteaya has quit IRC | 02:40 | |
| *** cody-somerville has joined #openstack-infra | 02:41 | |
| *** ^d has quit IRC | 02:44 | |
| *** rcleere has joined #openstack-infra | 02:46 | |
| *** dina_belova has joined #openstack-infra | 02:52 | |
| mordred | SpamapS: woot | 02:56 |
| *** dina_belova has quit IRC | 02:57 | |
| mordred | jog0: ping | 02:58 |
| clarkb | we should store our longs in https://github.com/philipl/pifs | 02:59 |
| clarkb | s/longs/logs/ | 03:00 |
| mordred | omg | 03:00 |
| SpamapS | mordred: so now we wait for something magical to happen and the tag to appear on pypi? | 03:03 |
| mordred | SpamapS: it sohuld have showed up a long time ago if it was gonna | 03:04 |
| SpamapS | https://pypi.python.org/pypi/os-apply-config | 03:04 |
| SpamapS | 0.0.1 still | 03:04 |
| mordred | SpamapS: hahahahahahaha | 03:05 |
| SpamapS | you did the old name | 03:05 |
| SpamapS | ? | 03:05 |
| mordred | SpamapS: you sure don't have pypi jobs | 03:05 |
| *** jfriedly has quit IRC | 03:06 | |
| jog0 | mordred: pong | 03:06 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Actually let os-*-config release to PyPI https://review.openstack.org/41832 | 03:08 |
| mordred | SpamapS, clarkb ^^ | 03:08 |
| mordred | jog0: re: the new pep8 thing... | 03:08 |
| mordred | jog0: a) do you think then that it should wind up as a thing that tox -epep8 just runs every time? | 03:08 |
| mordred | jog0: b) there is no b) | 03:08 |
| jog0 | mordred: you see my second comment | 03:09 |
| mordred | jog0: yup | 03:09 |
| jog0 | actually the logic could be if 'git status' shows nothing, run git diff HEAD^, else do git diff | 03:10 |
| mordred | jog0: that's what I'm saying - how do we not break that workflow, but still get the ability for me to check to see if I'm doing something evil in the future? | 03:10 |
| mordred | jog0: good call | 03:10 |
| mordred | jog0: maybe it wants to be a script that gets called from tools/* and gets added to oslo-incubator? | 03:10 |
| jog0 | mordred: I don't think there is away to not break the workflow | 03:10 |
| *** yaguang has quit IRC | 03:10 | |
| mordred | jog0: we could make hacking/flake8 MUCH smarter... | 03:11 |
| mordred | and be able to configure it to a) do this and b) what the expanded list is | 03:11 |
| jog0 | in part because if: a) the extra rigid checks are the default config, I run flake8 via the VIM plugin on the file I am working on and everything looks good so I push but then the new code only checks catch me | 03:11 |
| jog0 | err I got that backwards, if extra rigid is default then I fix things outside of my patch | 03:12 |
| mordred | oh yeah. hrm | 03:12 |
| jog0 | I do like the autofix a big bug in several 500 line patches and just ram it through | 03:13 |
| jog0 | and then fix for everything | 03:13 |
| jog0 | although we can tweak that to be per directory | 03:13 |
| jog0 | that may be a better middleground | 03:13 |
| *** ryanpetrello has joined #openstack-infra | 03:16 | |
| *** ryanpetrello has quit IRC | 03:17 | |
| jog0 | also this is a nova only problem I think | 03:18 |
| jog0 | due to size | 03:18 |
| jog0 | mordred: ^ | 03:18 |
| mordred | jog0: yeah. agree | 03:21 |
| mordred | jog0: that's sort of why I was thinking that the summit might be a nice time | 03:21 |
| mordred | BUT | 03:21 |
| mordred | even then there will be like 200 outstanding patches that are going to get rebase screwed | 03:21 |
| jog0 | mordred: do it right when Icehouse opens | 03:22 |
| lifeless | what are you brewing up? | 03:22 |
| jog0 | those few weeks between the release and summit when everyone is drunk and not coding | 03:22 |
| mordred | jog0: :) | 03:23 |
| jog0 | mordred: there is also the path taken with pylint | 03:23 |
| *** jerryz has quit IRC | 03:23 | |
| jog0 | which is make sure the number of errors doesn't go up | 03:23 |
| *** yaguang has joined #openstack-infra | 03:24 | |
| jog0 | err warnings* | 03:24 |
| jog0 | that should leave the workflow mostly the same if not totally | 03:24 |
| fungi | i think i want a test which does an sdist, then blows away everything except .git and .gitignore, then untars the sdist tarball in and confirms that git shows no changes. not sure if that warrants a separate job or can be combined into an existing one | 03:29 |
| fungi | seems a bit destructive for hacking or any flake8 plugin | 03:30 |
| *** vipul is now known as vipul-away | 03:30 | |
| *** vipul-away is now known as vipul | 03:30 | |
| *** vipul is now known as vipul-away | 03:33 | |
| *** vipul-away is now known as vipul | 03:33 | |
| mordred | fungi: well ... | 03:33 |
| mordred | fungi: that was a job that ttx wanted a long time ago - we did the other thing and did the setuptools-git route | 03:34 |
| mordred | fungi: although I believe you are wanting to prove different things | 03:34 |
| fungi | wanting to prove that we don't break our ability to consistently validate tarballs later | 03:35 |
| fungi | particularly guarding against mistakes in changes to the way tarballs get created | 03:37 |
| fungi | so that we can then have confidence that when a post-merge job expects to be able to do that to validate a release tarball is not tampered with or corrupted, we avoid false positives | 03:40 |
| locke105 | anyone used vagrant at all before? been playing with it for testing deployments with chef and stuff... wondering if anyone knows of an OpenStack provider plugin for it... | 03:43 |
| *** SergeyLukjanov has joined #openstack-infra | 03:43 | |
| mordred | warning - discussion about pip requirements incoming | 03:45 |
| mordred | locke105: vagrant has never worked for me at all | 03:45 |
| lifeless | dum-dum-dum-ddddduuuuummmmmm | 03:45 |
| mordred | locke105: but I hate ruby, and it hates me, and I have free openstack cloud accounts | 03:45 |
| locke105 | lol | 03:45 |
| mordred | locke105: so ignore me | 03:45 |
| locke105 | personally i'm more a python fan | 03:45 |
| lifeless | so context | 03:45 |
| locke105 | but ruby is alright besides the silly syntax | 03:45 |
| lifeless | mordred put up a patch that just tweaks minor versions on the pip requirements for os-collect-config | 03:46 |
| locke105 | "just tweaks" | 03:46 |
| locke105 | :) | 03:46 |
| lifeless | I whinged about it, not because I want to work with different versions [at least it's not an issue today], but because I don't want daily 'here is the latest tweak because $otherproject is incompat with that thing' | 03:46 |
| mordred | https://review.openstack.org/#/q/status:open+branch:master+topic:openstack/requirements,n,z | 03:46 |
| lifeless | So I want to understand why we're doing it the way we are | 03:46 |
| lifeless | and mordred quite reasonably said 'well, how else?' | 03:47 |
| lifeless | It seems to me that if we have a centralised list of what works | 03:47 |
| SpamapS | because we don't have a way to tell pip install -r "union with URL x" | 03:47 |
| lifeless | we should decouple 'things needed' and 'versions needed' | 03:47 |
| mordred | but when do you recouple them | 03:48 |
| mordred | ? | 03:48 |
| SpamapS | what if we put in requirements.txt "openstack-requirements>=2013.2" | 03:48 |
| lifeless | Now, if pip doesn't have support for that built-in - and we can look at that separately - we could always just compile a fixed central list in the global requirements repo, so there is one and only one place it's stored. | 03:48 |
| mordred | SpamapS: that breaks for other reasons | 03:48 |
| mordred | lifeless: I would like to go back to the problem you are trying to solve | 03:49 |
| lifeless | the reason I care about this is that slow changing projects - like oac/occ/orc will quickly become mostly requirements.txt changes. | 03:49 |
| lifeless | and by quickly I mean in a month or two. | 03:49 |
| mordred | so don't merge it | 03:49 |
| mordred | merge like, once every 2 months or so | 03:49 |
| lifeless | That will cause problems won't it ? | 03:49 |
| mordred | why? | 03:49 |
| lifeless | I mean the reason you're doing this is to avoid breaking devstack, isn't it? | 03:50 |
| mordred | right. so, devstack is goingto force upgrade you anyway | 03:50 |
| locke105 | what about an auto posted change like the transifex stuff | 03:50 |
| mordred | you don't get a choice about that | 03:50 |
| mordred | locke105: that's what we're doing/talking about | 03:50 |
| lifeless | mordred: upgrades are fine; *we're* very compatible. It's the *tighter* constraints your review adds that set of alarm bells for me. | 03:50 |
| mordred | lifeless: devstack's setup_develop function now runs update.py from trunk requirements.txt | 03:50 |
| SpamapS | mordred: ergo the point... "don't crap deps on X if you're going to get them from the actual place that needs them anyway in the deploy scenario you are testing" | 03:51 |
| mordred | SpamapS: sorry, can't parse that sentence | 03:51 |
| lifeless | mordred: if we merge any patch that sets an upper bound on $thing, pip will start whinging when trunk accepts a newer version if the newer version is already installed, and the havoc we're trying to fix will occur | 03:51 |
| clarkb | mordred: he is saying stop proposing changes if devstack will do it anyways | 03:51 |
| clarkb | the problem here is we are a lot more than devstack | 03:51 |
| mordred | right. unittests, etc | 03:52 |
| SpamapS | mordred: "regarding 'right. so, devstack is going to force upgrade you anyway'" | 03:52 |
| SpamapS | gah too many quotes | 03:52 |
| clarkb | specifically this is a major problem for people like tripleo trying to package this up distro like | 03:52 |
| mordred | we WANT everything to be COMPLETELY ALIGNED by release | 03:52 |
| clarkb | SpamapS: I think you are working around that by venv'ing everythin | 03:52 |
| lifeless | mordred: so I don't want to land overly tight requirements in the first place, because they are the cause of e.g. occ causing issues. | 03:52 |
| clarkb | ubuntu and fedora and rhel and suse and so on don't have that as an option | 03:52 |
| SpamapS | poor buggers | 03:52 |
| SpamapS | ;) | 03:52 |
| lifeless | clarkb: we do, because we may well end up with cross-release situations. | 03:52 |
| lifeless | clarkb: e.g. havan * + trunk nova-baremetal | 03:53 |
| mordred | lifeless: to a degree, strict alignment with openstack/reuqiremnets is just one of the prices you pay for being a part of openstack | 03:53 |
| *** dina_belova has joined #openstack-infra | 03:53 | |
| mordred | some projects don't strictly need it | 03:53 |
| mordred | but we're not really designed for 22 snowflakes | 03:53 |
| mordred | openstack is One Project | 03:53 |
| lifeless | mordred: sure, but this isn't about being part of openstack - I'm talking about the technical approach in solving the devstack pain for CI | 03:53 |
| SpamapS | mordred: can you unblock my head by explaining why we can't just depend on a single openstack-requirements pypi package? | 03:53 |
| mordred | this isn't about solving the devstack pain per se | 03:53 |
| mordred | that's solved | 03:53 |
| lifeless | mordred: and framing this as a snowflake question is also missing the point | 03:53 |
| mordred | now it's about solving requirements drift | 03:53 |
| mordred | we want projects to stop having divergent requirements files | 03:54 |
| clarkb | SpamapS: because openstack requirements is a super set that you almost never want (my issue with that approach, not sure if that is what mordred will say) | 03:54 |
| lifeless | so the reason divergence exists is because we have redundant information in them. | 03:54 |
| lifeless | I'm proposing that we tackle that directly. | 03:54 |
| mordred | ok. let's come back to that then | 03:55 |
| mordred | becuase I don't see how it's different in result | 03:55 |
| mordred | if you are going to make an artifact that one can pip instal | 03:55 |
| mordred | that artifact needs to list the versions that are in openstack/requirements, no? | 03:55 |
| mordred | and if you are going to ahve a git repo and you want tox -epy27 to work, you'll need the versions | 03:56 |
| lifeless | lets start by setting the use cases | 03:56 |
| mordred | so either they're self-contained, or somethign is going to have to make additional hits out to o/r | 03:56 |
| lifeless | there is install from git | 03:56 |
| lifeless | there is install from pypi | 03:56 |
| lifeless | and there is setup a developer test-running-environment | 03:56 |
| lifeless | are there more ? | 03:56 |
| mordred | nope, I think those are the main ones | 03:57 |
| lifeless | now, install from pypi is easy: we can materialise the current set of constraints at sdist time. | 03:57 |
| mordred | agreed | 03:57 |
| *** dina_belova has quit IRC | 03:58 | |
| SpamapS | clarkb: hrm. Not sure that is actually a big deal to me.. the overlap is probably "a few modules here and there", but "mmk" | 03:58 |
| SpamapS | err | 03:58 |
| mordred | SpamapS: think about if you are python-swiftclient | 03:58 |
| lifeless | SpamapS: uhm, it's huge. | 03:58 |
| SpamapS | not overlap, but the "inverse of overlap" | 03:58 |
| mordred | SpamapS: and o/r makes you install django | 03:58 |
| lifeless | SpamapS: postgresql client libs on every install, for instance... | 03:58 |
| SpamapS | Oh are we putting those things in reqs too? | 03:58 |
| SpamapS | I had thought deploy reqs where there are multiple choices are not in there. | 03:58 |
| mordred | no - but you need them to install the postgres python lib | 03:58 |
| lifeless | SpamapS: the python lib yes, which transitively ... | 03:58 |
| SpamapS | Indeed, if they're all in there, then it is not what I thought it was. :) | 03:59 |
| lifeless | SpamapS: they should be in extras, but the *version pins* for them are in o/r. | 03:59 |
| SpamapS | Ok so pip needs a --union-with X feature. :) | 03:59 |
| mordred | extras is bs, btw | 03:59 |
| mordred | although I'm willing to be swayed on that _later_ | 03:59 |
| lifeless | SpamapS: so the issue is that -r installs everything in the list. | 04:00 |
| lifeless | SpamapS: the way pip upstream seem to think it works is that you have one -r for each app you're installing. | 04:00 |
| locke105 | so.. we have something similar already with oslo-incubator | 04:00 |
| lifeless | SpamapS: the result of that here is that there is a requirements.txt *in the repos* for each app | 04:01 |
| SpamapS | right | 04:01 |
| locke105 | each project has an openstack-common.conf which says which modules in oslo it uses | 04:01 |
| lifeless | SpamapS: which is not the pip upstream story, because they are focused on web devs deploying their own thing | 04:01 |
| locke105 | so we could do the same with requirements | 04:01 |
| mordred | locke105: already done | 04:01 |
| mordred | locke105: we're debating whether it's a good idea to do so | 04:01 |
| locke105 | oh | 04:01 |
| lifeless | mordred: so, is there a requirement that the per-project requirements bumps be async ? | 04:01 |
| clarkb | lifeless: I think that is an important point | 04:01 |
| mordred | locke105: we're about to get way down in the weeds - it'll be fun | 04:02 |
| clarkb | lifeless: the existing tools are built around deploy your super specific app in secret | 04:02 |
| mordred | lifeless: there is a request from the projects about that so far | 04:02 |
| lifeless | mordred: or do we test that a new global requirements change works with all current repos already ? | 04:02 |
| locke105 | so thats what was meant by "superset" | 04:02 |
| mordred | lifeless: we do test that | 04:02 |
| locke105 | makes sense now | 04:02 |
| mordred | lifeless: requirements changes now also gate on devstack | 04:02 |
| clarkb | the use case of install this thing with a ton of optional requirements with a set of valid versions for each requirement across many distros is not something they try to solve | 04:02 |
| lifeless | mordred: so, I think the simplest thing with no need to evolve pip or whatever is to put the per-project requirements files in the global repo. | 04:03 |
| mordred | we do _not_ test that requirements changes break or not break all the unittests - although I believe we're getting closer to changing how that works | 04:03 |
| clarkb | its all geared around deploy one app to one place in a non open friendly manner | 04:03 |
| mordred | lifeless: how is that simpler? | 04:03 |
| mordred | lifeless: can we go back to the use cases? | 04:03 |
| jgriffith | mordred: BTW, I did some more digging and the issues I was having earlier were in fact due to vagrant caching of things | 04:03 |
| lifeless | mordred: it means you /can't/ have skew between versions. | 04:03 |
| mordred | lifeless: I believe we're jumping again | 04:03 |
| jgriffith | mordred: just FYI if somebody mentions it again | 04:03 |
| lifeless | mordred: almost certainly | 04:03 |
| mordred | jgriffith: awesome! locke105 was just asking about using vagrant | 04:03 |
| lifeless | mordred: use cases; pypi - tick. local dev and isntall from git. | 04:04 |
| mordred | although if you guys talk about that, can I request that it not be in here just right now | 04:04 |
| jgriffith | locke105: if you want to hear about my experience and new method gimmie a shout | 04:04 |
| lifeless | mordred: so 'pip install git+https://github.com/openstack/nova' should work | 04:04 |
| lifeless | mordred: which will pull the repo, and do setup.py stuff right? | 04:05 |
| mordred | yes | 04:05 |
| lifeless | mordred: could that sanely look for a subdirectory and if missing clone the global requirements repo to it ? | 04:05 |
| lifeless | mordred: I'm considering implementation obviously | 04:05 |
| mordred | now you're desribing the submodule approach, essentially - but, it might be able to do that | 04:06 |
| lifeless | so git submodules might be an answer, but that can be a bucket of worms, so I deliberately stayed away | 04:06 |
| mordred | right. I'm just saing - the idea is the same | 04:07 |
| lifeless | yes | 04:07 |
| lifeless | tox -epy27 could operate basically the same way | 04:07 |
| mordred | right. but then we're in a world of complexity that's hard to reason about | 04:07 |
| mordred | what happens if I do git clone nova nova-blah locally on my laptop | 04:08 |
| mordred | and then want to run setup.py in that dir | 04:08 |
| mordred | and I'm not on the internet | 04:08 |
| lifeless | mordred: well, today it blows up. | 04:08 |
| lifeless | mordred: because you're not on the internet. | 04:08 |
| mordred | I keep a local pypi mirror on my laopt | 04:08 |
| mordred | laptop | 04:08 |
| lifeless | so this scenario is one where git submodules look nicer still | 04:09 |
| mordred | which is configured in my ~/.pip/pip.conf | 04:09 |
| mordred | but we're still winding up with the specific versions in the project repo - we're just making the dev run the syncing code a bunch at times the dev may not know he's doing it | 04:10 |
| mordred | rather than storing the result of the sync in the repo | 04:10 |
| mordred | right? | 04:10 |
| lifeless | what you wrote appears to have a contradiction | 04:10 |
| lifeless | specific versions in the project repo <=> rather than storing the result of the sync in the repo | 04:11 |
| mordred | sorry - specific versions in the local working tree | 04:11 |
| mordred | or on the local dev env | 04:11 |
| lifeless | yes, at any point in time there are specific versions selected | 04:11 |
| lifeless | which is AIUI one of your goals | 04:12 |
| lifeless | to avoid spurious failures due to random incompat or shit hanging around. | 04:12 |
| mordred | right. | 04:12 |
| mordred | have you interacted with the transifex patches? | 04:12 |
| lifeless | a little | 04:12 |
| lifeless | none of tripleo is translated yet | 04:12 |
| mordred | constant stream of update patches, projects land them whenever | 04:12 |
| lifeless | but transifex is different, because they are project specific | 04:12 |
| mordred | it's still patch churn | 04:13 |
| lifeless | as in, if the project stops changing, they will stabilise as translators catch up | 04:13 |
| mordred | and the content is largely opaque to the reviewer | 04:13 |
| lifeless | also if the project doesn't merge them it has no impact on installability or CI | 04:13 |
| mordred | right. if the project does not merge a requirements change, nothing breaks | 04:13 |
| mordred | the only time you'll be forced to merge your outstanding reuqirements change | 04:14 |
| lifeless | I have a scenario I think will break. | 04:14 |
| lifeless | I'd be happy to know I'm wrong. | 04:14 |
| *** vogxn has joined #openstack-infra | 04:14 | |
| mordred | you may get pressure to do something if someone tries to land a change to requirements and it's not compatible with your project | 04:14 |
| mordred | but, by definition, your project will always be compatible with its own local set and the global set | 04:15 |
| mordred | because it's not possible land a breaking change to the global set - if you're in the d-g and your code actually gets tested there ) | 04:15 |
| lifeless | The scenario is - say we land the patch you put forward that caps testscenarios at <0.5 | 04:15 |
| mordred | :) | 04:15 |
| lifeless | [btw, why do that - I do backwards compat!] | 04:15 |
| clarkb | yes I questioned capping testscenarios and other lifeless libs | 04:16 |
| mordred | we've been bitten too many times - propose a change removing the caps | 04:16 |
| lifeless | Then, to have our project successfully install you have to have testscenarios < 0.5. | 04:16 |
| mordred | k | 04:16 |
| lifeless | When I then release 0.5 | 04:16 |
| lifeless | and the global project wants to raise the cap. | 04:16 |
| lifeless | The per-project cap will make my project not installable. | 04:17 |
| mordred | why? | 04:17 |
| lifeless | because it still says < 0.5 | 04:17 |
| mordred | where will it not be installable? | 04:17 |
| lifeless | well, it would have to downgrade testscenarios in the test environment | 04:17 |
| mordred | locally, it will be installable, because you will have not touched your cap, nor will you have touched your code, and g-r will not be in context | 04:18 |
| mordred | what test environment? | 04:18 |
| lifeless | devstack-gate | 04:18 |
| mordred | ah. see, that's just the thing | 04:18 |
| mordred | we update your requirements in devstack | 04:18 |
| mordred | before we install you | 04:18 |
| mordred | you are not asked about it | 04:18 |
| mordred | this is already landed | 04:18 |
| lifeless | ok | 04:18 |
| lifeless | so, if we said 'we merge these proposals before we cut a pypi release', would -infra be happy? | 04:19 |
| mordred | because that's where the major pain is | 04:19 |
| mordred | sure! | 04:19 |
| lifeless | [and if there is a minimum version that we actually need] | 04:19 |
| mordred | honestly - I'm not sure we _need_ auto proposals anymore | 04:19 |
| mordred | I think that devstack autoupdate | 04:19 |
| mordred | combined with gating reuqirements on that | 04:19 |
| mordred | combined with making sure that when you change reuqirements you only change to things in the gate | 04:19 |
| mordred | is probably fine moving forward | 04:20 |
| mordred | that said - I'd like to get people up to the current baseline as of when we started gating requirements | 04:20 |
| mordred | and then let's let it run for a while and see if anything breaks without doing autosync proposals, yeah? | 04:20 |
| lifeless | damn | 04:21 |
| lifeless | I just -2'd | 04:21 |
| mordred | hah | 04:21 |
| mordred | lifeless: _you_ are fine in this case | 04:21 |
| lifeless | mordred: ok back to -1, I'd like the testscenarios thing changed | 04:22 |
| lifeless | I realise the global set is different, I'll dig up the repo and propose a change to it there | 04:22 |
| mordred | cool. please do | 04:22 |
| mordred | and I'll happily update the patch | 04:22 |
| lifeless | for now, Ineed to pickup Cynthia | 04:22 |
| mordred | the main reason I want to sync everyone now | 04:22 |
| lifeless | thanks for the chat | 04:22 |
| mordred | is to get rid of referneces to d2to1, btw | 04:23 |
| mordred | lifeless: thank you! it was a good one | 04:23 |
| mordred | lifeless: I think you have strengthend the side of me leaning towards holding off on auto proposals for a while | 04:23 |
| clarkb | I don't see what is wrong with auto proposals | 04:24 |
| clarkb | there will only ever be one per project at any given time | 04:24 |
| clarkb | just ignore it | 04:24 |
| mordred | clarkb: I don't think they're wrong - but I do think I want to see how the recent changes we make affect things | 04:25 |
| mordred | clarkb: and whether or not we still feel like they are a thing that will be helpful | 04:25 |
| mordred | clarkb: btw - not related- https://review.openstack.org/#/c/41832/ | 04:25 |
| lifeless | clarkb: the thing about dashboards of 'I need to review X' is that having a thing there always is really annoying | 04:26 |
| clarkb | lifeless: you can ignore it with a -2 I think | 04:26 |
| lifeless | clarkb: I'd rather say 'make it reliable and automerge' or 'don't do it'; but thats a different discussion. | 04:26 |
| clarkb | because that will put it in the reviewed column until you remove the -2 | 04:27 |
| lifeless | clarkb: it will get refreshed on every o/r change. | 04:27 |
| clarkb | lifeless: but -2's are always reapplied | 04:27 |
| morganfainberg | mordred: any insight on what causes: https://review.openstack.org/#/c/36184/ the gerrit claiming a path conflict? I've been unable to duplicate the issue in cherry-picks, merges, etc locally but i might be missing something obvious. | 04:27 |
| lifeless | and then it's not there when you need it | 04:27 |
| mordred | morganfainberg: lemme try | 04:27 |
| morganfainberg | mordred: thanks :) | 04:27 |
| morganfainberg | mordred: sometimes my git-fu is lacking. | 04:27 |
| clarkb | morganfainberg: jgit | 04:28 |
| morganfainberg | clarkb: oh. right. | 04:28 |
| morganfainberg | clarkb: jgit is special. | 04:28 |
| clarkb | lifeless: I don't think we will have a perfect thing that makes everyone happy... some people want auto merge, others are firmly against it | 04:28 |
| clarkb | so we tend to be a little conservative and deal with it, but I really don't think the transifex or requirements changes are that intrusive | 04:29 |
| clarkb | I mean my queue of things is like 60 long a couple extra changes that I can ignore doesn't bother me | 04:29 |
| clarkb | morganfainberg: yup, it often fails where normal git is fine | 04:30 |
| mordred | morganfainberg: I rebased locally and repushed - for the reasons clarkb says | 04:30 |
| morganfainberg | mordred: thank you :) | 04:30 |
| mordred | clarkb: we have assininely long review lists, you and I | 04:31 |
| mordred | morganfainberg: and I've re-+2'd it - so you sohuld be good to go if you want to aprv | 04:31 |
| morganfainberg | mordred: i typically wait till jenkins doesn't explode if it's going to run the normal tests | 04:32 |
| morganfainberg | just paranoid | 04:32 |
| mordred | yup | 04:32 |
| clarkb | mordred: we appreciate the paranoia. It keeps zuul happy | 04:32 |
| morganfainberg | so i'll app once thats done tonight :) | 04:32 |
| clarkb | er morganfainberg ^ | 04:32 |
| morganfainberg | clarkb: i am running a zuul/gerrit internally for my company's code | 04:32 |
| morganfainberg | i totally get it :) | 04:32 |
| mordred | clarkb: any reason to not land the devstack don't relable patch? | 04:33 |
| clarkb | mordred: I have been letting jeblair coordinate those | 04:33 |
| morganfainberg | reminds me, i need to upgrade our zuul… it's ancient | 04:33 |
| clarkb | mordred: I think he ahs the whole thing planned out in his head | 04:33 |
| mordred | clarkb: good call. I'll just +2 | 04:33 |
| mordred | clarkb: by the whole thing, do you mean the universe? | 04:34 |
| clarkb | morganfainberg: keep in mind tip of master zuul does gearman with jenkins now | 04:34 |
| clarkb | mordred: yes, I am pretty sure when it rains up here jeblair is at fault | 04:34 |
| mordred | clarkb: ++ | 04:34 |
| morganfainberg | clarkb: thats why i need to upgrade :) | 04:34 |
| mordred | clarkb: btw - the queue for o/config is stupid | 04:35 |
| clarkb | queue? | 04:35 |
| clarkb | review queue? | 04:35 |
| mordred | yeah | 04:36 |
| clarkb | I was new change submission freeze | 04:36 |
| clarkb | no new changes until everything else is either merged or abandoned :) | 04:37 |
| mordred | yeah - but - but | 04:37 |
| mordred | ponies! | 04:37 |
| mordred | clarkb: you know - I'm thinking with custom dashboards I might be able to cook up what I really need right now | 04:38 |
| mordred | which is "show me things which I haven't reviewed, but which also don't have a -1 on them anywhere | 04:38 |
| *** melwitt has quit IRC | 04:39 | |
| lifeless | clarkb: /all/ proposals in tripleo projects are fully reviewed, every day. | 04:43 |
| lifeless | clarkb: you having 60 open is a defect in something :) | 04:43 |
| clarkb | lifeless: the number of changes submitted is far too large | 04:44 |
| clarkb | and the number of reviewers is too small | 04:44 |
| clarkb | lifeless: also having mordred submit code really makes it hard to keep up | 04:45 |
| clarkb | will submit like 15 patches a day | 04:45 |
| mordred | clarkb: dammit | 04:49 |
| mordred | clarkb: there does not seem to be a searchable boolean for Util.M.changesReviewedBy | 04:49 |
| mordred | there is one for changesReviewedBy | 04:50 |
| mordred | so, as best I can tell | 04:50 |
| mordred | there is no way for me to search for haveReviewed | 04:50 |
| clarkb | ya it is one of the weird things that the important changes had to work around | 04:51 |
| clarkb | and requires a patch which upstream does not want | 04:51 |
| * mordred is reading your patch | 04:51 | |
| mordred | they don't want me to be able to filter out shit I've dealt with? | 04:51 |
| mordred | because this: | 04:52 |
| mordred | https://review.openstack.org/#/q/watchedby:mordred%2540inaugust.com+-label:CodeReview%253C%253D-1+-label:Verified%253C%253D-1++-status:workinprogress,n,z | 04:52 |
| clarkb | apparently not, but I think gerrit 2.6 exposes all this through the REST API | 04:52 |
| mordred | is really close to being a todo list | 04:52 |
| clarkb | so, in addition to custom dashboards writing some js to show you what you want shouldn't be too bad | 04:52 |
| *** luisg has quit IRC | 04:52 | |
| clarkb | but that is all theory at this point | 04:52 |
| *** luisg has joined #openstack-infra | 04:52 | |
| mordred | god | 04:52 |
| mordred | all I need is one more single boolean search criteria | 04:52 |
| mordred | and I wouldnt' even need custom dashboards | 04:53 |
| clarkb | mordred: the problem is in the DB schema | 04:53 |
| clarkb | there is no good way to check if you ahve reviewed a thing | 04:53 |
| mordred | well, that's because their db schema is bonghits | 04:53 |
| *** dguitarbite has quit IRC | 04:53 | |
| *** dina_belova has joined #openstack-infra | 04:53 | |
| clarkb | I can't remember all the reasons now but it had to do with a vote of 0 being different than a non zero vote | 04:53 |
| clarkb | so important changes approximates and only looks at non zero votes | 04:53 |
| mordred | oh my. I just thought of an evil thing I could do | 04:54 |
| mordred | I could use that view | 04:54 |
| mordred | but put in -is:starred | 04:54 |
| mordred | and then I could star something to mean I'm done with it | 04:54 |
| clarkb | mordred: I have been doing the inverse lately | 04:54 |
| clarkb | I do a quick skim of the list to find things that need attention, start them all, then go through the starred list removing stars as I review | 04:55 |
| mordred | yeah - alternately I could use the above view to show me things I havne't seen at all | 04:55 |
| mordred | I could star those | 04:55 |
| mordred | then go through star list and unstar | 04:55 |
| mordred | yeah | 04:55 |
| mordred | I might try that | 04:55 |
| *** emagana has joined #openstack-infra | 04:57 | |
| *** dina_belova has quit IRC | 04:58 | |
| *** cody-somerville has quit IRC | 04:58 | |
| mordred | ok. I have constructed a "I need to deal with these reviews" link | 04:58 |
| mordred | and I'm going to try the star thing | 04:58 |
| *** vipul is now known as vipul-away | 05:02 | |
| *** vipul-away is now known as vipul | 05:02 | |
| *** rcleere has quit IRC | 05:10 | |
| *** vogxn has quit IRC | 05:20 | |
| *** blamar has quit IRC | 05:23 | |
| *** vogxn has joined #openstack-infra | 05:26 | |
| *** emagana has quit IRC | 05:34 | |
| openstackgerrit | Sergey Kolekonov proposed a change to openstack-infra/jenkins-job-builder: Existing plugins improvements https://review.openstack.org/41375 | 05:45 |
| *** nicedice_ has quit IRC | 05:46 | |
| *** SergeyLukjanov has quit IRC | 05:55 | |
| *** SergeyLukjanov has joined #openstack-infra | 05:56 | |
| *** SergeyLukjanov has quit IRC | 06:02 | |
| *** dguitarbite has joined #openstack-infra | 06:06 | |
| *** yaguang has quit IRC | 06:13 | |
| *** yaguang has joined #openstack-infra | 06:19 | |
| *** markmcclain has quit IRC | 06:27 | |
| *** odyssey4me has joined #openstack-infra | 06:28 | |
| Alex_Gaynor | There seem to be a few builds stuck in the queued state again. | 06:29 |
| *** yaguang has quit IRC | 06:32 | |
| *** ruhe has joined #openstack-infra | 06:33 | |
| *** dguitarbite has quit IRC | 06:36 | |
| *** odyssey4me has quit IRC | 06:45 | |
| *** afazekas has joined #openstack-infra | 06:50 | |
| *** yaguang has joined #openstack-infra | 06:52 | |
| *** dina_belova has joined #openstack-infra | 06:54 | |
| *** ruhe has quit IRC | 06:57 | |
| *** yaguang has quit IRC | 06:58 | |
| *** dina_belova has quit IRC | 06:58 | |
| *** dina_belova has joined #openstack-infra | 06:58 | |
| *** dina_belova has quit IRC | 07:03 | |
| *** afazekas_ has joined #openstack-infra | 07:08 | |
| *** nayward has joined #openstack-infra | 07:09 | |
| *** jhesketh has quit IRC | 07:12 | |
| *** afazekas_ has quit IRC | 07:13 | |
| *** yaguang has joined #openstack-infra | 07:17 | |
| *** Ryan_Lane has joined #openstack-infra | 07:37 | |
| *** vogxn has quit IRC | 07:47 | |
| *** jpich has joined #openstack-infra | 07:48 | |
| *** fbo_away is now known as fbo | 07:52 | |
| *** odyssey4me has joined #openstack-infra | 07:59 | |
| *** dina_belova has joined #openstack-infra | 07:59 | |
| *** dina_belova has quit IRC | 08:03 | |
| *** SergeyLukjanov has joined #openstack-infra | 08:04 | |
| *** vogxn has joined #openstack-infra | 08:04 | |
| *** odyssey4me has quit IRC | 08:06 | |
| *** odyssey4me has joined #openstack-infra | 08:07 | |
| *** dina_belova has joined #openstack-infra | 08:09 | |
| *** dina_belova has quit IRC | 08:14 | |
| *** locke105 has quit IRC | 08:26 | |
| *** fbo is now known as fbo_away | 08:26 | |
| *** derekh has joined #openstack-infra | 08:28 | |
| *** ruhe has joined #openstack-infra | 08:29 | |
| *** SergeyLukjanov has quit IRC | 08:44 | |
| *** fbo_away is now known as fbo | 08:46 | |
| *** SergeyLukjanov has joined #openstack-infra | 08:49 | |
| *** BobBall_Away is now known as BobBall | 08:51 | |
| *** psedlak has quit IRC | 08:56 | |
| *** morganfainberg is now known as morganfainberg_a | 09:10 | |
| *** mrmartin has joined #openstack-infra | 09:12 | |
| *** odyssey4me has quit IRC | 09:13 | |
| *** vogxn has quit IRC | 09:22 | |
| *** vogxn has joined #openstack-infra | 09:25 | |
| *** Ryan_Lane has quit IRC | 09:30 | |
| *** psedlak has joined #openstack-infra | 09:39 | |
| openstackgerrit | Mark McLoughlin proposed a change to openstack/requirements: Allow use of oslo.messaging 1.2.0a4 https://review.openstack.org/41696 | 09:46 |
| *** dstufft has quit IRC | 09:46 | |
| *** dstufft has joined #openstack-infra | 09:47 | |
| *** huangtianhua has joined #openstack-infra | 09:54 | |
| koolhead17 | mordred, https://review.openstack.org/#/c/41874/ <<--- needs immediate attention :) | 09:56 |
| *** w_ has joined #openstack-infra | 09:58 | |
| *** enikanorov has quit IRC | 10:00 | |
| *** olaph has quit IRC | 10:00 | |
| *** dina_belova has joined #openstack-infra | 10:10 | |
| *** dina_belova has quit IRC | 10:15 | |
| *** pcm__ has joined #openstack-infra | 10:36 | |
| *** odyssey4me has joined #openstack-infra | 10:41 | |
| *** pcm__ has quit IRC | 10:44 | |
| openstackgerrit | Petr Blaho proposed a change to openstack-infra/config: Adds Jenkins jobs for python-tuskarclient https://review.openstack.org/41887 | 10:44 |
| *** pcm__ has joined #openstack-infra | 10:44 | |
| *** vogxn has quit IRC | 10:44 | |
| *** koobs has quit IRC | 10:53 | |
| *** dina_belova has joined #openstack-infra | 10:55 | |
| *** yaguang has quit IRC | 10:57 | |
| *** enikanorov has joined #openstack-infra | 10:57 | |
| *** vogxn has joined #openstack-infra | 10:57 | |
| *** koobs has joined #openstack-infra | 11:00 | |
| *** koobs has quit IRC | 11:02 | |
| *** koobs has joined #openstack-infra | 11:02 | |
| *** yolanda has joined #openstack-infra | 11:20 | |
| yolanda | hi fungi, jeblair | 11:20 |
| yolanda | i moved my deployment to use geard instead of gearman | 11:20 |
| yolanda | but then i'm finding a problem, gate-noop job doesn't seem to be working | 11:20 |
| yolanda | and a telnet to geard with "status" shows no jobs, should i need to configure something by default? | 11:21 |
| *** fifieldt has quit IRC | 11:22 | |
| *** weshay has joined #openstack-infra | 11:22 | |
| *** sgviking has quit IRC | 11:30 | |
| *** sgviking has joined #openstack-infra | 11:31 | |
| *** boris-42 has joined #openstack-infra | 11:31 | |
| *** woodspa has joined #openstack-infra | 11:31 | |
| *** vogxn has quit IRC | 11:39 | |
| *** rfolco has joined #openstack-infra | 11:42 | |
| *** ArxCruz has joined #openstack-infra | 11:45 | |
| *** derekh has quit IRC | 11:50 | |
| *** psedlak_ has joined #openstack-infra | 11:54 | |
| *** markmc has joined #openstack-infra | 11:56 | |
| *** psedlak has quit IRC | 11:58 | |
| *** afazekas has quit IRC | 11:58 | |
| *** paul-- has joined #openstack-infra | 12:03 | |
| *** mikal has quit IRC | 12:06 | |
| *** afazekas has joined #openstack-infra | 12:07 | |
| *** mikal has joined #openstack-infra | 12:07 | |
| *** dprince has joined #openstack-infra | 12:08 | |
| *** HenryG has quit IRC | 12:11 | |
| *** Ryan_Lane has joined #openstack-infra | 12:11 | |
| *** prad_ has quit IRC | 12:12 | |
| *** adalbas has joined #openstack-infra | 12:15 | |
| *** afazekas is now known as afazekas_mtg | 12:15 | |
| *** w_ is now known as olaph | 12:17 | |
| *** HenryG has joined #openstack-infra | 12:23 | |
| *** SergeyLukjanov has quit IRC | 12:33 | |
| *** nayward has quit IRC | 12:38 | |
| *** nayward has joined #openstack-infra | 12:38 | |
| *** senk has joined #openstack-infra | 12:38 | |
| *** afazekas_mtg has quit IRC | 12:44 | |
| *** senk has quit IRC | 12:46 | |
| *** dina_belova has quit IRC | 12:51 | |
| *** weshay has quit IRC | 12:52 | |
| *** sandywalsh has quit IRC | 12:52 | |
| yolanda | seing that messages in zuul log: Unable to find change queue for project | 12:54 |
| yolanda | is that a normal issue? | 12:54 |
| *** dina_belova has joined #openstack-infra | 12:54 | |
| mordred | yolanda: did you change your config on jenkins to point at geard too? | 12:55 |
| yolanda | mordred, yes | 12:55 |
| mordred | you might also need to restart jenkins to cause it to run the registration code (I'm not sure, I'm just guessing) | 12:56 |
| *** dkliban has joined #openstack-infra | 12:59 | |
| *** SergeyLukjanov has joined #openstack-infra | 12:59 | |
| *** nayward has quit IRC | 13:00 | |
| yolanda | mordred, about the gate-noop that is solved, i hadn't added the gate-noop project in jenkins, now it works | 13:01 |
| yolanda | problem i have now is that gate pipeline isn't triggered, not sure why | 13:01 |
| *** psedlak__ has joined #openstack-infra | 13:01 | |
| mordred | ah - for that we'll have to wait for jeblair | 13:01 |
| yolanda | i have something like that: | 13:01 |
| yolanda | trigger: | 13:01 |
| yolanda | gerrit: | 13:01 |
| yolanda | - event: comment-added | 13:01 |
| yolanda | approval: | 13:01 |
| yolanda | - approved: 1 | 13:01 |
| yolanda | but although i approve some change, that isn't queued into to gate queue | 13:02 |
| mordred | yolanda: and you have a manager: field for that pipeline? | 13:03 |
| yolanda | manager: DependentPipelineManager | 13:04 |
| mordred | hrm. yeah. sorry - I'm no use - jeblair or fungi will be more help | 13:04 |
| yolanda | thx anyway :) | 13:05 |
| *** weshay has joined #openstack-infra | 13:05 | |
| *** nayward has joined #openstack-infra | 13:05 | |
| *** weshay has quit IRC | 13:05 | |
| *** psedlak_ has quit IRC | 13:05 | |
| *** sandywalsh has joined #openstack-infra | 13:06 | |
| *** weshay has joined #openstack-infra | 13:06 | |
| *** dkehn has quit IRC | 13:10 | |
| *** dkehn has joined #openstack-infra | 13:10 | |
| *** afazekas has joined #openstack-infra | 13:11 | |
| *** derekh has joined #openstack-infra | 13:13 | |
| *** blamar has joined #openstack-infra | 13:19 | |
| fungi | let's see if i'm of any use | 13:21 |
| fungi | gimme a sec to catch up on scrollback... i was coaching christine on prepping for a job interview she has coming up with one of red hat's marketing teams dealing with their openstack stuff | 13:23 |
| * fungi crosses fingers | 13:23 | |
| *** changbl has quit IRC | 13:25 | |
| *** pentameter has joined #openstack-infra | 13:25 | |
| *** mriedem has joined #openstack-infra | 13:25 | |
| *** rnirmal has joined #openstack-infra | 13:29 | |
| *** ryanpetrello has joined #openstack-infra | 13:31 | |
| *** dina_belova has quit IRC | 13:32 | |
| fungi | yolanda: the "unable to find change queue for project" message is potentially a benign warning. we see it when trying to match events against the silent pipeline, if memory serves | 13:32 |
| yolanda | fungi, main problem i had is that gate pipeline isn't working, so i was looking for all the potential problems | 13:33 |
| fungi | yolanda: if status shows no jobs, that may mean your jenkins isn't reporting the jobs it knows it can run into the geard? | 13:33 |
| fungi | i'm still a little fuzzy on the gear interface | 13:33 |
| yolanda | if i remove the approval: | 13:33 |
| yolanda | - approved: 1 conditional from trigger it works, just when i send some comment, and pushes job to jenkins | 13:33 |
| fungi | ahh, i see you got past that. still reading through what you've written | 13:34 |
| yolanda | but for some reason it doesn't like that filter, isn't that correct? | 13:34 |
| mordred | yolanda: have you added the approval column to your gerrit? | 13:34 |
| fungi | i think gerrit won't give a radio selection widget for approve in the interface if you don't? | 13:35 |
| *** burt has joined #openstack-infra | 13:35 | |
| yolanda | mordred, what i do is that i send a +2 to code review | 13:35 |
| mordred | yup. so that will be a different event | 13:35 |
| mordred | instead of approval | 13:35 |
| yolanda | mordred, do you have some doc for that events? | 13:35 |
| * fungi dredges up a link to the database update queries to add the column | 13:35 | |
| mordred | you'll want, um, I think code-review | 13:36 |
| mordred | fungi: I don't think they need the approval column per-se | 13:36 |
| *** nayward has quit IRC | 13:36 | |
| fungi | oh | 13:36 |
| mordred | yolanda: try changing to | 13:36 |
| mordred | - event: comment-added | 13:36 |
| mordred | codereview: | 13:37 |
| mordred | - codereview: 2 | 13:37 |
| fungi | ahh, i see they want to treat cdrv +2 as approval | 13:37 |
| yolanda | let me try | 13:37 |
| mordred | yah. that's the 'normal' gerrit approach | 13:37 |
| * fungi gets tunnel vision dealing with our hacked-up gerriting all the time | 13:38 | |
| *** nayward has joined #openstack-infra | 13:38 | |
| mordred | fungi: well, I'm watching stream-events right now | 13:39 |
| yolanda | mm, layout.yaml error | 13:39 |
| mordred | fungi: if you wanted to approve https://review.openstack.org/#/c/41832/ | 13:40 |
| mordred | fungi: I could watch what the vote looks lke | 13:40 |
| mordred | ok. | 13:41 |
| mordred | it seems lke it wants to be | 13:41 |
| mordred | approvals: | 13:41 |
| fungi | mordred: revewed and approved 41832 | 13:41 |
| fungi | er, reviewed | 13:41 |
| mordred | - code review: 2 | 13:41 |
| mordred | but I don't think a space there is avlid in yaml is it? | 13:42 |
| mordred | oh. it is | 13:42 |
| mordred | neat | 13:42 |
| yolanda | fungi, ok, fixed | 13:43 |
| yolanda | gerrit: | 13:43 |
| yolanda | - event: comment-added | 13:43 |
| yolanda | approval: | 13:43 |
| yolanda | - code-review: 2 | 13:43 |
| openstackgerrit | A change was merged to openstack-infra/config: Actually let os-*-config release to PyPI https://review.openstack.org/41832 | 13:43 |
| yolanda | that's the right syntax | 13:43 |
| *** mestery_ has joined #openstack-infra | 13:43 | |
| fungi | too cool. this could use some examplishness in the zuul docs, methinks | 13:43 |
| fungi | also, SpamapS, you should be able to tag releases and get them uploading automatically now, finally :/ | 13:44 |
| *** mestery has quit IRC | 13:46 | |
| mordred | fungi: at this point, there is no way for us to manually trigger just the pypi upoad job is there? | 13:47 |
| fungi | mordred: in fact there is | 13:47 |
| fungi | mordred: jeblair wrote an awesome quick-n-dirty script | 13:47 |
| mordred | neat! because the tag is already pushed and the tarball already uploaded | 13:47 |
| yolanda | fungi, do you have some sample of the gate-noop jenkins task? i just created a dummy shell project, but it complains with Reported result: ERROR, although the job is executed correctly | 13:48 |
| fungi | basically uses the gear library to connect to the gearman server and request the job be run with the right set of params | 13:48 |
| *** prad_ has joined #openstack-infra | 13:48 | |
| fungi | yolanda: sure, just a sec | 13:48 |
| yolanda | thx | 13:48 |
| fungi | yolanda: https://git.openstack.org/cgit/openstack-infra/config/tree/modules/openstack_project/files/jenkins_job_builder/config/misc.yaml#n1 | 13:49 |
| fungi | oh, wait, wrong file | 13:49 |
| yolanda | we are not using jenkins_job_builder at the moment, just creating some jobs manually on jenkins for testing | 13:50 |
| fungi | huh, actually i don't see any more specific scripting in the yaml for it. i'll pastebin the xml | 13:51 |
| fungi | yolanda: http://paste.openstack.org/show/44118/ | 13:53 |
| *** cppcabrera has joined #openstack-infra | 13:54 | |
| *** huangtianhua has quit IRC | 13:54 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Add job to propose tag merges https://review.openstack.org/41927 | 13:54 |
| fungi | i guess that's what we get from jjb when all we specify is a job name and a node label | 13:54 |
| mordred | ttx, fungi ^^ turns out it's really easy | 13:54 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Add job to propose tag merges https://review.openstack.org/41927 | 13:55 |
| mordred | oops. forgot to add the bug tag | 13:55 |
| fungi | aha! so that fixes the versioning i guess... you tried this approach with the swift merge back to master? | 13:55 |
| *** cppcabrera has left #openstack-infra | 13:56 | |
| *** mestery_ is now known as mestery | 13:57 | |
| yolanda | fungi, i was just trying to execute some shell job | 13:57 |
| yolanda | that's a bit different | 13:58 |
| mordred | fungi: I did - although I mostly did it by hand | 13:58 |
| fungi | ttx: on the topic of capping folsom dependencies, i just realized another hole in the idea... the last time it broke was because we released a new client which broke backward compatibility and was getting sucked into the folsom tests--but we *do* want to catch that right? so we should not cap our own projects when listed as dependencies and let those breaks pop up? | 13:58 |
| *** markmc has quit IRC | 13:59 | |
| mordred | yes. that's correct | 13:59 |
| fungi | unfortunately that took a couple months for someone to care enough about folsom to fix it. capping outside dependencies is not going to save us from ourselves :/ | 14:00 |
| mordred | yup | 14:01 |
| mordred | and nope | 14:01 |
| fungi | did we get anywhere with backward-compat gating jobs for changes to our clients/libs? | 14:01 |
| *** dina_belova has joined #openstack-infra | 14:01 | |
| *** markmcclain has joined #openstack-infra | 14:01 | |
| fungi | unfortunately, like the requirements testing, the holy grail would be to calculate reverse-dependencies of the project being changed and run all the tests associated with those projects | 14:02 |
| fungi | on all affected branches | 14:03 |
| mordred | nope. we've gotten nowhere with those | 14:03 |
| mordred | I think it could maybe be done with a flag to d-g | 14:04 |
| mordred | so that it uses different logic when figuring out the branches of the other projects to pull | 14:05 |
| fungi | well, i'll get changes proposed to cap all of the third-party dependency tree for each project supported in folsom and hope for the best, for now | 14:05 |
| mordred | zomg | 14:06 |
| mordred | I just realized I think it's easy | 14:06 |
| mordred | new d-g job that simply overrides ZUUL_BRANCH | 14:06 |
| mordred | proposed changes get pulled based on their refs... | 14:06 |
| mordred | so they wou;dn't be affected | 14:06 |
| mordred | and we only add that job to the client libs | 14:07 |
| fungi | oh, and then override the branch so the remainder of the projects integrated get chosen correctly and anchored to that branch | 14:07 |
| mordred | yup | 14:07 |
| mordred | so we make a job that overrides ZUUL_BRANCH=stable/grizzly | 14:07 |
| fungi | d-g bases the services/projects list on ZUUL_BRANCH right? | 14:07 |
| mordred | and one that does stable/folsom | 14:07 |
| mordred | yup | 14:07 |
| fungi | i think it might work | 14:08 |
| fungi | that'll at least get you devstack-tempest covered for backward-compat. doesn't directly help unit testing | 14:08 |
| fungi | but on the other hand, broken unit tests are going to be specific to one project, so they won't hold up every integrated project on that branch while waiting for a fix | 14:10 |
| fungi | so i guess that's less of a risk to stable branch management overall | 14:10 |
| mordred | also, I'm more and more of the opinion that we should be doing unitests on devstack nodes | 14:10 |
| *** mrodden has joined #openstack-infra | 14:11 | |
| mordred | fungi: also, it means we'll be running stable branches more regularly - so the periodic jobs will probably cease being important | 14:11 |
| annegentle | hey guys, I think I have a fundamental conceptual gap for spinning up a slave server... sounds like a crisis eh? | 14:11 |
| annegentle | :) | 14:11 |
| mordred | it does! | 14:11 |
| *** locke105 has joined #openstack-infra | 14:11 | |
| annegentle | so I have a Rackspace Cloud server. | 14:11 |
| fungi | annegentle: it is no small feat | 14:11 |
| annegentle | My thinking is I just "turn it into" a slave server | 14:11 |
| annegentle | so I made a local.pp | 14:12 |
| annegentle | turned my cloud server into a puppet server | 14:12 |
| annegentle | ran the local.pp | 14:12 |
| fungi | sounds right so far | 14:12 |
| annegentle | but then what? Would the scripts be available in usr/? | 14:12 |
| *** cody-somerville has joined #openstack-infra | 14:13 | |
| *** cody-somerville has quit IRC | 14:14 | |
| *** cody-somerville has joined #openstack-infra | 14:14 | |
| fungi | annegentle: so, for example with a server destined to run python 2.7-based jobs we use an ubuntu 12.04 lts (precise pangolin) vm and instantiate openstack_project::slave for it | 14:14 |
| jeblair | mordred: why do you want to run unit tests on devstack nodes? | 14:14 |
| annegentle | fungi: ok so maybe I don't want to "puppet apply" the local.pp I made? | 14:15 |
| fungi | annegentle: something along these lines (but you'd have variable assignments in place of hiera lookups)... https://git.openstack.org/cgit/openstack-infra/config/tree/manifests/site.pp#n516 | 14:15 |
| jeblair | annegentle, fungi: we also have this script: https://git.openstack.org/cgit/openstack-infra/config/tree/install_jenkins_slave.sh | 14:15 |
| fungi | ahh, good point. why do i keep forgetting about install_jenkins_slave.sh (rhetorical question) | 14:16 |
| annegentle | jeblair: fungi: maybe I should try the install_jenkins_slave.sh then | 14:16 |
| jeblair | fungi: though i wonder if it works now that the 27/33 stuff has made things more complicated? | 14:16 |
| fungi | annegentle: that is probably what you want, yes | 14:16 |
| *** ruhe has quit IRC | 14:16 | |
| annegentle | jeblair: fungi: ok trying it now, but I get "host is missing hostname and/or domain | 14:17 |
| fungi | jeblair: i think the py33 bits shouldn't complicate it unless you specifically want to test python 3.3 since it depends on a variable being added and set to true | 14:17 |
| annegentle | fungi: and clarkb told me what to change but I guess I didn't do it right? /etc/hosts and /etc/hostnames? | 14:17 |
| jeblair | fungi: makes sense | 14:17 |
| * annegentle is grasping at straws | 14:18 | |
| fungi | annegentle: well, the hostname shouldn't matter so much in this case, depending on what you're trying to do | 14:18 |
| fungi | but in theory you should just be able to create an ubuntu precise vm in rackspace, log into it, pull down the current install_jenkins_slave.sh and run it as root | 14:19 |
| * fungi tries that to see if it's still working as intended | 14:20 | |
| annegentle | fungi: okie, trying it now | 14:20 |
| *** anteaya has joined #openstack-infra | 14:20 | |
| mordred | jeblair: cause I'm crazy. ignore me | 14:21 |
| mordred | jeblair: in other news- python-swiftclient does not participate in the devstack-gate | 14:21 |
| fungi | ...yet? ;) | 14:21 |
| jeblair | mordred: neat. i don't think we've quite fully integrated django-openstack-auth either | 14:21 |
| mordred | also, neither does ceilometerclient | 14:22 |
| mordred | jeblair: did you see my idea above on how to run trunk client with stable branches? am I missing a clear logic hole? | 14:22 |
| jeblair | mordred: lemme reread and think more. i'm pre-breakfast here. | 14:23 |
| mordred | jeblair: ok. I'm about to send up a patch that might explain it more succinctly | 14:23 |
| *** datsun180b has joined #openstack-infra | 14:24 | |
| BobBall | :( Can I bang my head on something? | 14:25 |
| yolanda | fungi, jeblair, problem i have now is that job is executed correctly in jenkins, but it seems like it doesn't report it back to zuul | 14:25 |
| BobBall | https://bugs.launchpad.net/devstack/+bug/1212280 - python-crypto is actually broken on xenserver-core because it's now not uninstalled... | 14:25 |
| uvirtbot | Launchpad bug 1212280 in devstack "python-crypto mismatch on CentOS 6.4 when using XenServer-Core" [Undecided,New] | 14:25 |
| yolanda | there is an error in jenkins like: Reported result: ERROR | 14:25 |
| yolanda | although job ends with SUCCESS | 14:26 |
| yolanda | is there some extra config needed for jobs or for jenkins communication, rather than having gearman plugin installed? | 14:26 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Gate current clients on stable branches https://review.openstack.org/41931 | 14:27 |
| mordred | jeblair: ^^ there you go (you too clarkb and fungi) | 14:27 |
| *** adalbas has quit IRC | 14:27 | |
| mordred | I _think_ that might just do what we want | 14:27 |
| *** SergeyLukjanov has quit IRC | 14:28 | |
| BobBall | mordred: any thoughts on other ways to get the python-crypto/python-lxml mess sorted on xenserver-core? a fresh install shows that your patch at https://review.openstack.org/#/c/40431/ doesn't work after all :/ (sorry - I should have tested from fresh rather than on the system that was close-to-working) | 14:28 |
| mordred | BobBall: yes. my main thoughts in devstack right now are single-global-venv | 14:28 |
| * fungi wonders if xenserver works in/with a virtualenv | 14:29 | |
| BobBall | That sounds like fun | 14:29 |
| *** SergeyLukjanov has joined #openstack-infra | 14:29 | |
| BobBall | why might it not fungi? | 14:29 |
| fungi | eh, i don't currently fathom the depths to which xenserver is something nova runs in vs on vs manages vs is managed by | 14:30 |
| mordred | BobBall: here is what I'm at with it so far https://review.openstack.org/#/c/40534/ | 14:30 |
| mordred | BobBall: it's not fully operational, as you can see | 14:30 |
| BobBall | haha fungi :) Got it :) Xenserver-core is the thing I'm worried about at the moment - which is just running in a CentOS6.4 at the same level as devstack | 14:31 |
| jeblair | mordred: i think there's a problem with that -- if the queue is A[nova/master] <- B[novaclient/master], then to test B you will checkout grizzly projects and checkout the zuul ref for B. but you will also checkout the zuul ref for A, which is not on the grizzly branch. | 14:31 |
| BobBall | i.e. no different architecturally to libvirt | 14:31 |
| mordred | jeblair: I didn't add it to the server projects | 14:31 |
| mordred | jeblair: so the queue should only be clientlib projects | 14:31 |
| fungi | BobBall: ahh, okay, so there are no xenserver tools which would need to be able to run within the context of the same virtualenv or anything | 14:31 |
| jeblair | mordred: however, the branch _is_ encoded in the zuul ref, so you can probably update the devstack-gate setup logic to ignore zuul refs in that circumstance | 14:32 |
| jeblair | mordred: i didn't say that you did. :) | 14:32 |
| BobBall | fungi: not AFAIK, no. All of the calling will go out of the venv and nothing needs to then run from within it. | 14:32 |
| jeblair | mordred: there is a single queue for all changes to all devstack projects | 14:32 |
| mordred | jeblair: of the same name | 14:32 |
| fungi | BobBall: okay, sounds like it should work then | 14:32 |
| jeblair | mordred: wha? | 14:32 |
| BobBall | fungi: because we're coming from a Dom0/DomU environment I think we can guarantee that separation more than most others ;) | 14:32 |
| mordred | jeblair: the single global virtual queue thing is about job name maching, no? | 14:33 |
| jeblair | mordred: look at http://status.openstack.org/zuul/ | 14:33 |
| fungi | BobBall: yeah, makes sense | 14:33 |
| jeblair | mordred: you will see that it includes server projects (cinder, horizon) as well as client projects (python-cinderclient) | 14:33 |
| mordred | jeblair: yes. I undersatnd that. can you explain how it constructs that queue - because clearly my undersatnding is erroneous | 14:34 |
| mordred | it's projects sharing a job, right? | 14:34 |
| jeblair | mordred: all projects that share at least one job have their queues merged. | 14:34 |
| mordred | AHHHH | 14:34 |
| mordred | k | 14:34 |
| mordred | with you | 14:34 |
| jeblair | mordred: (because that means they have the ability to affect each others outcome in each of those jobs) | 14:35 |
| mordred | yup. so I'd probably also want to add a flag or something that lets d-g know I'm doing this | 14:35 |
| mordred | and then add logic to d-g to strip out changes... hrm. that just became way more complex. dammit | 14:36 |
| jeblair | i think that's about it though. | 14:36 |
| annegentle | fungi: I see the same error on a fresh precise server, abotu the fqdn | 14:36 |
| mordred | because I can't tell it that I want it to strip out any master changes | 14:36 |
| mordred | because master changes to other client libs in the queue we'd want to keep | 14:37 |
| fungi | annegentle: can you paste the error? | 14:37 |
| mordred | annegentle: in addition to editing /etc/hosts | 14:37 |
| mordred | annegentle: you should run sudo hostname my.new.fqdn.com | 14:37 |
| mordred | annegentle: also, you should join me in yelling and your bosses and my bosses about providing a cloud that sets fdqn's when it spins up hosts | 14:38 |
| mordred | because it's pretty much exactly zero times when you explicitly want an invalid hostname | 14:38 |
| mordred | jeblair: do you think it would be super ugly to have the flag to d-g be a list of projects that should not have their master branch refs stripped? | 14:39 |
| annegentle | mordred: heh. but help me understand fqdns... so I can yell appropriately. I have to hand-edit /etc/hosts? | 14:39 |
| ttx | fungi: clearly we won't catch all cases.. But if we can get rid of the stupid ones maybe there will be more focus on the really interesting ones | 14:39 |
| mordred | annegentle: main thing is, you need to tell the current running server what its hostname is - so you need to run the hostname comand with an argument to set the hostname | 14:39 |
| fungi | ttx: yes, i think once mordred polishes https://review.openstack.org/41931 that will address the concern | 14:40 |
| annegentle | mordred: ok that definitely gets me further. I'll see if I can write this up | 14:40 |
| ttx | fungi: indeed. It's about the only way to preserve backward compat anyway | 14:41 |
| jeblair | mordred: i'm not sure i fully comprehend the case you're concerned about... but one second. | 14:41 |
| fungi | ttx: i'm feeling much better about our chances of keeping it stable at this point, actually | 14:41 |
| jeblair | mordred, fungi, ttx: it's worth mentioning that this would only test library source code api backwards-compatibility, not network api backwards-compatibility (which is also something we want to test) | 14:42 |
| BobBall | mordred: Does that patch depend on anything else? "pip_install pip" fails because it's not running things as root. cache_download (pip/util.py) tries to copy the cache file from /var/cache/pip/... which is owned by root. | 14:42 |
| jeblair | just wanted to make sure we're clear on that and don't forget the other thing by using terms losely. | 14:42 |
| mordred | BobBall: yes. it does, but both of those things should be self contained | 14:42 |
| mordred | jeblair: totally. purely library backwards compat | 14:43 |
| ttx | jeblair: ok | 14:43 |
| fungi | jeblair: our biggest pain point recently on stable/folsom was when a new version of a client was released which was not backward-compatible, so devstack-tempest no longer worked and we couldn't land security fixes for folsom for a couple months until someone cared enough to fix the client lib | 14:43 |
| *** adalbas has joined #openstack-infra | 14:43 | |
| mordred | right. and I believe that a working version of this would prevent that from happening | 14:44 |
| jeblair | fungi: yeah, i know, i'm not knocking this -- i just want to avoid us saying "oh, yeah, we test backwards compat. totally." | 14:44 |
| mordred | :) | 14:44 |
| fungi | so preventing changes from landing which could put us in that position, hopefully | 14:44 |
| jeblair | this is really important. | 14:44 |
| * fungi agrees | 14:44 | |
| fungi | i mean, we also acknowledge that we probably won't (any time soon, maybe ever) test that a change to project 1 doesn't break unit tests for project 2 depending on it, much less across dev and stable branches | 14:46 |
| mordred | fungi, jeblair ^^ that, actually, is why I'm crazy and think we should just start running unittests on a prepped devstack node | 14:46 |
| mordred | but seriously, ignore me - too many other things right now | 14:47 |
| jeblair | mordred: the set of projects where you don't want to mask their zuul refs is equal to the set of projects that only have a master branch, right? | 14:47 |
| mordred | jeblair: ah. yes. good call | 14:47 |
| jeblair | mordred: 'this project does not have this branch; therefore use master' is already a logic bit in d-g; you could re-use that to avoid the mask operation | 14:48 |
| mordred | jeblair: ++ | 14:48 |
| mordred | jeblair: can you point mein the right direction in d-g to where this happens? setup_workspace doesn't seem to be the right place | 14:48 |
| jeblair | mordred: in fact, because it sets BRANCH=master in that case, depending on exactly how you implement the branch mask, it may already be taken care of? | 14:48 |
| *** emagana has joined #openstack-infra | 14:49 | |
| jeblair | mordred: setup_workspace | 14:49 |
| *** jjmb has joined #openstack-infra | 14:49 | |
| jeblair | mordred: https://git.openstack.org/cgit/openstack-infra/devstack-gate/tree/devstack-vm-gate-wrap.sh#n105 | 14:49 |
| mordred | yup. that's what I'm loking at | 14:50 |
| mordred | looking | 14:50 |
| mordred | I grok how this gets the right ref for the current change to be tested | 14:50 |
| mordred | I don't grok how it handles the previous projects in the queue | 14:50 |
| jeblair | mordred: zuul sets a ref for every project with a change | 14:50 |
| mordred | but there is only one ZUUL_REF variable | 14:51 |
| jeblair | mordred: it sets the same ref. :) | 14:51 |
| jeblair | mordred: i will type more, be patient: | 14:51 |
| *** pcrews has joined #openstack-infra | 14:52 | |
| *** dina_belova has quit IRC | 14:52 | |
| jeblair | mordred: so change A to project foo causes foo:refs/zuul/Z1 to be created; change B to project bar causes foo:refs/zuul/Z2, bar:refs/zuul/Z2 to be created | 14:52 |
| mordred | gotcha. I follow now | 14:52 |
| jeblair | mordred: so all devstack gate has to do is say "i'm supposed to test Z2, i'll check it out on every project that has one" | 14:53 |
| *** markmcclain has quit IRC | 14:53 | |
| Alex_Gaynor | Just a heads up, there are a few jobs stuck in queued on the zuul status page, probably need to be killed / cycled / whatevered. | 14:54 |
| fungi | jeblair: on a separate note, Alex_Gaynor pointed out earlier that we have accumulated a couple more jobs stuck in a queued state in zuul... is that what the latest gearman-plugin patch (41488) is supposed to solve? | 14:56 |
| jeblair | mordred: i _think_ (and i really need breakfast before i can say this for sure) that if you perform the mask operation using the $BRANCH variable after it's reset on line 109, you probably don't have to worry about the clientlib case. | 14:56 |
| BobBall | mordred: /var/cache/pip is only accessible by root in the generic case - so I think that your change requires PIP_DOWNLOAD_CACHE be set to a new directory that the stack user can create? do you agree? | 14:56 |
| fungi | heh, clearly i had you on my mind, Alex_Gaynor | 14:56 |
| openstackgerrit | Anne Gentle proposed a change to openstack-infra/config: Adds info about building a Jenkins slave with script https://review.openstack.org/41942 | 14:56 |
| mordred | jeblair: yes. I think I see a simple way to do it | 14:56 |
| jeblair | fungi: yes; we should try rolling that out today. | 14:57 |
| mordred | BobBall: as in, it's only READABLE by root? | 14:57 |
| fungi | k | 14:57 |
| BobBall | mordred: no, writeable - but that means you can't store the pip tarball in there running as the stack user | 14:57 |
| BobBall | mordred: and if you had already downloaded it as root then the file is likely to be read-only (at least it was in my setup :) ) | 14:58 |
| mordred | BobBall: yes. I follow you now | 14:58 |
| mordred | I thought I'd put in a patch ot help with that - but might have missed it | 14:58 |
| *** dina_belova has joined #openstack-infra | 14:58 | |
| mordred | the idea was to copy /var/cache/pip to somewhere writable by the stack user and then set PIP_DOWNLOAD_CACHE to that locaiton | 14:58 |
| mordred | and if /var/cache/pip happens to be empty, then an empty dir forthe stack user will e just fine :) | 14:59 |
| *** _TheDodd_ has joined #openstack-infra | 14:59 | |
| SpamapS | fungi: thanks, do I have to manually push 0.1.0 ? | 15:00 |
| BobBall | yup - I can fix it for me like that, but I guess we need a change to the patch (or a depends on another one) | 15:00 |
| SpamapS | fungi: or can I delete the tag, re-ad, re-push, and have it do something meaningful? | 15:00 |
| fungi | ttx: thinking through workflow, if i had capped, say, horizon stable/folsom's requirements yesterday we'd have set Django>=1.4,<=1.4.5 but today a security fix 1.4.6 was released. would we need to keep tabs on announcements and adjust pip-requires accordingly for the remainder of the support period of folsom? | 15:01 |
| *** SergeyLukjanov has quit IRC | 15:01 | |
| fungi | SpamapS: i'll need to run a script to trigger the job for you | 15:02 |
| fungi | SpamapS: now that it actually exists | 15:02 |
| BobBall | mordred: there is a change that sets it to ~/pipcache - but this is after pip_install pip (which fails for me) | 15:02 |
| jeblair | fungi: will you help me with some clicking? | 15:02 |
| fungi | jeblair: gladly | 15:02 |
| *** SergeyLukjanov has joined #openstack-infra | 15:03 | |
| jeblair | fungi: would you mind logging into jenkins01? | 15:03 |
| mordred | BobBall: ah. then I just suck | 15:03 |
| fungi | jeblair: lemme get my clicky finger warmed up. logging in | 15:03 |
| *** ruhe has joined #openstack-infra | 15:03 | |
| ttx | fungi: I'm not sure who that would serve... | 15:03 |
| ttx | fungi: I'd rather react to distros asking us to bump so that they don't get caught when upgrading deps... maybe | 15:03 |
| fungi | ttx: okay, so we just leave the cap there as a historical artifact and don't need to test that 1.4.6 broke nothing i guess | 15:03 |
| jeblair | fungi: in a minute (not just yet), i'd like you to manually abort all of the running jobs (except the ones that start with "ceilometer-") | 15:04 |
| fungi | jeblair: okay | 15:04 |
| fungi | jeblair: i'm there, so just say when | 15:04 |
| ttx | fungi: right. FWIW distros would try to backport security patches only in that 1.4.6 | 15:04 |
| jeblair | just waiting for a change to merge | 15:04 |
| fungi | k | 15:04 |
| *** emagana has quit IRC | 15:04 | |
| ttx | so we might actually serve them better by NOT upgrading ours | 15:04 |
| fungi | ttx: and you're good with leaving the lower bound untouched but changing/adding the upper bound on everything to <= today's releases | 15:05 |
| fungi | (except for things we release) | 15:05 |
| fungi | ttx: and as for new transitive dependency entries i add in the process (so we can cap those) should i bother specifying a lower bound? | 15:06 |
| openstackgerrit | A change was merged to openstack/requirements: Allow use of oslo.messaging 1.2.0a4 https://review.openstack.org/41696 | 15:06 |
| jeblair | zuul is stopped | 15:07 |
| fungi | because if so, i'm not really sure what a meaningful one would be. maybe >=0 | 15:07 |
| jeblair | fungi: please proceed | 15:07 |
| fungi | jeblair: okay, so start clicking through? | 15:07 |
| fungi | on it | 15:07 |
| jeblair | fungi: oh, don't abort devstack-launch jobs | 15:07 |
| *** emagana has joined #openstack-infra | 15:07 | |
| fungi | ah, oops. i only did one i think | 15:08 |
| jeblair | is ok | 15:08 |
| *** mrmartin has quit IRC | 15:09 | |
| ttx | fungi: probably not | 15:09 |
| jeblair | fungi: let me know when you're done | 15:11 |
| fungi | will do--still clicking | 15:11 |
| jeblair | (i'll pitch in too) | 15:12 |
| *** markmcclain has joined #openstack-infra | 15:13 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/devstack-gate: Add ability to do smart backwards compat testing https://review.openstack.org/41945 | 15:13 |
| fungi | kill the launch jobs too, or no? | 15:13 |
| jeblair | fungi: nope | 15:13 |
| jeblair | all done | 15:14 |
| fungi | i think that's all then | 15:14 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Gate current clients on stable branches https://review.openstack.org/41931 | 15:14 |
| fungi | so once these finish, you'll be able to restart zuul? | 15:14 |
| mordred | jeblair: after breakfast, I think those two changes together might do it | 15:14 |
| fungi | gah, two keystone jobs popped up, so knocked them down | 15:15 |
| *** pabelanger has quit IRC | 15:15 | |
| jeblair | fungi: that's cause i restarted zuul | 15:15 |
| fungi | oh, oops | 15:15 |
| jeblair | i'll abort the rest and recheck that change | 15:15 |
| fungi | i thought i had simply missed those two jobs down at the bottom of the list. my bad. didn't realize zuul was already back up | 15:16 |
| *** mkerrin has joined #openstack-infra | 15:19 | |
| *** Protux has quit IRC | 15:21 | |
| *** Protux has joined #openstack-infra | 15:21 | |
| jeblair | mordred: you should recheck those changes (zuul was down) | 15:22 |
| *** changbl has joined #openstack-infra | 15:23 | |
| mordred | jeblair: I will | 15:24 |
| mordred | speaking of - I had an idea just now walking for coffee | 15:24 |
| mordred | what if - instead of the huge amount of non-voting jobs we have - we added a feature similar to recheck that would allow someone to request a non-voting job be run against their patch | 15:25 |
| mordred | so something like "please-tests gate-devstack-vm-tempest-cells" | 15:25 |
| jeblair | i'm not sure we have a huge number of non-voting jobs; i think we have 4, one of which is about to flip. | 15:26 |
| jeblair | mordred: but yes, we could do that. | 15:26 |
| mordred | by huge - I meant huge number of runs of those jobs that are not cared about :) | 15:26 |
| jeblair | ah, that, yes. :) | 15:26 |
| mordred | more to enable the people actually working on those issues to get real gate feedback | 15:27 |
| jeblair | mordred: we could have an 'experimental' pipeline, and if you leave a comment, it triggers that. | 15:27 |
| mordred | yes. that would actually be even easier - and so what if you also get another experimental test | 15:27 |
| mordred | (I was thinking about this while thinking about adding two more devsack runs to every client change :) ) | 15:28 |
| jeblair | exactly. i think there's a special flag you have to set to get it to leave a comment but not vote. | 15:28 |
| * mordred seems to be hacking on an area of our testing that he does not normally hack on a lot this morning | 15:30 | |
| * mordred wonders what he ate... | 15:30 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Add missing tests to swiftclient https://review.openstack.org/41951 | 15:32 |
| *** Shrews has quit IRC | 15:35 | |
| *** Shrews has joined #openstack-infra | 15:35 | |
| *** mrodden has quit IRC | 15:37 | |
| *** emagana has quit IRC | 15:40 | |
| *** ruhe has quit IRC | 15:43 | |
| *** mrodden has joined #openstack-infra | 15:46 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Move non-voting jobs to an experimental queue https://review.openstack.org/41954 | 15:48 |
| mordred | jeblair: ^^ there's a stab at that | 15:48 |
| fungi | ooh, the docs team is looking at doing parallel manual rebuilds and seeing some serious speed-up. it'll be nice to see those not running nearly so long in the future... http://lists.openstack.org/pipermail/openstack-docs/2013-August/002509.html | 15:48 |
| jeblair | gah | 15:49 |
| jeblair | the new right-alignment sometimes causes every job result to shift down a row. | 15:50 |
| *** jergerber has joined #openstack-infra | 15:51 | |
| BobBall | mordred: Why are you setting up the virtualenv with --system-site-packages? This is causing me problems because I've got an env where pip has installed things previously (and broken the python-crypto library) | 15:51 |
| openstackgerrit | Yegor Minin proposed a change to openstack-infra/gerritlib: Add option to query commit messages https://review.openstack.org/38865 | 15:51 |
| fungi | BobBall: i think nova needs it for libvirt, since pip can't install that into the virtualenv? | 15:53 |
| BobBall | Poo | 15:53 |
| fungi | BobBall: though i may be misunderstanding the question | 15:53 |
| BobBall | understood | 15:54 |
| fungi | and apparently there's been little interest from libvirt upstream in generating a setup.py-able python-libvirt with the necessary library bindings so that it can be uploaded to pypi as part of their release automation | 15:55 |
| fungi | (it gets built directly out of their c lib sources currently) | 15:56 |
| jeblair | which is interesting, considering who the libvirt authors are | 15:56 |
| BobBall | Do you think breaking devstack for libvirt would encourage them to fix it? :) | 15:57 |
| mordred | BobBall: no | 15:57 |
| BobBall | shame | 15:57 |
| mordred | it's an upstream thing and a HUGE engineering issue | 15:58 |
| mordred | that said - if we _start_ from a venv-based machine | 15:58 |
| mordred | we shouldn't get you in to broken state in the first place | 15:58 |
| mordred | and if we pip install -U everything, we shoudl install what we ask for into the venv even if there is something in the system location | 15:58 |
| mordred | so it _should_ actually have the result that we're wanting | 15:58 |
| mordred | (other than you having a broken base system to start from :) ) | 15:59 |
| BobBall | I think the problem is that the pycrypto in the system libs was broken but new enough - so the venv didn't think it needed it's own one | 15:59 |
| _TheDodd_ | Greetings, jeblair! I hope you are well. anteaya mentioned something to the affect that you have some thoughts on the recent jeepyb changes. | 15:59 |
| BobBall | (i.e. broken in the python-crypto vs pip pycrypto sense) | 15:59 |
| mordred | BobBall: right. so in the future, it will do-the-right-thing and not break it | 15:59 |
| mordred | BobBall: you can try uninstall/re-installing the yum version to get the system version working | 16:00 |
| mordred | to test that | 16:00 |
| BobBall | yeah - it's just a shame that we can't rely on an isolated system | 16:00 |
| mordred | totally agree | 16:00 |
| BobBall | i.e. we can't use virtual environments to be confident that it'll work irrespective of how broken the outside world is | 16:00 |
| mordred | I would welcome redhat fixing the build system around libvirt | 16:00 |
| jeblair | _TheDodd_: i don't think so; i think there was some suspicion that there was something wrong with the backwards-compatability, but that turned out to be incorrect on further inspection. | 16:00 |
| mordred | however, even if they decided to-I expect that to be at least 6-8 months of work | 16:00 |
| BobBall | wow, that huge | 16:00 |
| * BobBall thinks that's a rabbit hole he should steer well clear from | 16:01 | |
| _TheDodd_ | jeblair, I see. Well, I am glad that it has been working! I've been a bit out of the loop these last few weeks, so I've been somewhat concerned about it. | 16:02 |
| *** hub_cap has left #openstack-infra | 16:02 | |
| _TheDodd_ | I am still planing on writing up some test code for a few of the major aspects of update_bug. At some point I would like to code up a mock object representing an lp bug test which we can automate our testing against. | 16:03 |
| fungi | _TheDodd_: the few backwards compatibility concerns which surfaced mainly just highlighted the broad variety of different and often overly-wordy ways people had gotten used to referencing fixed bugs in their commit messages | 16:03 |
| _TheDodd_ | Some of that stuff just seems far to risky to be deploying without some good test code. | 16:03 |
| fungi | i am inclined to consider no longer supporting most of the references i saw as a feature, not a drawback | 16:04 |
| _TheDodd_ | fungi, yea. I was stressing about that big time when I first wrote up the changes. | 16:04 |
| _TheDodd_ | I was concerned that I may have left some important words out, but I didn't know how far I should go with it. | 16:05 |
| fungi | _TheDodd_: it's interesting how quickly people will accept changes in workflow as long as there's a good reason for it ;) | 16:05 |
| _TheDodd_ | Ha. Indeed. | 16:05 |
| fungi | i didn't really see anyone get upset about it. just confused until it was explained and then, "ah! okay, i should start using the new header format anyway--thanks!" | 16:05 |
| _TheDodd_ | Nice | 16:06 |
| _TheDodd_ | Glad to hear that. | 16:06 |
| clarkb | morning | 16:06 |
| *** vogxn has joined #openstack-infra | 16:06 | |
| _TheDodd_ | It's always nice when peace prevails in the community. | 16:06 |
| fungi | it mostly caught people unawares because even though they'd seen the announcement they already had outstanding commits using non-forward-compatible phrasing | 16:06 |
| _TheDodd_ | Morning, clarkb! | 16:06 |
| fungi | i think they just assumed their existing commits would get "grandfathered in" somehow, but there wasn't a great way to skin that without a lengthy transition period and a fair amount of extra work | 16:08 |
| *** pabelanger has joined #openstack-infra | 16:08 | |
| jeblair | clarkb: so i refreshed the plugin manager on jenkins-dev, and it doesn't seem to indicate there is an update... | 16:09 |
| jeblair | clarkb: i'm uninstalling it and will restart jenkins | 16:09 |
| jeblair | clarkb: i'm wondering if it may have been installed manually before, so doesn't check the update center or something | 16:09 |
| clarkb | jeblair: I am not sure | 16:10 |
| clarkb | jeblair: I have had to manually tell the upadte manager to update its list | 16:10 |
| jeblair | i did do that already | 16:10 |
| clarkb | jeblair: also check that it shows up on the jenkins-ci.org side | 16:10 |
| jeblair | clarkb: where's that? | 16:10 |
| clarkb | jeblair: http://updates.jenkins-ci.org/download/plugins/zmq-event-publisher/ | 16:11 |
| clarkb | jeblair: looks like 0.0.2 is there | 16:11 |
| jeblair | clarkb: yeah, it shows up in the available list after i removed the manually installed version | 16:12 |
| jeblair | clarkb, zaro: so i think the jenkins publishing jobs are all good. :) | 16:13 |
| clarkb | awesome | 16:13 |
| *** jfriedly has joined #openstack-infra | 16:17 | |
| *** emagana has joined #openstack-infra | 16:17 | |
| *** gyee has joined #openstack-infra | 16:18 | |
| *** dina_belova has quit IRC | 16:19 | |
| *** vogxn has quit IRC | 16:19 | |
| *** pcrews has quit IRC | 16:21 | |
| *** dina_belova has joined #openstack-infra | 16:22 | |
| *** markmcclain has left #openstack-infra | 16:22 | |
| *** ^d has joined #openstack-infra | 16:23 | |
| *** ^d has joined #openstack-infra | 16:23 | |
| fungi | SpamapS: i have a few minutes to trigger your release now... what was the project name and tag number? | 16:23 |
| openstackgerrit | Ryan Petrello proposed a change to openstack-infra/config: Configure Pecan to re-publish docs through RTFD on release. https://review.openstack.org/41969 | 16:24 |
| *** _TheDodd_ has quit IRC | 16:24 | |
| *** SergeyLukjanov has quit IRC | 16:26 | |
| *** dina_belova has quit IRC | 16:26 | |
| *** jjmb has quit IRC | 16:27 | |
| *** _TheDodd_ has joined #openstack-infra | 16:27 | |
| SpamapS | fungi: os-apply-config , 0.1.0 | 16:28 |
| SpamapS | fungi: and thanks (hopefully you still have a minute :) | 16:28 |
| fungi | SpamapS: thanks. working on it | 16:28 |
| *** odyssey4me has quit IRC | 16:29 | |
| fungi | SpamapS: okay, so looks like i need to trigger os-apply-config-tarball and then once that succeeds trigger os-apply-config-pypi-upload | 16:31 |
| *** morganfainberg_a is now known as morganfainberg | 16:32 | |
| *** markmcclain has joined #openstack-infra | 16:32 | |
| fungi | i have the first running now | 16:34 |
| fungi | SpamapS: i think we may want to hold off on the pypi upload? it looks like your generated tarball is unversioned... http://tarballs.openstack.org/os-apply-config/os-apply-config-60b1928.tar.gz | 16:40 |
| clarkb | fungi: pbr should apply a version if it is tagged. that version happens when there is no tag | 16:40 |
| fungi | clarkb: hmmm... maybe i ran the job wrong... i used the --refname refs/tags/0.1.0 (it does exist) and the output of git show-ref 0.1.0 for the --newrev value | 16:41 |
| fungi | /opt/zuul/tools/trigger-job.py --job os-apply-config-tarball --pro | 16:42 |
| fungi | ject stackforge/os-apply-config --pipeline release --newrev 60b1928075033ef03966773c1d851407770ead0d --refname refs/tags/0.1.0 | 16:42 |
| fungi | gah, stray newline there | 16:42 |
| SpamapS | ? | 16:42 |
| fungi | SpamapS: pbr isn't finding the tag when doing setup.py sdist it seems | 16:43 |
| fungi | but i do see the tag in the repo | 16:43 |
| * fungi checks the jenkins log for the job | 16:43 | |
| SpamapS | did I tag the wrong commit? | 16:43 |
| *** dkliban is now known as dkliban_afk | 16:44 | |
| fungi | SpamapS: clarkb: https://jenkins01.openstack.org/job/os-apply-config-tarball/1/console | 16:45 |
| *** nati_ueno has joined #openstack-infra | 16:46 | |
| mtreinish | jeblair: so I've got: https://tinyurl.com/pdfz93l | 16:46 |
| fungi | i wonder if trigger-job.py needs to be modernized | 16:46 |
| mtreinish | is there a way to scale one of the plots to make it bigger | 16:46 |
| *** jjmb has joined #openstack-infra | 16:47 | |
| SpamapS | fungi: I am pbr stupid, so I don't really know what that shows except "not a tarball with a version" | 16:49 |
| *** fbo is now known as fbo_away | 16:49 | |
| *** ^demon has joined #openstack-infra | 16:49 | |
| SpamapS | fungi: I can confirm that sdist locally does not produce a versioned tarball | 16:49 |
| fungi | SpamapS: yeah, i'm right there with you... still trying to figure it out myself | 16:49 |
| *** ^d has quit IRC | 16:51 | |
| clarkb | mtreinish: look at the graphs on the zuul status page | 16:52 |
| *** jjmb has quit IRC | 16:52 | |
| fungi | SpamapS: i too get dist/os-apply-config-60b1928.tar.gz when running git checkout 0.1.0 && python setup.py sdist | 16:52 |
| SpamapS | hm | 16:52 |
| SpamapS | git describe doesn't show the tag | 16:52 |
| fungi | SpamapS: aha! | 16:52 |
| SpamapS | unsigned | 16:52 |
| SpamapS | _DOH_ | 16:52 |
| clarkb | mtreinish: you should be able to construct a graph that way and scale it | 16:52 |
| *** saper has quit IRC | 16:52 | |
| fungi | SpamapS: yup | 16:52 |
| *** saper has joined #openstack-infra | 16:52 | |
| clarkb | mtreinish: you can right click on one, copy url and open in a new tab | 16:52 |
| SpamapS | fungi: will delete 0.1.0 and re-issue with a signature | 16:52 |
| jeblair | mtreinish: it looks like if you click and drag a box it zooms in | 16:53 |
| *** dina_belova has joined #openstack-infra | 16:53 | |
| fungi | jeblair: mordred: clarkb: what was the proper approach we decided on for replacing unsigned release tags with signed ones? don't? | 16:53 |
| SpamapS | oh | 16:53 |
| * SpamapS waits | 16:53 | |
| *** wenlock has joined #openstack-infra | 16:53 | |
| fungi | i think without --fetch tags the build slave may not see the signed tag, if my memory is correct | 16:53 |
| fungi | because it's already gotten the unsigned one into its refs | 16:54 |
| SpamapS | ok so I can do 0.1.1 signed | 16:54 |
| fungi | SpamapS: that will work as long as your team is cool with it | 16:55 |
| SpamapS | I really don't mind if 0.1.0 was "the release that never was" | 16:55 |
| mtreinish | jeblair: ok cool thanks | 16:55 |
| mtreinish | clarkb: ok, I'll take a look | 16:55 |
| fungi | SpamapS: and should automagically worm its way through zuul-land without need of manual attention | 16:55 |
| mtreinish | clarkb: do you think graphing the failures that way is useful (the greater the negative amplitude on the red the more failures on parallel that weren't on serial) | 16:56 |
| *** boris-42 has quit IRC | 16:57 | |
| *** markmcclain has quit IRC | 16:58 | |
| *** SergeyLukjanov has joined #openstack-infra | 16:58 | |
| fungi | SpamapS: i highly recommend following the exact commands on https://wiki.openstack.org/wiki/GerritJenkinsGithub#Tagging_a_Release | 16:58 |
| fungi | SpamapS: modulo the branch/commit you're releasing from and the tag name itself of course | 16:59 |
| ryanpetrello | can anybody shed some light on the difference between 'release' and 'pre-release': http://ci.openstack.org/zuul.html#overview ? | 16:59 |
| fungi | ryanpetrello: it's based on a regexp match on the tag | 17:00 |
| *** dina_belova has quit IRC | 17:01 | |
| fungi | ryanpetrello: https://git.openstack.org/cgit/openstack-infra/config/tree/modules/openstack_project/files/zuul/layout.yaml#n48 | 17:01 |
| SpamapS | fungi: ok, 0.1.1 wending its way through the system | 17:01 |
| fungi | SpamapS: great! let's see how far it gets this time | 17:02 |
| * SpamapS feels like he just set off a rube goldberg machine | 17:02 | |
| *** jpich has quit IRC | 17:03 | |
| fungi | SpamapS: it does often feel just like that, i agree | 17:03 |
| mordred | SpamapS: w00t! | 17:03 |
| SpamapS | https://jenkins01.openstack.org/job/os-apply-config-tarball/2/console | 17:03 |
| SpamapS | yaaay | 17:03 |
| mordred | woot | 17:03 |
| mordred | of course, I just got a varnish error from pypi | 17:03 |
| mordred | dstufft: ^^ | 17:04 |
| SpamapS | https://jenkins01.openstack.org/view/All/job/os-apply-config-pypi-upload/ <-- shouldn't that go next? | 17:04 |
| clarkb | SpamapS: oui, but not on that master | 17:05 |
| SpamapS | oh duh | 17:05 |
| clarkb | SpamapS: iirc only jenkins.o.o is hooked up to slaves that can talk to pypi | 17:05 |
| *** rcleere has joined #openstack-infra | 17:05 | |
| SpamapS | https://jenkins.openstack.org/view/All/job/os-apply-config-pypi-upload/ | 17:05 |
| SpamapS | so that one? | 17:06 |
| *** BobBall is now known as BobBall_Away | 17:06 | |
| clarkb | yes, that is where I expect it to have run | 17:06 |
| mordred | SpamapS: you like finding all the problems don't you? | 17:06 |
| *** markmcclain has joined #openstack-infra | 17:07 | |
| SpamapS | mordred: its why you love to hate me and hate to love me. | 17:07 |
| * mordred hate loves SpamapS | 17:07 | |
| * SpamapS is the Charlie Sheen of OpenStack | 17:07 | |
| SpamapS | #winning | 17:07 |
| *** sarob has joined #openstack-infra | 17:07 | |
| clarkb | there is a two and a half men joke here somewhere | 17:08 |
| clarkb | I am not sure why that job did not run | 17:08 |
| clarkb | is the zuul config not correct? | 17:09 |
| * clarkb AFKs to dock laptop | 17:09 | |
| *** yolanda has quit IRC | 17:09 | |
| BobBall_Away | mordred: Next problem - /usr/bin/glance-api is invoked through /usr/bin/python rather than .venv/bin/python, so it breaks out of the venv and doesn't have the stuff it needs (e.g. the right version of python-crypto). Thoughts? | 17:09 |
| *** chuckieb has joined #openstack-infra | 17:10 | |
| mordred | BobBall_Away: awesome! why is it doing that? | 17:10 |
| mordred | ah | 17:10 |
| BobBall_Away | mordred: #!/usr/bin/python :) | 17:10 |
| mordred | $GLANCE_BIN_DIR/glance-api | 17:10 |
| mordred | we've got some explicit BIN_DIR settings we shoudl fix | 17:11 |
| mordred | lemme take a stab | 17:11 |
| *** nayward has quit IRC | 17:11 | |
| *** beagles has quit IRC | 17:11 | |
| mordred | get_python_exec_prefix | 17:11 |
| mordred | BobBall_Away: http://paste.openstack.org/show/44141 | 17:13 |
| mordred | but I'm going to amend my patch - did you make fixes to it already? | 17:14 |
| mordred | or - feel free to fix and reupload, btw | 17:14 |
| BobBall_Away | mordred: only locally to move the PIP_CACHE_DIR | 17:14 |
| BobBall_Away | (or whatever it was) | 17:14 |
| mordred | ok. I've made that change locally here too | 17:15 |
| BobBall_Away | mordred: but it's not working yet for me - and my day is almost over so I have to leave! If you haven't had a chance to fix it tonight I'll look at a new change tomorrow if I get it working | 17:16 |
| mordred | BobBall_Away: cool. well, the glance hint was a good one | 17:16 |
| *** beagles has joined #openstack-infra | 17:16 | |
| BobBall_Away | glad I'm useful for something ;) | 17:17 |
| BobBall_Away | *gone* | 17:17 |
| *** pcrews has joined #openstack-infra | 17:17 | |
| clarkb | SpamapS: fungi: so pypi upload did not run? | 17:18 |
| fungi | clarkb: i think i see why | 17:18 |
| *** _TheDodd_ has quit IRC | 17:18 | |
| clarkb | oh good I can go find caffeine then | 17:19 |
| *** sarob has quit IRC | 17:19 | |
| *** sarob has joined #openstack-infra | 17:19 | |
| fungi | well, no | 17:21 |
| fungi | clarkb: http://paste.openstack.org/show/44142/ "handle: None name: build:os-apply-config-pypi-upload unique: 9d3f4315d0304687926c164b5578e238> is not registered with Gearman" | 17:22 |
| *** markmcclain has quit IRC | 17:23 | |
| *** beagles has quit IRC | 17:23 | |
| fungi | but /var/lib/jenkins/jobs/os-apply-config-pypi-upload/config.xml exists on jenkins.o.o (which should be the one to farm it out, right?) | 17:23 |
| *** beagles has joined #openstack-infra | 17:23 | |
| *** sarob has quit IRC | 17:24 | |
| *** gyee has quit IRC | 17:25 | |
| dstufft | mordred: still happening? | 17:25 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/nodepool: Initial commit https://review.openstack.org/41805 | 17:25 |
| jeblair | clarkb, fungi, mordred: ^ that is ripe. | 17:26 |
| mordred | dstufft: it seems to be back | 17:26 |
| mordred | jeblair: I will look at that for real once I have coffee | 17:27 |
| dstufft | mordred: probably a temporary blip then | 17:27 |
| mordred | clarkb: did you see the pile of exciting new patches I uploaded this morning? | 17:27 |
| mordred | dstufft: yeah. agree | 17:27 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/nodepool: Initial commit https://review.openstack.org/41805 | 17:27 |
| mordred | the openstack project is excellent early-warning monitoring :) | 17:27 |
| mordred | dstufft: oh - btw - I think jeblair wrote the hard part of the code that needs to be written for me to be able to trigger openstack builds on changes to the pip repo and let you know if you're going to break us | 17:28 |
| *** cthulhup has joined #openstack-infra | 17:28 | |
| dstufft | mordred: I never bothered to get the guy with the keys to add my info to pingdom for pypi, becasue twitter lets me know faster than pingdom does | 17:28 |
| jeblair | clarkb, mordred, fungi: the jenkins queue is managable now; so i'm going to rolling upgrade zmq and gearman plugins | 17:28 |
| dstufft | my info == cell phone, it does get sent to my email | 17:29 |
| mordred | jeblair: ++ | 17:29 |
| mordred | dstufft: we have on more than one occiasion alerted gys at rax and hp about an issue they were about to start seeing ni ntheir clouds :) | 17:29 |
| jeblair | jenkins02 is in shutdown mode | 17:30 |
| mordred | a chorus of yelling developers is the new pingdom | 17:30 |
| dstufft | mordred: +1 | 17:30 |
| dstufft | build thing lots of people use, never need monitoring again | 17:30 |
| * fungi needs to run an errand... brb | 17:31 | |
| * jeblair imagines a world where twitter is pingdom's only user | 17:31 | |
| *** danger_fo_away is now known as danger_fo | 17:32 | |
| mgagne | cgit is not well conceived to be easily skinnable... =( | 17:34 |
| openstackgerrit | Jason Meridth proposed a change to openstack-dev/hacking: Adds ability to ignore H231, python3x except compatability check https://review.openstack.org/41713 | 17:34 |
| * jeblair -> afk for a bit | 17:35 | |
| *** rnirmal has quit IRC | 17:40 | |
| clarkb | mordred: I did not | 17:42 |
| clarkb | I am still catching up this morning | 17:42 |
| *** _TheDodd_ has joined #openstack-infra | 17:42 | |
| mordred | clarkb: I may have done things we've been wanting for a bit | 17:42 |
| clarkb | jeblair: fungi: is that a bug in the gearman plugin? potentially fixed by the upgrade jeblair is about to do | 17:42 |
| clarkb | on my walk to the office this morning I was thinking about a useful first real use case for logstash/elasticsearch | 17:44 |
| clarkb | I am thinking a daily job that searches the last days devstack screen logs for ERROR level messages, and collates them into a reasonable format for humans and further consumption would be a good place to start | 17:45 |
| clarkb | then we can start by handing a thing to openstack-qa and eventually feed that into a thing that deals with bugs | 17:45 |
| *** derekh has quit IRC | 17:45 | |
| mordred | ++ | 17:46 |
| *** ArxCruz_ has joined #openstack-infra | 17:47 | |
| clarkb | neat cgit shows you change refs in the git log tree | 17:47 |
| *** sarob has joined #openstack-infra | 17:47 | |
| *** nayward has joined #openstack-infra | 17:47 | |
| *** ArxCruz has quit IRC | 17:48 | |
| clarkb | I think we may need to give git.o.o more horsepower though | 17:49 |
| * fungi is back | 17:49 | |
| clarkb | mordred: can you look at https://review.openstack.org/#/c/41790/ ? this is the start of our mysql backup stuff | 17:50 |
| clarkb | mordred: and the child change of that shows you how I intend on using it | 17:50 |
| clarkb | mordred: I hope to test it on etherpad-dev today | 17:50 |
| mordred | clarkb: cool. you know you don't have to use $defaults_file = '/etc/mysql/debian.cnf' if ythe mysql was set up by puppet, yeah? | 17:51 |
| clarkb | mordred: right, but not all of our mysql is setup by puppet iirc | 17:51 |
| mordred | indeed. | 17:51 |
| clarkb | mordred: and I don't like the puppetlabs mysql backup class | 17:52 |
| mordred | just pointing out - as debian.cnf is debian specific | 17:52 |
| mordred | no - I don't think you should use their backup class | 17:52 |
| clarkb | mordred: right that is why it is a parameter | 17:52 |
| clarkb | does centos not create a defaults file that lets root in? | 17:52 |
| mordred | dunno | 17:52 |
| fungi | yeah, debian.cnf is convenient on debian systems, but i always feel sort of guilty for using it | 17:52 |
| mordred | great. also - there may be a time when our databases are in a cloud trove somewhere | 17:52 |
| fungi | like i really should set up separate backup credentials instead | 17:52 |
| clarkb | mordred: we can use a non debian "defaults" file when we make that switch | 17:53 |
| mordred | so we may want to just be able to pass the hostname, username, and password itself in as a parameter to the backup class and let it manage what it does with it | 17:53 |
| mordred | actually, ^^ is more my concern | 17:53 |
| clarkb | wouldn't that first require managing all of your mysql servers with puppet? | 17:53 |
| mordred | why? | 17:54 |
| clarkb | otherwise we won't know what the credentials are | 17:54 |
| mordred | if we create a database in the cloud | 17:54 |
| mordred | we will get credentials back from that action | 17:54 |
| mordred | then we put those creds into heira | 17:54 |
| clarkb | right, but we aren't there today | 17:54 |
| mordred | righ. I know | 17:54 |
| clarkb | today we need a thing that can backup databases on ubuntu that may or may not be configured by puppet | 17:54 |
| mordred | I'm just saying that is a thing we are going to want to be able to do in the future | 17:54 |
| clarkb | definitely | 17:55 |
| fungi | do we have a mysql database for pbx.o.o? | 17:55 |
| mordred | I am fine refactoring the backup module later to deal with that | 17:55 |
| clarkb | (I don't disagree with what you are saying, but think that the needs today are different than the needs tomorrow) | 17:55 |
| clarkb | mordred: and it might be easist to capture that with a second define. backup-trove-db vs backup-local-db or somesuch | 17:56 |
| clarkb | fungi: I don't think so | 17:56 |
| fungi | okay, then all debuntu things | 17:56 |
| *** Ryan_Lane has quit IRC | 17:57 | |
| clarkb | btw cloning nova from git://git.o.o is a full minute faster than cloning from github using https | 17:57 |
| fungi | that is thrilling, truly | 17:57 |
| clarkb | fungi: grep on pbx says no mysql | 17:57 |
| clarkb | er | 17:57 |
| clarkb | pgrep | 17:57 |
| *** Ryan_Lane has joined #openstack-infra | 17:58 | |
| * fungi is switching gears to build a temporary system to build a precise cacti security backport, because ubuntu | 17:58 | |
| mordred | fungi: lovely | 17:59 |
| mordred | fungi: I'm going to grab a coffee - when I get back do you want to talk about apt repos? | 17:59 |
| fungi | yeah, i want cacti back up before our calls friday, and i also want to be able to see if git.o.o could stand upgrading | 18:00 |
| fungi | mordred: i can talk, sure | 18:00 |
| fungi | get coffee | 18:00 |
| clarkb | mtreinish: tempest testr seems to potentially be unhappy noq | 18:00 |
| clarkb | *now | 18:00 |
| fungi | anybody know if there's a trick to running novaclient out of a virtualenv? i pip installed it into one but get "ERROR: 'NoneType' object has no attribute 'authenticate'" | 18:01 |
| *** dina_belova has joined #openstack-infra | 18:01 | |
| fungi | hen trying to use it (installing it works just fine) | 18:02 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Use cgit server instead of github for everything https://review.openstack.org/38177 | 18:02 |
| *** Ryan_Lane has quit IRC | 18:02 | |
| mordred | speaking of that ^^ | 18:02 |
| *** dina_belova has quit IRC | 18:02 | |
| *** dina_bel_ has joined #openstack-infra | 18:02 | |
| *** Ryan_Lane has joined #openstack-infra | 18:03 | |
| clarkb | fungi: no clue. is that the most recent release? | 18:03 |
| * clarkb tries | 18:03 | |
| fungi | clarkb: most recent release on pypi, though i am running it on a debian/jessie (testing) vm | 18:03 |
| *** jerryz has joined #openstack-infra | 18:04 | |
| fungi | probably roughly equivalent to running on raring | 18:04 |
| *** Ryan_Lane has quit IRC | 18:04 | |
| *** ladquin_afk is now known as ladquin | 18:04 | |
| clarkb | fungi: nova list works on precise out of a venv | 18:05 |
| fungi | clarkb: ha... bug 1197745 | 18:05 |
| uvirtbot | Launchpad bug 1197745 in python-novaclient "Rackspace authentication no longer supported" [Wishlist,Confirmed] https://launchpad.net/bugs/1197745 | 18:05 |
| clarkb | ah, I am was talking to hpcloud | 18:06 |
| fungi | reported on 2013-07-04 | 18:06 |
| fungi | after also installing rackspace-auth-openstack into the venv and setting the new OS_AUTH_SYSTEM=rackspace var, it works | 18:08 |
| fungi | i guess we'll eventually want to do similar things on our systems which launch servers or which are used by us to do so (d-l.slave, ci-puppetmaster) | 18:10 |
| anteaya | mordred: when you are back from coffee, this file disappeared from patchset 3: https://review.openstack.org/#/c/38177/2/modules/openstack_project/files/git/cgitrc | 18:12 |
| jeblair | fungi: i believe we switched to using passwords rather than rax api keys | 18:12 |
| anteaya | mordred: I think it should be in there but the only change is git:// for https:// | 18:12 |
| anteaya | though I could be wrong | 18:12 |
| SpamapS | fungi: so, what came of the pypi upload not running? | 18:13 |
| fungi | SpamapS: jenkins-gearman plugin bug is what clarkb suggested. we should probably wait until jeblair finishes the plugin rolling upgrades and manually retrigger it | 18:14 |
| SpamapS | I am so glad to ferret out all of your bugs :) | 18:14 |
| jeblair | fungi: who what? | 18:14 |
| *** mrmartin has joined #openstack-infra | 18:14 | |
| fungi | SpamapS: if only those really were all of our bugs | 18:14 |
| fungi | jeblair: http://paste.openstack.org/show/44142/ | 18:15 |
| fungi | jeblair: zuul claims that the job is not registered, but it clearly exists on jenkins.o.o | 18:16 |
| *** nayward has quit IRC | 18:16 | |
| clarkb | jeblair: I think possible related to executor management which the new verison of the plugin should handle much better | 18:16 |
| *** luisg has quit IRC | 18:16 | |
| *** luisg has joined #openstack-infra | 18:16 | |
| SpamapS | fungi: ah, so something has to be HUP'd or restarted maybe? | 18:16 |
| *** _TheDodd_ has quit IRC | 18:18 | |
| *** _TheDodd_ has joined #openstack-infra | 18:18 | |
| fungi | SpamapS: well, clarkb's assertion is that it may be due to a bug which may be fixed by jenkins plugin upgrades jeblair is in the middle of applying right now. if so, then trying after he's finished may work out | 18:19 |
| *** _TheDodd_ has quit IRC | 18:19 | |
| jeblair | i'll upgrade jenkins.o.o now | 18:19 |
| *** _TheDodd_ has joined #openstack-infra | 18:19 | |
| jeblair | i also restarted puppet because it had a zombie jenkins-jobs process | 18:22 |
| *** afazekas has quit IRC | 18:25 | |
| *** melwitt has joined #openstack-infra | 18:27 | |
| *** chuckieb has quit IRC | 18:27 | |
| *** lifeless has quit IRC | 18:28 | |
| *** hartsocks has joined #openstack-infra | 18:28 | |
| jeblair | okay, jenkins.o.o is back up. i had to start it twic; it seems to have hung the first time. | 18:29 |
| *** psedlak__ has quit IRC | 18:29 | |
| jeblair | fungi, SpamapS: the job is registered now | 18:29 |
| fungi | jeblair: awesome. retriggering it now | 18:29 |
| jeblair | restarting jenkins02 | 18:29 |
| fungi | well, awesome about the job now being registered. not so awesome about jenkins getting itself into a sorry state when starting | 18:30 |
| fungi | SpamapS: https://pypi.python.org/pypi/os-apply-config/0.1.1 | 18:33 |
| SpamapS | \o/ | 18:33 |
| SpamapS | now if I can just figure out why neutron is crashing... | 18:34 |
| nati_ueno | Hi Could you guys review this one? https://review.openstack.org/#/c/41799/ This is needed to identify gating issues | 18:34 |
| *** emagana has quit IRC | 18:35 | |
| mrmartin | hi | 18:37 |
| jeblair | i'm seeing the same deadlock on jenkins02; i'm getting stacktraces | 18:38 |
| mordred | clarkb: mind taking a loot at https://review.openstack.org/#/c/41283 ? or dhellmann ? | 18:38 |
| mordred | (pbr is broken for the windows folks) | 18:38 |
| *** markmcclain has joined #openstack-infra | 18:39 | |
| clarkb | woo shrews just helped me find a testr bug | 18:40 |
| clarkb | mordred: ya, let me file this bug against testrepository first | 18:40 |
| mordred | clarkb: kk | 18:40 |
| *** chuckieb has joined #openstack-infra | 18:40 | |
| *** thomasbiege has joined #openstack-infra | 18:41 | |
| *** dkliban_afk is now known as dkliban | 18:42 | |
| clarkb | hmm no lifeless on the freenodes | 18:46 |
| clarkb | https://bugs.launchpad.net/testrepository/+bug/1212390 is the bug | 18:46 |
| uvirtbot | Launchpad bug 1212390 in testrepository "Testr load cannot handle subunit v1." [Undecided,New] | 18:46 |
| jeblair | clarkb, zaro: so i think some of those methods i changed to synchronized need to go back to a finer grained explit synchronization on the gewtHandles object | 18:47 |
| clarkb | jeblair: you think that is causing the deadlock? | 18:47 |
| mordred | jeblair: poop. I had liked the sync methods approach | 18:48 |
| dhellmann | mordred: why not just make _run_shell_command() take a list of args? | 18:48 |
| *** thomasbiege has quit IRC | 18:48 | |
| clarkb | I have not started any of the owrk that I wanted to start yet. | 18:48 |
| mordred | dhellmann: cause it's ugly to read at the calling location? I dunno. it's just what we've always done :) | 18:48 |
| dhellmann | mordred: uglier than escaping quotes to make shlex work? | 18:49 |
| mordred | dhellmann: you do make a good point | 18:49 |
| *** rscottcoyle has joined #openstack-infra | 18:49 | |
| mordred | there are times when I miss Java multi-methods | 18:49 |
| dhellmann | mordred: ah, I thought this was your patch, I'll add a comment | 18:50 |
| *** emagana has joined #openstack-infra | 18:50 | |
| mordred | dhellmann: nope, it's alex from hyperv | 18:50 |
| * dhellmann nods | 18:51 | |
| dhellmann | do we have any examples of jenkins jobs that take a multiline string as a named parameter? | 18:52 |
| *** psedlak__ has joined #openstack-infra | 18:53 | |
| clarkb | dhellmann: I don't think so | 18:53 |
| dhellmann | clarkb: :-/ I want to pass file contents to a builder to write a little temp file. I'm passing it through sed to convert spaces to newlines right now, but that feels hackish. | 18:54 |
| NobodyCam | dhellmann: do you have a free minute to to take a look at the Fix Driver loading Ironic patch? (https://review.openstack.org/#/c/39617/) | 18:56 |
| dhellmann | NobodyCam: looking | 18:56 |
| NobodyCam | :) | 18:56 |
| NobodyCam | TY | 18:56 |
| *** ^demon has quit IRC | 18:59 | |
| *** psedlak__ has quit IRC | 18:59 | |
| mrmartin | jeblair, I added a missing .gitreview to the groups project. Do I need to be a member of core reviewers for this project? https://review.openstack.org/#/c/41988/ | 19:01 |
| *** cthulhup has quit IRC | 19:03 | |
| *** ArxCruz_ has quit IRC | 19:03 | |
| fungi | jeblair: i can add mrmartin to groups-core if desired | 19:04 |
| fungi | mrmartin: taking a look at your patch now | 19:04 |
| jeblair | fungi: please, thanks | 19:04 |
| mrmartin | fungi: thanks | 19:04 |
| fungi | mrmartin: you're in the core group for it now | 19:04 |
| mrmartin | it is perfect patch | 19:04 |
| *** ArxCruz has joined #openstack-infra | 19:05 | |
| jeblair | i'm going to downgrade to 0.0.3.1 on jenkins02 and restart | 19:05 |
| mordred | jeblair: k | 19:05 |
| mrmartin | ok I have a few questions, I made the puppet manifests to create the drupal environment for groups-staging.openstack.org, and it runs well in my local dev environment. | 19:05 |
| fungi | jeblair: do we want openstack-ci-core included in groups-core too? | 19:06 |
| mrmartin | I needed to add a pear puppet module due drush cli support: rafaelfc-pear (v1.0.3) | 19:06 |
| mrmartin | what is the proper way to add it to openstack-ci/config repo? | 19:06 |
| jeblair | fungi: yeah, it's under openstack-infra | 19:07 |
| mordred | mrmartin: you'll see a list of modules ... | 19:07 |
| mrmartin | Does it enough if I add it to install_modules.sh? How will this populated after a commit? | 19:07 |
| mordred | yes | 19:07 |
| mordred | our puppet cron stuff runs install_modules | 19:08 |
| mrmartin | oh great | 19:08 |
| *** gyee has joined #openstack-infra | 19:08 | |
| mrmartin | the other question is, that we'll have a running staging site at groups-staging.openstack.org. Currently I have a bash build script that populates the drupal site under virtualhost directory, it runs nicely when I launching a new node with drupal manifest. | 19:09 |
| NobodyCam | TY dhellmann :) | 19:10 |
| mrmartin | If I like to update this environment after every commit, what is the proper way to do that? Can I trigger some notification from gating, and execute this script on this node somehow? | 19:11 |
| mordred | dhellmann: alexpilotti was wondering if we could do the run_shell_command-takes-list refactor next when he's past not being able to install anything | 19:11 |
| dhellmann | mordred: sure | 19:11 |
| *** alexpilotti has joined #openstack-infra | 19:11 | |
| *** emagana has quit IRC | 19:11 | |
| *** mjfork has quit IRC | 19:11 | |
| mrmartin | hi Alessandro | 19:11 |
| fungi | mrmartin: we'd have the vcsrepo module watch that repository and then make an exec for that script in puppet and subscribe the exec to the vcsrepo | 19:12 |
| mordred | alexpilotti: dhellmann said 'sure' | 19:12 |
| dhellmann | alexpilotti, mordred : +2 | 19:12 |
| mrmartin | fungi: thanks, I'll check it | 19:12 |
| *** emagana has joined #openstack-infra | 19:13 | |
| fungi | mrmartin: that should cause the script to run each time vcsrepo pulls the latest commit from the remote branch | 19:13 |
| mordred | alexpilotti: but you have to promise to do the refacto :) | 19:13 |
| alexpilotti | thanks dhellmann ! | 19:13 |
| * alexpilotti hides | 19:13 | |
| clarkb | mrmartin: fungi: would the vhost need to be updated each time? or do we simply need to update the code behind it? | 19:13 |
| mrmartin | fungi: ok, this is the last step missing from auto-update process of staging | 19:13 |
| * dhellmann wonders if yaml even honors multiline strings in jjb params | 19:13 | |
| alexpilotti | kidding, I'm on it | 19:13 |
| mrmartin | clarkb: I mean we need just update the code behind it, not the entire vhost (vhost = apache vhost here) | 19:14 |
| clarkb | mrmartin: fungi: it almost sounds to me like the current situation depends on the script to update the code which is why it is run whereas the new sitaution can have puppet do that | 19:14 |
| alexpilotti | We have an issue on hooks.py on Neutron on Windows: https://bugs.launchpad.net/neutron/+bug/1212385 | 19:14 |
| uvirtbot | Launchpad bug 1212385 in neutron "Setup fails on Windows due to changes in neutron/hooks.py" [Undecided,New] | 19:14 |
| fungi | clarkb: i took "under virtualhost directory" to mean that the repository updates get pulled into the document root corresponding to that apache vhost | 19:15 |
| alexpilotti | basically, on Windows we have to remove some requirements and add some others | 19:15 |
| alexpilotti | https://github.com/openstack/neutron/blob/ee3fe4e836ca1c81e50a8324a9b5f982de4fa97f/neutron/hooks.py#L25 | 19:15 |
| mordred | alexpilotti: ah - you need to _remove_ some of them | 19:15 |
| fungi | clarkb: or s/pulled into/generated into based on the git repo content/ | 19:15 |
| mrmartin | clarkb: I have a working bash script for that, I'll check whether I could puppetize it. This could be a more elegant solution. | 19:16 |
| alexpilotti | now as requires_dist is empty, I cannot remove them (actually it's only pyudev) | 19:16 |
| mordred | yeah. I have a n idea | 19:16 |
| *** mjfork has joined #openstack-infra | 19:16 | |
| alexpilotti | is there a different way to handle it now? | 19:16 |
| mordred | alexpilotti: for now, put in an else after the win32, add pyudev there, and remove it from the reuqirements.txt list | 19:17 |
| mordred | alexpilotti: we have a todo list item to let people put markerlib strings in requirements.txt - but I have to sell dstufft on pip support for that first | 19:17 |
| alexpilotti | oki, so requires_dist is still used in the latest pbr builds, correct? | 19:17 |
| mordred | yup | 19:18 |
| mordred | it's what pbr injects things that are in requirements.txt into | 19:18 |
| alexpilotti | oki, sounds good! | 19:18 |
| clarkb | mordred: I left a comment on your experimental pipeline change | 19:19 |
| mordred | clarkb: yay! | 19:19 |
| clarkb | mordred: would be curious to see what you have to say about what I have to say | 19:19 |
| mordred | clarkb: so - the outstanding question with that is what happens with the voting | 19:20 |
| mordred | clarkb: the difference is that the experimental pipeline does not run by default | 19:20 |
| mordred | it only runs when you request it | 19:20 |
| *** hartsocks has left #openstack-infra | 19:20 | |
| mordred | so we don't have to run cells 1000 times a day so that devs can ignore the red color and still know that cells is broken | 19:20 |
| mordred | but the guys working on cells can easily request a run of the cells job | 19:20 |
| *** rnirmal has joined #openstack-infra | 19:23 | |
| *** ruhe has joined #openstack-infra | 19:24 | |
| *** vipul is now known as vipul-away | 19:24 | |
| *** thomasbiege has joined #openstack-infra | 19:25 | |
| *** nicedice_ has joined #openstack-infra | 19:25 | |
| *** vipul-away is now known as vipul | 19:28 | |
| *** emagana has quit IRC | 19:28 | |
| *** vipul is now known as vipul-away | 19:32 | |
| *** thomasbiege has quit IRC | 19:32 | |
| *** lifeless has joined #openstack-infra | 19:39 | |
| *** jbjohnso has joined #openstack-infra | 19:39 | |
| *** ruhe has quit IRC | 19:39 | |
| jbjohnso | question in this whole gerrit workflow, the way I usually push tags in git doesn't work, how to push tags? | 19:39 |
| mgagne | jbjohnso: which project? | 19:41 |
| fungi | jbjohnso: https://wiki.openstack.org/wiki/GerritJenkinsGithub#Tagging_a_Release | 19:41 |
| *** afazekas has joined #openstack-infra | 19:41 | |
| mgagne | jbjohnso: what fungi said, otherwise it could be an ACL issue in gerrit for this specific project | 19:43 |
| mordred | jeblair: you're not on the twitter, but I figured you'd like this: http://www.slate.com/blogs/business_insider/2013/08/14/gmail_security_google_says_users_have_no_legitimate_expectation_of_privacy.html | 19:43 |
| mordred | jbjohnso: yeah. we just have to set you as being a person who can push tags for a project | 19:44 |
| *** vipul-away is now known as vipul | 19:45 | |
| *** pabelanger has quit IRC | 19:45 | |
| *** wenlock has quit IRC | 19:49 | |
| *** chuckieb has quit IRC | 19:50 | |
| *** hartsocks has joined #openstack-infra | 19:54 | |
| jbjohnso | thanks, that worked | 19:54 |
| mtreinish | clarkb: sorry I was in a really long meeting were the tests failing? | 19:55 |
| hartsocks | So I have a question about a patch I'm trying to get Jenkins to recheck. Is there something going on today? | 19:56 |
| mordred | hartsocks: yeah. there were some issues earlier with race conditions inside of jenkins | 19:57 |
| mordred | we're trying to track them down | 19:57 |
| hartsocks | mordred: thanks | 19:58 |
| *** SergeyLukjanov has quit IRC | 19:58 | |
| * fungi needs to disappear again for a bit... bbiaw | 19:58 | |
| clarkb | mtreinish: they were failing, but seem to be happy again | 19:59 |
| clarkb | mtreinish: I may have overreacted | 19:59 |
| clarkb | mordred: thats the difference. I completely missed that for some reason | 20:00 |
| clarkb | mordred: now that I know that I love the idea | 20:00 |
| clarkb | because cells failing always and neutron-full failing always is getting old | 20:01 |
| *** rscottcoyle has quit IRC | 20:02 | |
| *** afazekas has quit IRC | 20:02 | |
| *** rcleere has quit IRC | 20:03 | |
| mgagne | first attempt at skinning cgit for openstack: http://imgur.com/Xw7CIy9 | 20:04 |
| pleia2 | mgagne: that looks great | 20:05 |
| pleia2 | I still have my test cgit instance up, so if you want to toss in review I can test | 20:06 |
| mtreinish | clarkb: ok, yeah I'm looking at the trend graph it looks to be about the same there were a couple of failures that happened around the same time | 20:06 |
| mtreinish | but overall it looks pretty good | 20:06 |
| pleia2 | oh, I need to set up a static place for files | 20:07 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/gearman-plugin: Use more fine-grained synchronization in GearmanProxy https://review.openstack.org/42003 | 20:07 |
| mtreinish | but definitely not as good as serial | 20:07 |
| jeblair | clarkb, mordred, zaro: ^ | 20:07 |
| jeblair | mgagne: NICE! | 20:07 |
| mgagne | pleia2: I'm not ready for a review, I'm copy/pasting my css in the developer tool of chrome, my setup is ghettho | 20:07 |
| *** pcrews has quit IRC | 20:08 | |
| *** emagana has joined #openstack-infra | 20:08 | |
| jeblair | pleia2, mgagne: can we disable the "owner" column in the display? | 20:08 |
| mgagne | pleia2: I can send you the css if you want to work on it and improve it. I used duct tape. Otherwise I'll check later this week | 20:08 |
| *** pcrews has joined #openstack-infra | 20:08 | |
| mgagne | jeblair: sure! | 20:08 |
| mgagne | jeblair: in my "todo" list | 20:08 |
| mordred | jeblair: exciting, looking | 20:08 |
| pleia2 | mgagne: sure | 20:08 |
| mgagne | jeblair: enable-index-owner: 0 | 20:09 |
| mgagne | pleia2: don't laugh: http://paste.openstack.org/show/44161/ | 20:09 |
| mgagne | pleia2: I'm mostly taking css from the gerrit theme | 20:10 |
| mgagne | pleia2: I did not test other pages as my setup is limited in that aspect. I would have to pop a cgit install to test and I'm lazy | 20:10 |
| pleia2 | mgagne: thanks! which openstack icon are you using, the one from gerrit? | 20:11 |
| mgagne | pleia2: yes, check the url =) | 20:12 |
| jeblair | i've put jenkins02 back in shutdown mode, hopefully we can upgrade the gearman plugin again there. | 20:12 |
| mgagne | pleia2: there is already a copy of the logo in the openstack_project module | 20:12 |
| pleia2 | mgagne: saw bg, not os icon | 20:12 |
| pleia2 | cool, I'll make sure I install that one | 20:12 |
| mgagne | pleia2: it's only a matter of installing static files to the right folder | 20:12 |
| * pleia2 nods | 20:12 | |
| pleia2 | halfway through my basic static files patch now (logo and favicon are config options, so those are easy to do quickly) | 20:13 |
| pleia2 | ...but my lunch just arrived! be back soon | 20:13 |
| mgagne | pleia2: modules/openstack_project/files/openstack.png? | 20:13 |
| clarkb | mgagne: if I have a puppet module with a define it it as module::foo do I need an init.pp for that module? | 20:14 |
| mgagne | clarkb: no | 20:14 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/nodepool: Initial commit https://review.openstack.org/41805 | 20:14 |
| clarkb | mgagne: thanks | 20:14 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Switch etherpad_lite backups to mysql_backup. https://review.openstack.org/41791 | 20:16 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Add new mysql_backup module. https://review.openstack.org/41790 | 20:16 |
| *** whayutin_ has joined #openstack-infra | 20:16 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Add job to propose tag merges https://review.openstack.org/41927 | 20:17 |
| *** _TheDodd_ has quit IRC | 20:17 | |
| *** _TheDodd_ has joined #openstack-infra | 20:18 | |
| mordred | clarkb: lifeless is back | 20:20 |
| *** weshay has quit IRC | 20:20 | |
| *** ^d has joined #openstack-infra | 20:22 | |
| *** whayutin_ is now known as weshay | 20:23 | |
| *** adalbas has quit IRC | 20:23 | |
| *** ftcjeff has joined #openstack-infra | 20:26 | |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Switch etherpad_lite backups to mysql_backup. https://review.openstack.org/41791 | 20:27 |
| lifeless | mordred: clarkb: yes, apparently my server didn't want to be on overnight. So whatever you asked... I don't know. | 20:28 |
| mordred | lifeless: clarkb made you a new bug present | 20:28 |
| *** sandywalsh has quit IRC | 20:28 | |
| mordred | lifeless: https://bugs.launchpad.net/testrepository/+bug/1212390 | 20:28 |
| uvirtbot | Launchpad bug 1212390 in testrepository "Testr load cannot handle subunit v1." [Undecided,New] | 20:28 |
| *** ftcjeff_ has joined #openstack-infra | 20:30 | |
| clarkb | mordred: jeblair: fungi 41791 and its parent have been applied to etherpad-dev and look to be good | 20:31 |
| *** jtzl has joined #openstack-infra | 20:31 | |
| jeblair | clarkb, zaro: are either of you interested in reviewing https://review.openstack.org/#/c/42003/ or should i let production jenkins review it? :) | 20:31 |
| clarkb | jeblair: I am reviewing it now | 20:31 |
| jeblair | ok | 20:31 |
| lifeless | mordred: clarkb: ah - wontfix. | 20:32 |
| openstackgerrit | Jason Meridth proposed a change to openstack-dev/hacking: Adds ability to ignore H231, python3x except compatability check https://review.openstack.org/41713 | 20:32 |
| clarkb | lifeless: why not? | 20:32 |
| zaro | jeblair: me too, was reviewing clarks puppet backup. | 20:32 |
| *** ftcjeff has quit IRC | 20:32 | |
| clarkb | lifeless: you are outputting content in a format you cannot consume | 20:32 |
| lifeless | clarkb: no, we're not outputting that content at all | 20:33 |
| clarkb | lifeless: also I now know why testr run --analyze-isolation hasn't been working for anyone | 20:33 |
| clarkb | because testr load is loading an empty test log | 20:33 |
| clarkb | jeblair: is that change basically a revert on that one file? | 20:33 |
| jeblair | clarkb: i very much doubt it, but have not diffed the two versions | 20:33 |
| jeblair | clarkb: there should still be substantial changes to method bodies in both this and the previous change | 20:34 |
| lifeless | clarkb: huh, where from ? | 20:35 |
| clarkb | lifeless: testr load of a v1 subunit file doesn't load anything | 20:35 |
| clarkb | lifeless: so when I tell people to grab the subunit log fron jenkins, load it and run analyze-isolation it is a noop | 20:35 |
| clarkb | either testr can handle both or I have to also tell people to do a conversion first | 20:35 |
| clarkb | and the fact that it isn't a failure to load the old stuff makes it silently break | 20:36 |
| lifeless | clarkb: oh, I thought you were saying analyze-isolation was internally broken. | 20:36 |
| *** thomasbiege has joined #openstack-infra | 20:36 | |
| *** briancli1e is now known as briancline | 20:36 | |
| lifeless | clarkb: so you may remember extensive discussions about auto probing back oh 8 months | 20:36 |
| *** thomasbiege has quit IRC | 20:36 | |
| clarkb | well it might also be since it has to know bout the last run | 20:36 |
| lifeless | clarkb: it's hard to get /right/, the failure modes are subtle | 20:36 |
| clarkb | but testr can't read the last run apparently | 20:36 |
| clarkb | I actually did not check if testr last works properly after testr run | 20:37 |
| lifeless | badfile | subunit-1to2 | testr load ; should be enough | 20:37 |
| lifeless | clarkb: v1 needs to go far far away as soon as possible | 20:38 |
| jeblair | clarkb, fungi, mordred: quick question: are you okay with a nodepool.openstack.org host? if so, i will spin one up and use it to develop puppet manifests (for later review) | 20:38 |
| lifeless | clarkb: The last thing I want to do is add more support for it anywhere | 20:38 |
| *** dkliban has quit IRC | 20:38 | |
| *** dina_bel_ has quit IRC | 20:39 | |
| lifeless | clarkb: if you want to avoid education, we could do 1to2 when capturing the file, so users don't need to know that the current store is v1 | 20:39 |
| lifeless | clarkb: by capturing, I mean when you copy it to storage | 20:39 |
| *** adalbas has joined #openstack-infra | 20:40 | |
| clarkb | testr last does work correctly, so I think analyize-isolation works in the general case | 20:40 |
| clarkb | lifeless: education is fine. I think my biggest concern right now is that there is no indication that something went wrong | 20:41 |
| *** sandywalsh has joined #openstack-infra | 20:41 | |
| clarkb | instead it says PASS | 20:41 |
| *** dprince has quit IRC | 20:41 | |
| mordred | jeblair: yes | 20:42 |
| *** jjmb has joined #openstack-infra | 20:42 | |
| clarkb | jeblair: I left a question in that review | 20:42 |
| clarkb | jeblair: I am okay with nodepool.o.o | 20:43 |
| mordred | lifeless, clarkb why aren't we capturing in v2 anyway? | 20:44 |
| mtreinish | mordred: I think it's for the html results output | 20:45 |
| clarkb | mtreinish: that can actually handle either version | 20:46 |
| mtreinish | oh, ok | 20:46 |
| clarkb | I think it may have to do with v2 being ugly to look at as a human maybe? | 20:46 |
| clarkb | I do like the v1 is the default log format. It is easier on us humans | 20:46 |
| clarkb | but v2 isn't so bad that I can't see switching to it everywhere | 20:46 |
| mgagne | what OS/version is running on git.o.o ? | 20:49 |
| lifeless | I just haven't inished the internal migration sufficiently to move the backend to 2 | 20:49 |
| lifeless | ouch, HTF did I do that. | 20:49 |
| clarkb | mgagne: CentOS 6.4 | 20:49 |
| mgagne | clarkb: thanks | 20:49 |
| mordred | pleia2: do you know if there is a way to make 'tree' the default think I git when I go to https://git.openstack.org/cgit/openstack/keystone ? | 20:52 |
| mrmartin | guys: I have this review: https://review.openstack.org/#/c/42007/ but no email notification was arrived to my inbox. Do I need to setup something else in the project to receive notifications about new patches? | 20:53 |
| jeblair | mrmartin: check out https://review.openstack.org/#/settings/projects | 20:53 |
| clarkb | and by default you don't get mail that would be generated by yourself | 20:53 |
| clarkb | there is a check box somewhere to change that under settings | 20:54 |
| mrmartin | clarkb: ok, thanks | 20:54 |
| clarkb | jeblair: did you see my comments on 42003? | 20:54 |
| *** jtzl has quit IRC | 20:54 | |
| jeblair | clarkb: yes, am about to respond | 20:54 |
| mrmartin | and when I commit a new patch, do I need to always add myself as a reviewer? | 20:54 |
| clarkb | mrmartin: you do not need to add yourself as a reviewer if you watch the project | 20:55 |
| mgagne | mordred: what would be the link to go back to summary if that link is tree by default? | 20:55 |
| mordred | mgagne: hrm. I'm not sure | 20:55 |
| clarkb | mrmartin: and I think you are automatically added to the list of people hwne it is your own change | 20:55 |
| *** jtzl has joined #openstack-infra | 20:55 | |
| mrmartin | clarkb: yeah, I thought the same, but not happened | 20:55 |
| mordred | mgagne: I should probably just get used to it I guess | 20:56 |
| mrmartin | clarkb: I try to tune the project watch settings | 20:56 |
| mgagne | mordred: it's not that I don't agree with you =) | 20:56 |
| mordred | mgagne: yeah - I think it's funny - pretty much everyone decides to show you something other than the files as the primary view | 20:56 |
| mordred | launchpad, bitbucket, gitorious, cgit | 20:57 |
| mordred | github is like the only one that starts from a tree view | 20:57 |
| mgagne | mordred: tell me about it, add gitlab to the list | 20:57 |
| *** ArxCruz has quit IRC | 20:57 | |
| mordred | except, almost without exception, the _only_ reason I go to a repo in a web browser is to, you guessed it, browse the files | 20:57 |
| mgagne | mordred: they feel "activities" are most important than files themselves | 20:57 |
| jeblair | clarkb: responded; thanks. | 20:58 |
| *** rfolco has quit IRC | 20:58 | |
| *** briancline has quit IRC | 20:58 | |
| mordred | mgagne: yup. except there is no workflow in cgit - it's just a browser - and "what tags are there" is a less frequent question than "what's in the requirements.txt of that project" | 20:58 |
| *** jjmb has quit IRC | 20:59 | |
| mordred | they've already got a branch dropdown - if they'd add a tag dropdown, I believe the summary page could actually just go away | 20:59 |
| *** dkliban has joined #openstack-infra | 20:59 | |
| mgagne | mordred: top feedback for gitlab: http://feedback.gitlab.com/forums/176466-general/suggestions/3907476-add-readme-to-project-home-view-or-make-files-t "we think the activity feed is the most important view for frequent visitors to the project" | 20:59 |
| mordred | mgagne: funny. none of the visitors to the project seem to agree | 21:00 |
| mgagne | mordred: IMO, he is wrong and should open his eyes to the reality of the users | 21:00 |
| mgagne | mordred: tell me about it, I got totally turned off by this answer and moved to an other git repo manager | 21:01 |
| mordred | mgagne: | 21:01 |
| mordred | :) | 21:01 |
| mordred | that's ok - we asked github a question back in the early days, they told us they didnt' care, we went elsewhere | 21:02 |
| mgagne | mordred: =) | 21:02 |
| *** woodspa has quit IRC | 21:03 | |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/gearman-plugin: Use more fine-grained synchronization in GearmanProxy https://review.openstack.org/42003 | 21:04 |
| jeblair | clarkb, mordred: ^ | 21:04 |
| clarkb | looking | 21:05 |
| jeblair | i guess the thread member should have the same kind of synch around it, huh? | 21:06 |
| jeblair | clarkb: yeah, let me refine that a little more | 21:08 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/gearman-plugin: Use more fine-grained synchronization in GearmanProxy https://review.openstack.org/42003 | 21:11 |
| jeblair | clarkb: ok, so if we combine what i just did with your idea, then i think we have protection for .thread and .worker ^ | 21:11 |
| *** changbl has quit IRC | 21:12 | |
| mtreinish | clarkb, jeblair: is there a good way to pull stats from zuul. Like if I wanted to know how many runs of a certain job succeed on the gate pipeline in an interval. | 21:13 |
| jeblair | mtreinish: i think those are the graphite keys you were looking at earlier? | 21:14 |
| mtreinish | jeblair: but is there a way to pull the raw number from graphite? | 21:14 |
| jeblair | mtreinish: oh, yeah, i think you can form a json graphite url... | 21:15 |
| jeblair | mtreinish: can you paste the graph url you were looking at earlier? | 21:15 |
| mtreinish | jeblair: this one: https://tinyurl.com/pdfz93l | 21:15 |
| mtreinish | ? | 21:15 |
| jeblair | mtreinish: http://graphite.openstack.org/render/?from=-24hour&until=-0hour&target=stats.zuul.pipeline.gate.job.gate-tempest-devstack-vm-testr-full.SUCCESS&format=json | 21:16 |
| jeblair | mtreinish: http://graphite.openstack.org/render/?from=-24hour&until=-0hour&target=summarize%28stats.zuul.pipeline.gate.job.gate-tempest-devstack-vm-testr-full.SUCCESS,%20%2724h%27%29&format=json | 21:17 |
| mtreinish | jeblair: ok cool thanks | 21:18 |
| mgagne | with cgit, do you think we should "care" (show) about how many files/lines are modified by a commit? | 21:18 |
| jeblair | mgagne: tentatively, yes? unless there's a good reason not to? | 21:19 |
| jeblair | mgagne: are you thinking about the Lines column in the log view? | 21:19 |
| mgagne | jeblair: yes | 21:19 |
| mgagne | jeblair: mostly motivating by the desire to unclutter the UI | 21:19 |
| *** pcm__ has quit IRC | 21:20 | |
| *** thomasbiege has joined #openstack-infra | 21:20 | |
| jeblair | mgagne: ok, yeah, i could be convinced that's not useful. it seems rare that you'd chose to look at a commit because of the number of lines changed. | 21:21 |
| jeblair | (and if that's what you are doing, perhaps a more specialized tool is appropriate) | 21:21 |
| mtreinish | jeblair: do you know how it is counting? because it goes from ~.0333 to ~.91666 which shows the increase in successful jobs but not how many more (or least not in a raw count)? | 21:21 |
| *** thomasbiege has quit IRC | 21:23 | |
| jeblair | mtreinish: http://graphite.openstack.org/render/?from=-24hour&until=-0hour&target=hitcount%28stats.zuul.pipeline.gate.job.gate-tempest-devstack-vm-testr-full.SUCCESS,%20%2724h%27%29&format=json | 21:24 |
| jeblair | mtreinish: does that do what you want? you probably want to double check that. | 21:24 |
| mtreinish | jeblair: yeah that's it. I really need to read the graphite docs now | 21:25 |
| mtreinish | jeblair: it seems like a reasonable number | 21:25 |
| *** alexpilotti_ has joined #openstack-infra | 21:26 | |
| *** fbo_away is now known as fbo | 21:27 | |
| *** alexpilotti has quit IRC | 21:29 | |
| *** alexpilotti_ is now known as alexpilotti | 21:29 | |
| mordred | jeblair: in stopAll() - what's the point of doing stopHandles = new ArrayList<AbstractWorkerThread>(); right before the sync block where you do stopHandles = new ArrayList<AbstractWorkerThread>(gmwtHandles); ? | 21:30 |
| *** thomasbiege has joined #openstack-infra | 21:31 | |
| clarkb | jeblair: and with thread.start() within the synchronized block should the stops move into the synchronized blocks as well? | 21:31 |
| jeblair | mtreinish: i changed it to include all pipelines for the past 1350 minutes, and got 279 runs; the current zuul log says 274 runs; that's fairly close (there could still be a little time skew i'm not accounting for) | 21:31 |
| jeblair | mtreinish: (the zuul log doesn't make grepping by pipeline easy) | 21:31 |
| jeblair | mordred: that does seem pointless. | 21:32 |
| mordred | jeblair: also probably harmless | 21:33 |
| *** alexpilotti has quit IRC | 21:34 | |
| jeblair | clarkb: so i'm counting on ewt.start not actually doing any work (it just spawns a Thread); but having it in the sync block means you can't stop it until run has been set and a thread object exists | 21:34 |
| jeblair | clarkb: (you can't stop it because you can't retrieve it from gewthandles) | 21:35 |
| jeblair | clarkb: but stop could take a while and actually execute real code, so i'd rather not have it in the sync block | 21:35 |
| *** alexpilotti has joined #openstack-infra | 21:35 | |
| jeblair | clarkb: (but i think we should also probably move the management worker start into its sync block) | 21:36 |
| * fungi returns | 21:37 | |
| fungi | ahh, scrollback | 21:37 |
| clarkb | jeblair: gotcha that makes sense to me | 21:37 |
| clarkb | fungi: hahahaha | 21:37 |
| mordred | woot | 21:37 |
| *** sarob has quit IRC | 21:37 | |
| *** sarob has joined #openstack-infra | 21:38 | |
| clarkb | jeblair: however tests do not pass with patchset 3 | 21:38 |
| clarkb | I wonder if the tests have bad assumptions about the order of things now that the locking is slightly changed | 21:38 |
| * fungi so far... likes mgagne's openstack cgit skin! | 21:38 | |
| mordred | fungi, clarkb: if you're bored, you could look at https://review.openstack.org/#/c/41215/ | 21:39 |
| *** rcleere has joined #openstack-infra | 21:39 | |
| clarkb | I am super bord /sarcasm :) I did get the current iteration of the mysql backup stuff working on etherpad-dev thouhg | 21:39 |
| *** dina_belova has joined #openstack-infra | 21:40 | |
| clarkb | if folks want to review those changes I can apply to them to etherpad.o.o as well then we can start looking at backing up all of the other mysql servers | 21:40 |
| mordred | clarkb: oh yeah - I'll go review them now | 21:40 |
| openstackgerrit | Mathieu Gagné proposed a change to openstack-infra/config: Cleanup cgit interface https://review.openstack.org/42016 | 21:40 |
| fungi | clarkb: third half. nice | 21:41 |
| *** sarob has quit IRC | 21:42 | |
| *** mriedem has quit IRC | 21:42 | |
| mordred | clarkb: +2 | 21:43 |
| *** dina_belova has quit IRC | 21:44 | |
| clarkb | fungi: I didn't realize the pie had that many slices, but it was too late to give up on halves :) | 21:44 |
| *** jhesketh has joined #openstack-infra | 21:44 | |
| *** thomasbiege has quit IRC | 21:46 | |
| *** pabelanger has joined #openstack-infra | 21:47 | |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/gearman-plugin: Use more fine-grained synchronization in GearmanProxy https://review.openstack.org/42003 | 21:50 |
| *** dina_belova has joined #openstack-infra | 21:50 | |
| jeblair | clarkb, mordred: ^ with the change i mentioned and passing tests | 21:50 |
| mordred | jeblair: I think passing tests are an overrated luxury | 21:50 |
| mordred | holy crap - that's many more files | 21:51 |
| mgagne | would people be interested in source highlight in cgit? | 21:52 |
| mordred | mgagne: I think so - I think they're used to it everywhere else | 21:52 |
| *** rcleere has quit IRC | 21:52 | |
| lifeless | where to pip bugs go? | 21:52 |
| lifeless | bah | 21:52 |
| *** sarob has joined #openstack-infra | 21:52 | |
| lifeless | where do pip bugs go ? I ended up looking into pip cache safety, and it's not concurrency safe. | 21:52 |
| lifeless | would like to capture that for upstream. | 21:52 |
| lifeless | dstufft: ^ | 21:52 |
| dstufft | github.com/pypa/pip | 21:53 |
| lifeless | thanks | 21:53 |
| mgagne | mordred: highlight seems to be the "default" proposed tool to do highlight. I will propose pygments instead which is used by github and support more languages | 21:53 |
| mordred | mgagne: yah. we use pygments for paste.o.o too | 21:54 |
| openstackgerrit | A change was merged to openstack-dev/pbr: Fixes issue with command escaping on Windows https://review.openstack.org/41283 | 21:54 |
| openstackgerrit | A change was merged to openstack-dev/pbr: Stop checking periods in commit messages https://review.openstack.org/40838 | 21:54 |
| mordred | alexpilotti: ^^ | 21:54 |
| mordred | alexpilotti: as soon as the next two land, I will cut a new release for you | 21:54 |
| alexpilotti | mordred: sorry, I just got reconnected | 21:55 |
| *** dina_belova has quit IRC | 21:55 | |
| *** rnirmal has quit IRC | 21:56 | |
| openstackgerrit | A change was merged to openstack-dev/pbr: Fix python-ldap mirroring. https://review.openstack.org/40732 | 21:57 |
| mordred | come on baby. you can do one more | 21:57 |
| fungi | jeblair: still churning through scrollback, but a nodepool.o.o fqdn seems fime | 21:57 |
| fungi | fine | 21:57 |
| mordred | I think it seems fime too | 21:57 |
| mtreinish | jeblair: now I'm not so sure of those numbers. It shows 65 successes and 27 failures for parallel and 64 successes and 14 failures for serial. | 21:59 |
| mtreinish | shouldn't they be about the same | 21:59 |
| jeblair | mtreinish: it may have aborted some | 21:59 |
| mtreinish | oh ok | 21:59 |
| *** anteaya has quit IRC | 22:00 | |
| *** alexpilotti_ has joined #openstack-infra | 22:01 | |
| *** alexpilotti has quit IRC | 22:02 | |
| *** vipul is now known as vipul-away | 22:03 | |
| *** emagana has quit IRC | 22:03 | |
| marun | I'm curious - are some projects simply not configured to update launchpad bug status from gerrit events? | 22:03 |
| *** datsun180b has quit IRC | 22:04 | |
| clarkb | marun: yes | 22:04 |
| clarkb | however that is mostly stackforge projects. all openstack/* projects should work just fine | 22:05 |
| clarkb | though we occasionally find one that doesn't have the correct user in the bug driver group | 22:05 |
| marun | clarkb: Is devstack an exception? | 22:05 |
| clarkb | marun: do you have a specific example? | 22:05 |
| *** alexpilotti has joined #openstack-infra | 22:05 | |
| *** alexpilotti_ has quit IRC | 22:05 | |
| *** burt has quit IRC | 22:05 | |
| clarkb | marun: devstack shouldn't be. can you link to a change that isn't doing the right bug stuff? | 22:05 |
| marun | will do | 22:05 |
| marun | https://review.openstack.org/#/c/41270/ | 22:05 |
| marun | This was submitted on sunday and merged yesterday. | 22:06 |
| *** dkliban has quit IRC | 22:06 | |
| marun | No automatic updates to the bug status that I can see. | 22:06 |
| marun | https://bugs.launchpad.net/devstack/+bug/1210664 | 22:06 |
| uvirtbot | Launchpad bug 1210664 in devstack "neutron can't setup basic network connnectivity in gate jobs" [Undecided,Fix released] | 22:06 |
| *** mrmartin has quit IRC | 22:06 | |
| clarkb | marun: that commit message is using the old format | 22:07 |
| marun | clarkb: ah, I'm doing it wrong? | 22:07 |
| openstackgerrit | Mathieu Gagné proposed a change to openstack-infra/config: Cleanup cgit interface https://review.openstack.org/42016 | 22:07 |
| clarkb | marun: yeah, I am trying to find a link to the email thread that explains it all | 22:07 |
| clarkb | marun: http://lists.openstack.org/pipermail/openstack-dev/2013-August/012945.html is the thread that announced the changes | 22:08 |
| clarkb | marun: http://lists.openstack.org/pipermail/openstack-dev/2013-August/013424.html is a thread asking a similar question to yours | 22:08 |
| Alex_Gaynor | wowa, are we really out of devstack nodes, not sure I've seen that before | 22:09 |
| *** alexpilotti has quit IRC | 22:09 | |
| *** vipul-away is now known as vipul | 22:09 | |
| *** weshay has quit IRC | 22:09 | |
| clarkb | jeblair: are we running on one jenkins right now? | 22:09 |
| marun | clarkb: not that it's your responsibility, but this new implementation is a usability fail. | 22:09 |
| *** wannatest has joined #openstack-infra | 22:10 | |
| clarkb | marun: how so? this was proposed back in san diego as a usability win | 22:10 |
| clarkb | and it was brought up again in portland iirc | 22:10 |
| marun | clarkb: all he had to do was allow 'bug xyz' as well as 'bug #xyz' and the old functionality would have continued to work | 22:10 |
| fungi | marun: mostly, it's the fact that "fixes" doesn't come at the beginning of a line and is preceded by * so the backwards-compat regex in the new matching doesn't catch it, but yeah use the new style headers in the future and rejoice | 22:10 |
| *** acabrera has joined #openstack-infra | 22:10 | |
| *** acabrera has left #openstack-infra | 22:10 | |
| marun | fungi: fair enough | 22:11 |
| *** jerryz has quit IRC | 22:12 | |
| zaro | jeblair: not sure if this related to the latest patch, but i'm seeing lost jobs. | 22:12 |
| jeblair | zaro: where? | 22:12 |
| fungi | i am pleased enough that we can now differentiate complete, partial and related bug references in commit messages that i can personally cope with the minor change in workflow | 22:13 |
| jeblair | clarkb: yes | 22:13 |
| *** mkirk_ has joined #openstack-infra | 22:13 | |
| jeblair | zaro: the word "LOST" does not appear on the zuul status page | 22:14 |
| zaro | jeblair: when i run i see all of the jobs i submitted go on the gearman queue but not all of them get run by jenkins. | 22:14 |
| *** alexpilotti has joined #openstack-infra | 22:14 | |
| *** mrodden has quit IRC | 22:14 | |
| zaro | jeblair: it seems like gearman will passes out number of jobs equal to the number of available workers to run them. then it doesn't run anymore jobs that are on the queue. | 22:15 |
| zaro | jeblair: i am still using the C gearman server ver 1.1.7, could that be the problem? | 22:16 |
| *** alexpilotti has quit IRC | 22:16 | |
| *** alexpilotti has joined #openstack-infra | 22:16 | |
| openstackgerrit | A change was merged to openstack-infra/gearman-plugin: Use more fine-grained synchronization in GearmanProxy https://review.openstack.org/42003 | 22:16 |
| *** alexpilotti has quit IRC | 22:19 | |
| *** sarob_ has joined #openstack-infra | 22:20 | |
| jeblair | zaro: i am unable to replicate that. can you be more specific about how many jobs you are running, when, and how may executors are available to run them? | 22:20 |
| jeblair | zaro: or try again with geard? | 22:20 |
| *** gyee has quit IRC | 22:21 | |
| zaro | jeblair: sorry about that. i have a problem with the way i was submitting jobs in my client. | 22:22 |
| jeblair | zaro: https://jenkins01.openstack.org/job/gearman-plugin-hpi-artifact/3/console | 22:23 |
| *** sarob has quit IRC | 22:24 | |
| fungi | dstufft: now you're taunting rubygems people on oss-security. you know they're new to this whole internet thing... be gentle ;) | 22:24 |
| * Alex_Gaynor goes to find oss-security and popcorn | 22:24 | |
| jeblair | zaro: 2013-08-14 22:19:28.323 | [EnvInject] - Unset unresolved 'PROJECT_VER' variable. | 22:24 |
| *** sarob_ has quit IRC | 22:24 | |
| jeblair | zaro: that doesn't make much sense to me. | 22:25 |
| *** blamar has quit IRC | 22:25 | |
| dstufft | fungi: oh I wasn't actually :] I was just trying to say that pip has a CVE for the same thing so it probably makes sense to assign one, and making sure that they actually verify TLS and don't just use HTTPS w/o verifying since that's barely better than HTTP | 22:25 |
| fungi | Alex_Gaynor: though network security news of the week is that windows ping-of-death has returned wearing a cloak of ipv6 | 22:25 |
| zaro | jeblair: looking into it. | 22:26 |
| Alex_Gaynor | dstufft: can we have cert pinning for pip and pypi? | 22:26 |
| fungi | dstufft: yes, i agree that their fix is almost certainly too premature to be useful to their community | 22:26 |
| dstufft | Alex_Gaynor: It's on my list | 22:26 |
| Alex_Gaynor | dstufft: <3 | 22:26 |
| mordred | dstufft: what about pnies. can we have pnies? | 22:26 |
| Alex_Gaynor | mordred: I have a pony on my desk | 22:27 |
| fungi | Alex_Gaynor: dstufft: definitely. i don't want to trust joe ca, bait and tackle with verifying my pypi downloads any longer than needed | 22:27 |
| dstufft | Alex_Gaynor: Ideally I want to see if there's something that can be done generically so people can pin certs other than PyPI too. But worst case we'll just hardcode in the CA or something | 22:27 |
| dstufft | I need to talk to Noah and probably fastly to make sure we can count on them using digicert | 22:27 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Switch etherpad_lite backups to mysql_backup. https://review.openstack.org/41791 | 22:28 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Add new mysql_backup module. https://review.openstack.org/41790 | 22:28 |
| clarkb | fungi: ^ addressed your comments, thanks | 22:28 |
| *** _TheDodd_ has quit IRC | 22:29 | |
| fungi | dstufft: something in .pip/pip.conf which overrides the verification ca directory or cert would be great | 22:29 |
| dstufft | fungi: you can already use your own ca bundle | 22:29 |
| fungi | dstufft: wfm then | 22:29 |
| mordred | dstufft: also, can I use pip to install ruby gems? | 22:30 |
| *** ^d has quit IRC | 22:30 | |
| mordred | what about dbs? | 22:30 |
| mordred | debs | 22:30 |
| mordred | damn keyboard | 22:30 |
| clarkb | mordred: so I am about to say something that I may regret. I think the tools around gems are better than pip | 22:30 |
| dstufft | I forget the exact command for it, but it should be settable via cli flag, env var, or conf file | 22:30 |
| jeblair | restarting jenkins02 | 22:30 |
| clarkb | but virtualenv is better than whatever they use | 22:30 |
| fungi | dstufft: though also being able to treat a single public cert as the ca bundle to match a remote server would be great | 22:30 |
| clarkb | now I need to go wash my mouth out or otherwise punish myself | 22:30 |
| fungi | dstufft: meaning, "signed by or actually being the same as this cert" | 22:31 |
| * mordred admonishes clarkb | 22:31 | |
| clarkb | mordred: because they get dependency resolution correct | 22:31 |
| mordred | clarkb: dstufft is working on that | 22:31 |
| dstufft | fungi: should be able to make a single cert bundle, we just pass that shit into OpenSSL | 22:31 |
| dstufft | clarkb: I'm learning how to write SAT solver | 22:31 |
| fungi | dstufft: then what else is needed for pinning? | 22:31 |
| *** vipul is now known as vipul-away | 22:31 | |
| dstufft | fungi: to handle it for random people automatically | 22:31 |
| mordred | dstufft: I hear lifeless may have spent some time in a former life thinking about dependency graphs, btw... | 22:32 |
| mordred | dstufft: in python | 22:32 |
| fungi | dstufft: ahh, so providing a default which is neither "all the ca certs you have in /etc/ssl" nor "whatever i find on the server" | 22:32 |
| Alex_Gaynor | former lives are the best lives | 22:32 |
| fungi | Alex_Gaynor: none of my former lives are better than this one | 22:33 |
| clarkb | mordred: there is a dependency graph solver thin in testr/testools/subunit I forget which | 22:33 |
| clarkb | but it doesn't deal with ranges which is where SAT comes in iirc | 22:33 |
| mordred | yah | 22:33 |
| mordred | dstufft: also, I keep forgetting - what's the story with markerlib and pip? | 22:33 |
| mordred | dstufft: is it that there is no story if I do pip install -r requirements.txt - but there is a way to make it consume the information if it's in install_requires? | 22:34 |
| dstufft | fungi: basically pip is used for sites other than PyPI itself so we want to encourage people to use TLS and making that easier by shipping with a sane ca bundle is a generally good thing, but to make PyPI more secure we want to pin PyPI's cert. But we need to make sure we can trust that it will be digicert for the foreseable future, and that we don't lock out other servers from using other certs | 22:34 |
| dstufft | mordred: requirements.txt doesn't support anything like markerlib correct | 22:35 |
| fungi | dstufft: ahh, digicert better live up to your expectations then ;) | 22:35 |
| jeblair | this is a job for dnssec | 22:35 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Switch etherpad_lite backups to mysql_backup. https://review.openstack.org/41791 | 22:35 |
| fungi | jeblair: i couldn't agree more | 22:35 |
| dstufft | DNSSEC you just trust someone different than the CA's | 22:35 |
| clarkb | ^ last one assuming that doesn't do anything crazy | 22:35 |
| uvirtbot | clarkb: (last [--{from,in,on,with,without,regexp} <value>] [--nolimit]) -- Returns the last message matching the given criteria. --from requires a nick from whom the message came; --in requires a channel the message was sent to; --on requires a network the message was sent on; --with requires some string that had to be in the message; --regexp requires a regular expression the message must match; --nolimit returns (1 more message) | 22:35 |
| clarkb | sorry uvirtbot | 22:35 |
| *** markmcclain has quit IRC | 22:36 | |
| fungi | jeblair: i will trust the isc farther than i can kick any commercial ca | 22:36 |
| dstufft | it just moves the problem, doesn't actually solve it :V | 22:36 |
| *** markmcclain has joined #openstack-infra | 22:36 | |
| jeblair | ^ last sorry | 22:37 |
| uvirtbot | jeblair: (last [--{from,in,on,with,without,regexp} <value>] [--nolimit]) -- Returns the last message matching the given criteria. --from requires a nick from whom the message came; --in requires a channel the message was sent to; --on requires a network the message was sent on; --with requires some string that had to be in the message; --regexp requires a regular expression the message must match; --nolimit returns (1 more message) | 22:37 |
| clarkb | jeblair: you did it too :) | 22:37 |
| jeblair | ^ last --with sorry | 22:37 |
| uvirtbot | jeblair: [22:37:12] <jeblair> ^ last sorry | 22:37 |
| fungi | jeblair: soren won't like you hacking his 'bot | 22:38 |
| jeblair | i'm excited. i actually learned a uvirtbot command | 22:38 |
| mordred | wiw | 22:38 |
| mordred | yeah | 22:38 |
| fungi | heh | 22:38 |
| mordred | I like that a lot | 22:38 |
| jeblair | it might even be useful | 22:38 |
| dstufft | fungi: jeblair We use DynEct at PSF so we should have DNSSEC capabilities, need to figure out a way to get it into pip when the network stack pip is running on might not support it | 22:38 |
| mordred | ^ last from soren | 22:38 |
| uvirtbot | mordred: (last [--{from,in,on,with,without,regexp} <value>] [--nolimit]) -- Returns the last message matching the given criteria. --from requires a nick from whom the message came; --in requires a channel the message was sent to; --on requires a network the message was sent on; --with requires some string that had to be in the message; --regexp requires a regular expression the message must match; --nolimit returns (1 more message) | 22:38 |
| jeblair | ^ last --from soren | 22:39 |
| uvirtbot | jeblair: Error: I couldn't find a message matching that criteria in my history of 1000 messages. | 22:39 |
| jeblair | aww that is sad | 22:39 |
| mordred | it is sad | 22:39 |
| clarkb | only 1000 messages. I talk to much for that buffer | 22:39 |
| mordred | yah | 22:39 |
| fungi | dstufft: that is a problem for the future. everyone should be supporting dnssec recursion by the time you have a good answer, right? | 22:39 |
| zaro | jeblair: i think build problem is a intermittent bug with the jenkins plugin. | 22:39 |
| zaro | jeblair: for the same patch 42003 | 22:39 |
| jeblair | zaro: what? | 22:40 |
| zaro | jeblair: jenkins run for patch #1 did not set env correctly, #3 worked correctly, then last one didn't work. | 22:40 |
| fungi | anyway, i trust the former bsd greybeards at isc way more than netsol/verisign/symantec, so hedging my bets on dnssec | 22:40 |
| *** vipul-away is now known as vipul | 22:41 | |
| jeblair | zaro: ah, you mean it shows up in the gate-build job too. i agree | 22:41 |
| zaro | jeblair: actually the good and bad runs are on different precise slaves. so it might be puppet not getting updated? | 22:42 |
| jeblair | zaro: very possible; can you give me a good and bad slave? | 22:42 |
| *** pentameter has quit IRC | 22:43 | |
| zaro | jeblair: bad one was precise5, good one is precise15 | 22:43 |
| jeblair | zaro: /usr/local/jenkins/slave_scripts/maven-properties.sh is the same on both | 22:45 |
| *** ryanpetrello has quit IRC | 22:45 | |
| zaro | jeblair: can we manually run same build on precise5? | 22:45 |
| lifeless | dstufft: mordred: I have spent a ridiculous amount of time with dependency graphs | 22:46 |
| jeblair | zaro: not easily. do you think that the first build on a host fails? | 22:47 |
| zaro | jeblair: don't think so because it failed to set the env twice on precise15. | 22:48 |
| *** changbl has joined #openstack-infra | 22:48 | |
| jeblair | zaro: how are the variables supposed to be injected? i only see that that script is called; but that's not supposed to set variables. | 22:48 |
| zaro | jeblair: the jenkins run on patch set 3 and #4 both ran on precise15 | 22:49 |
| zaro | jeblair: vars injected with wrapper | 22:49 |
| jeblair | zaro: perhaps only the first one succeeds? | 22:49 |
| zaro | jeblair: vars injected with wrapper '-inject' | 22:49 |
| zaro | jeblair: it's in jenkins-plugin-jobs.yaml | 22:49 |
| zaro | jeblair: first one? | 22:50 |
| *** prad_ has quit IRC | 22:50 | |
| dstufft | lifeless: heh, my current thoughts are using a SAT solver, but I'm learning about them first :] | 22:50 |
| *** dina_belova has joined #openstack-infra | 22:50 | |
| marun | lifeless: for tempest, tox now uses testtools instead of nose, but on rhel64 (py26) testtools complains that without py27 the discover module must be installed. | 22:50 |
| marun | lifeless: is there any reason not to add discovery to tempest's list of dependencies so py26 'just works'? | 22:51 |
| clarkb | marun: add discover to the test-requirements or requirements | 22:51 |
| clarkb | marun: there isn't we should be doing this for all of the othe rprojects too | 22:51 |
| mordred | marun: yeah. we have discover in the test-requirements.txt in several projects | 22:51 |
| clarkb | however | 22:51 |
| mordred | I do not fully understand why it is needed but anecdotally, it is | 22:51 |
| jeblair | zaro: when i look in the web interface, i only see the script file path, not the prorperties file path. | 22:51 |
| clarkb | tempest + testr + py26 won't get along in all causes due to setUpClass aiui | 22:52 |
| marun | clarkb: are you telling me testr is now broken on rhel64?!?!?! | 22:52 |
| clarkb | marun: yes | 22:52 |
| marun | <unprintable> | 22:52 |
| clarkb | and not now, always has been | 22:52 |
| zaro | jeblair: which web ui? | 22:52 |
| jeblair | zaro: jenkins | 22:52 |
| marun | clarkb: Was anyone from redhat cognizant of this fact? | 22:53 |
| mordred | clarkb: didn't mtreinish fix that? | 22:53 |
| marun | clarkb: because that seems like a pretty huge foobar. | 22:53 |
| clarkb | marun: we have avoided this issue by not using setUpClass because setUpClass is extremely broken | 22:53 |
| jeblair | zaro: can you look into fixing this? you may want to set up a jenkins server, with a slave, and use jjb. | 22:53 |
| clarkb | marun: its not a fubar | 22:53 |
| zaro | jeblair: ohh yes, it's because there are two instacnes of '-inject' in the jjb job. | 22:53 |
| clarkb | marun: setUpClass is bad bad bad, don't use it | 22:53 |
| zaro | jeblair: only one will appear on the UI. | 22:53 |
| *** jerryz has joined #openstack-infra | 22:53 | |
| marun | clarkb: I'm confused. | 22:53 |
| *** ^d has joined #openstack-infra | 22:53 | |
| jeblair | zaro: pretty sure that means it's broken | 22:53 |
| clarkb | marun: it doesn't work properly with nose either, so nose was also broken | 22:53 |
| marun | clarkb: is or is tempest not broken right now? | 22:53 |
| *** ^d has quit IRC | 22:53 | |
| *** ^d has joined #openstack-infra | 22:53 | |
| marun | (on py26) | 22:54 |
| clarkb | tempest + testr + py26 does not quite work right | 22:54 |
| clarkb | in theory tempest + nose + py26 mostly works but the issue is around setUpClass which nose does not do properly | 22:54 |
| clarkb | so the same bits of code were always broken, the severity of brokeness is changing | 22:54 |
| marun | clarkb: Tempest was working fine with nose as of 2 weeks ago, but I don't run the full suite. | 22:54 |
| *** dina_belova has quit IRC | 22:55 | |
| jeblair | zaro: that could be why it didn't work. | 22:55 |
| marun | clarkb: What symptoms can I expect trying py26 + testr? | 22:55 |
| clarkb | marun: nose + setUpClass breaks around failures | 22:55 |
| zaro | jeblair: ok. i thought that maybe we could set multiple instances of a jenkins plugin in jjb. but i guess it's not reliable. | 22:55 |
| Alex_Gaynor | mtreinish: FWIW the numbers might be slightly off about testr parallel stuff right now, there's some horizon/tempest issues which is causing a ton of failures :( | 22:55 |
| jeblair | zaro: i would not do anything that the web ui doesn't let you do. | 22:55 |
| clarkb | marun: anything that depends on a setUpClass to prep things will probably fail | 22:55 |
| clarkb | marun: because testr will not run setUpClass under py26 | 22:55 |
| zaro | jeblair: dangit! will fix. | 22:56 |
| marun | clarkb: *sigh* | 22:56 |
| jeblair | zaro: if there's no other way to deal with it, i'd consider making the job a freestyle job and run maven directly | 22:56 |
| clarkb | so instead of seeing brokeness occasionally you will get it 100% of the time | 22:56 |
| marun | clarkb: well, I appreciate the insight | 22:56 |
| mordred | clarkb, marun: I believe the suggestion for fixing that was "put code in the setupClass into a fixture, but have the setupclass load the fixture" | 22:56 |
| mtreinish | Alex_Gaynor: yeah I noticed that I'm working on a way to get accurate stats right now. | 22:56 |
| clarkb | mordred: that won't fix it | 22:56 |
| zaro | jeblair: cool. i suggested that to clarkb but he was resistant. | 22:56 |
| mordred | clarkb, marun: then add a testresources for the class, which only testr will run | 22:57 |
| lifeless | marun: you should add discover yes, but it's not a complete answer | 22:57 |
| mordred | and not nose | 22:57 |
| clarkb | mordred: you need something else to load the fixture because setUpClass is not run by py26 unittest | 22:57 |
| Alex_Gaynor | mtreinish: I would have suggseted yelling at horizon people people as a solution, but I guess better stats is a reasonable approach | 22:57 |
| mtreinish | Alex_Gaynor: it's hard to filter out a random failure that's not parallel specific but occurs randomly on a parallel run | 22:57 |
| clarkb | mordred: or that | 22:57 |
| mtreinish | Alex_Gaynor: the yelling will come tomorrow | 22:57 |
| marun | lifeless: I'm mainly wanting to keep the quantum smoke test running. Sounds like I need to update to testfixture asap. | 22:57 |
| clarkb | but we erally need to get off ot he setUpClass crutch and stop acting like it wasn't already broken | 22:57 |
| mordred | right. and _then_ we can start pulling the setupcalsses out | 22:57 |
| mtreinish | Alex_Gaynor: right now I'm going to go home for the day | 22:57 |
| Alex_Gaynor | mtreinish: night, enjoy | 22:58 |
| jeblair | putting jenkins01 in shutdown mode | 22:58 |
| clarkb | mtreinish: ftr nose + setUpClass is broken in some cases where setUpClass will throw and exception, nose will then skip all tests in that class and report success | 22:58 |
| clarkb | marun: ^ | 22:58 |
| jeblair | and jenkins.o.o | 22:58 |
| marun | clarkb: Sure, that's easy to fix though. | 22:58 |
| clarkb | marun: it should report failure but doesn't | 22:58 |
| clarkb | marun: it is? | 22:58 |
| marun | clarkb: add error handling to setupClass | 22:58 |
| marun | clarkb: No thrown exception, nose doesn't lose it's shit. | 22:59 |
| mordred | or stop using setupclass :) | 22:59 |
| clarkb | marun: but then no tests run and you report success | 22:59 |
| clarkb | marun: you need some way of getting nose to die gracefully and report failure | 22:59 |
| clarkb | there doesn't seem to be a way to do that | 22:59 |
| clarkb | without fixing nose because their setUpClass stuff is a hack | 22:59 |
| dstufft | pytest! | 22:59 |
| dstufft | :] | 23:00 |
| dstufft | (because the solution to a problem in tool a is switch to tool b you see) | 23:00 |
| mordred | marun: https://pypi.python.org/pypi/fixtures + https://pypi.python.org/pypi/testresources | 23:00 |
| marun | clarkb: From one hack to another, merrily we go... | 23:00 |
| clarkb | dstufft: when tool a is no longer maintained... | 23:00 |
| clarkb | marun: there is no hack in the new stuff :) it just properly doesn't work | 23:00 |
| mordred | dstufft: or - just actually finish the current migration :) | 23:00 |
| mordred | dstufft: :) | 23:00 |
| marun | clarkb: sure, keep telling yourself ;) | 23:00 |
| mordred | gah | 23:00 |
| clarkb | marun: all that said, fixtures + testresources is so much better | 23:01 |
| mordred | LAG LAG LAG NETWORK | 23:01 |
| dstufft | mordred: :D | 23:01 |
| marun | clarkb: I like it, don't get me wrong. but it's not free. It is necessarily more complicated. | 23:01 |
| marun | The price of progress. | 23:01 |
| clarkb | is it more complicated? | 23:01 |
| marun | One more api to learn | 23:01 |
| clarkb | I ported all of nova's unittests without any prior nova unittest knowledge. | 23:02 |
| marun | *sigh* | 23:02 |
| marun | I'm just getting worn out keeping up with all these infra changes and it makes it hard to get real work done. | 23:02 |
| marun | /rant | 23:02 |
| lifeless | dstufft: what problem are you solving ? | 23:02 |
| marun | bad choice of words, though. real == stuff i have to do. | 23:03 |
| *** danger_fo is now known as danger_fo_away | 23:03 | |
| clarkb | marun: it is fair that it is a new api and that requires some amount of learning. My experience learning them has been that you get better tests because the tools are simpler not more complicated | 23:03 |
| dstufft | lifeless: Given a set of version constraints and a set of available versions find out which versions can be installed to satisfy all the constraints | 23:04 |
| clarkb | the cost is >0 | 23:04 |
| clarkb | but the long term cost is far less than dragging along the brokeness that we have/had | 23:04 |
| dstufft | lifeless: trivially written is a boolean expression afaict | 23:04 |
| dstufft | as a* | 23:04 |
| marun | clarkb: i just wear too many hats (neutron, tempest, rh) and do all of them poorly as you can tell. | 23:04 |
| pleia2 | goes through massive patch and leaves comments on wrong set \o/ | 23:04 |
| dstufft | lifeless: https://gist.github.com/dstufft/86ee27f147e53e19fc66 | 23:04 |
| clarkb | pleia2: the worst part is when new patchset is pushed just before you leave your comments ;) | 23:05 |
| pleia2 | clarkb: yeah | 23:05 |
| mordred | marun: you wear them great | 23:05 |
| lifeless | dstufft: ok, so 'what to install' :> | 23:05 |
| jeblair | jenkins, jenkins-dev, and jenkins02 have all been upgraded to ${project-version} of gearman-plugin | 23:05 |
| jeblair | yay. | 23:05 |
| clarkb | jeblair: woot | 23:06 |
| lifeless | dstufft: have you thought about how that interacts with incremental discovery ? | 23:06 |
| mordred | jeblair: yay! | 23:06 |
| mordred | jeblair: my favorite version! | 23:06 |
| lifeless | dstufft: by which I mean pip's basically recursing at the moment to build an egg etc | 23:07 |
| dstufft | lifeless: the paper i'm reading is for an incremental SAT solver, so as more version specs are found I can add additional constraints efficiently resolve the problem, and see if there are new things I need to download | 23:08 |
| dstufft | so it'd still do recursing for right now, until the new API is done on PyPI which allows querying that info | 23:09 |
| dstufft | (and new packaging formats) | 23:09 |
| lifeless | dstufft: I'm thinking about 'install version A, then find A is blacklisted' | 23:09 |
| dstufft | lifeless: pip doesn't install anything until after all the dependencies are resolved | 23:09 |
| dstufft | it jsut downloads them and runs egg-info | 23:09 |
| lifeless | dstufft: oh | 23:09 |
| lifeless | dstufft: so setup_requires isn't pip managed | 23:09 |
| marun | lifeless: how best to conditionally skip a test class using testr? The standard nose-centric way in tempest is to use a check in setupClass. | 23:10 |
| mordred | yeah. setup_requires is a little bitch | 23:10 |
| dstufft | lifeless: correct. Although I have plans to make it pip managed, but setup_requires aren't globally installed they are installed only for that invocation and I want to make it pip controlled and available only to the package that setup_require'd it | 23:10 |
| *** dhellmann is now known as dhellmann_ | 23:10 | |
| lifeless | marun: put a check at the top of setUp if you want it to affect every test | 23:11 |
| marun | lifeless: so instead of per-class it becomes per-method? | 23:11 |
| dstufft | I was talking with jaraco (Setuptools maintainer) the other day about making a sort of setup.py.py egg-info --setup-requires which will just give you the setup requires and won't install anything | 23:11 |
| lifeless | marun: it always is per method | 23:11 |
| lifeless | marun: the shim that makes it look per class in nose is a lie | 23:12 |
| dstufft | so pip can go get them | 23:12 |
| lifeless | dstufft: please god please | 23:12 |
| mordred | dstufft: O M G +1000 | 23:12 |
| marun | lifeless: Fair enough. And if the check is expensive (check if neutron is reachable), would you suggest a way to cache that result for future checks? | 23:12 |
| lifeless | dstufft: I /so/ want run-mirror to not have to install stuff to build a targeted mirror | 23:12 |
| lifeless | marun: you can stash a variable on the class trivially | 23:13 |
| marun | lifeless: ok | 23:13 |
| lifeless | marun: or use a TestResource | 23:13 |
| mordred | dstufft: what can I do to to help that? | 23:13 |
| *** dnavale has joined #openstack-infra | 23:13 | |
| dstufft | lifeless: pip will still need to sort of install stuff (because the point of setup_requires is that they can change how setup.py is interpreted), but it will just install it to a temporary directory which is only made available to the setup.py that requested it | 23:13 |
| lifeless | dstufft: so yes incremental SAT sounds like a good approach | 23:14 |
| dstufft | mordred: I need to make a patch against setuptools and pip to verify it will work, but jaraco seemed amiable to adding it | 23:14 |
| jeblair | mordred: have you reviewed 38177 yet? | 23:14 |
| lifeless | mordred: something odd with https://review.openstack.org/#/q/status:open+project:openstack-dev/pbr,n,z | 23:15 |
| mordred | lifeless: what? | 23:15 |
| dstufft | Also amiable to pulling pkg_resources out of setuptools so projects that use entry points can jsut depend on pkg_resources, which is one of the major blockers to making setuptools not a required dependency of pip | 23:15 |
| mordred | jeblair: I wrote 38177 | 23:16 |
| jeblair | mordred: i know. | 23:16 |
| lifeless | mordred: oh, I loaded it at the wrong time; its fine :) | 23:16 |
| jeblair | mordred: it looks like the output of a search and replace. several people have pointed out obivous errors in it already. i started reviewing it and found some as well. | 23:16 |
| jeblair | mordred: i just want to know if you've reviewed it yourself. | 23:16 |
| jeblair | it's a big patch and a lot of work to review. | 23:17 |
| mordred | jeblair: locally via git diff - but good point - let me walk through it ttw real quick | 23:17 |
| jeblair | mordred: k, thx | 23:17 |
| dstufft | mordred: so I don't want to get anyones hopes up, but if this all works in the future we can have a pip that won't requier setuptools (so none of this setuptools breaks pip nonesense), where ``pip install`` never uses setuptools except as a build tool to actually prefer the install (like it does today) but pip always does the downloading and dependency resolver, and an actual resolver instead of "first specifier is the only specifier" | 23:18 |
| dstufft | s/prefer/preform/ | 23:18 |
| jeblair | mordred: it looks like pleia2 actually just finished what i started. | 23:18 |
| mordred | dstufft: I would like that a lot | 23:18 |
| mordred | oh neat! | 23:18 |
| dstufft | mordred: oh, and probably pip will be available by default in Python too, possibly only 3.4+ but I'm going to try and get Linux distros and stuff to do it for all versions (if I can get them to do it at all) | 23:19 |
| dstufft | s/in Python/with Python/ | 23:20 |
| mordred | dstufft: I don't want it in python by default | 23:20 |
| dstufft | mordred: it'll still be indendepntly upgradeable | 23:20 |
| dstufft | it's not getting added to the stdlib | 23:20 |
| dstufft | it'll just be bundled with Python (or in Linux land python and pip will depend on each other) | 23:20 |
| dstufft | so that if you have Python you also have pip | 23:20 |
| mordred | dstufft: that's how shit gets broke | 23:20 |
| mordred | because you can never fix it | 23:20 |
| *** nati_ueno has quit IRC | 23:21 | |
| dstufft | mordred: the outcome will be exactly the same as if you immediately installed pip after installing Python | 23:21 |
| dstufft | it will just do it for you | 23:21 |
| dstufft | https://github.com/dstufft/peps/blob/master/pip-bundling.rst <- in progress | 23:21 |
| fungi | dstufft: i have someone asking what tool to use these days to create and update a full private pypi mirror for their company... bandersnatch or is there something better? | 23:23 |
| *** dnavale has quit IRC | 23:24 | |
| dstufft | fungi: bandersnatch will mirror PyPI | 23:24 |
| dstufft | it's the best tool for that ATM | 23:24 |
| *** nati_ueno has joined #openstack-infra | 23:24 | |
| fungi | dstufft: perfect. i only skim distutils-sig unfortunately. your reinforcement of my recollection is much appreciated | 23:24 |
| dstufft | fungi: ther'es also dev-pi if they just want a caching mirror that also supports uploading private packages | 23:24 |
| fungi | perhaps. i'll suggest that too | 23:25 |
| dstufft | bandersnatch mirrors everything up front, dev-pi does it on demand | 23:25 |
| dstufft | I think dev-pi also caches things not hosted on PyPI so there's that too | 23:25 |
| fungi | ahh | 23:25 |
| clarkb | jeblair: I just remembered. The bug that has us turnning off some compression in jenkins to avoid the runaway threads was fixed in a semi recent jenkins release which we should now be running | 23:26 |
| dstufft | https://pypi.python.org/pypi/devpi/0.9.4 | 23:26 |
| clarkb | jeblair: should we try removing the workaround from the default file? | 23:26 |
| dstufft | fungi: devpi actually works by having multiple indexes inside on install I believe | 23:26 |
| dstufft | so you have one index that's a on demand mirror of PyPI (and external things if I recall), then you have extra indexes for your own personal private repos | 23:26 |
| dstufft | s/repos/packages/ | 23:26 |
| fungi | huh, neat | 23:27 |
| clarkb | jaypipes: subnets and networks are different :P | 23:27 |
| dstufft | so it depends on what they want to do exactly, but either of those should work fine | 23:27 |
| *** jerryz_ has joined #openstack-infra | 23:27 | |
| fungi | dstufft: thanks again. i passed along the suggestions | 23:27 |
| dstufft | bandersnatch is just a more or less byte for byte copy of what's on PyPI | 23:27 |
| dstufft | fungi: yup np | 23:27 |
| *** jerryz has quit IRC | 23:28 | |
| fungi | dstufft: yeah, bandersnatch talks about implementing pep-381 and then in turn pep-381 talks about getting permission/registering your mirror, so it was daunting (tried to explain pep-449 is ripping out the official mirrors anyway) | 23:28 |
| dstufft | fungi: yea 381 has both a mirroring protocol in it, and a method for registering and naming your mirror, 449 removes the latter part | 23:29 |
| fungi | 381 shoulda been two peps in my opinion, but hindsight yadda yadda | 23:30 |
| dstufft | yea | 23:30 |
| dstufft | it's ok 449 removes part of 381, and then a future PEP will remove the other part of 381 | 23:30 |
| dstufft | whenever I get around to making a new mirroring protocol | 23:30 |
| fungi | s/new/actually secure/ | 23:31 |
| dstufft | yea basically | 23:31 |
| dstufft | basically everything was broken | 23:31 |
| dstufft | and we're trying to inplace fix it all :V | 23:31 |
| fungi | i followed that much of the thread. wholeheartedly approve | 23:31 |
| openstackgerrit | Monty Taylor proposed a change to openstack-infra/config: Use cgit server instead of github for everything https://review.openstack.org/38177 | 23:33 |
| *** zehicle has quit IRC | 23:33 | |
| *** ryanpetrello has joined #openstack-infra | 23:35 | |
| mordred | pleia2: thanks for finding those! good point on the help blurbs - so I just completely reworked them | 23:37 |
| lifeless | mordred: we're still going to mirror to github right ? | 23:37 |
| pleia2 | mordred: yeah, looks great :) | 23:37 |
| pleia2 | lifeless: yes | 23:37 |
| mordred | lifeless: yes. apparently some people still think its useful for something | 23:38 |
| mordred | lifeless: but we're going to be able to remove more of our dependnce on it | 23:38 |
| *** fbo is now known as fbo_away | 23:38 | |
| mordred | lifeless: qI've also batted around the idea of also replicating to bitbucket, so as that we're not playing favorites | 23:39 |
| mordred | lifeless: but I don't think we care enough to do that yet :) | 23:39 |
| lifeless | mordred: meh | 23:40 |
| mgagne | mordred: I actually prefer browsing code using the github interface, but it's a matter of personal preference I guess | 23:40 |
| jeblair | clarkb, mordred: regarding mysql backups, wouldn't it be better to back all of the databases on the server rather than specifying the database name? | 23:41 |
| lifeless | mgagne: git clone; vim. :) | 23:41 |
| mgagne | lifeless: sure, when I'm on my laptop which isn't always the case unfortunately | 23:42 |
| mordred | jeblair: I like the named database approach, because it will map easier to having a single backup host and a bunch of cloud databases in the future | 23:42 |
| jeblair | mordred: i don't see how that affects it... | 23:42 |
| jeblair | mordred: i'm assuming we'd get a different set of credentials for each database, yeah? | 23:43 |
| mordred | jeblair: yeah. good point | 23:43 |
| fungi | mgagne: part of the goal, i think, is to get the cgit browsing experience close enough to as usable as github that you don't care so much, and then there's how much faster it runs | 23:43 |
| *** vipul is now known as vipul-away | 23:43 | |
| *** vipul-away is now known as vipul | 23:43 | |
| mordred | jeblair: ok, I'm down with that too | 23:43 |
| *** UtahDave has quit IRC | 23:44 | |
| mgagne | fungi: that's why I'm proposing updating the skin it sucks less =) | 23:44 |
| jeblair | mordred: so yeah, we'd want to say "backup gerrit" and "backup paste", but each of those would back up all of related schemas, (so "all of gerrit" and then "all of paste") | 23:44 |
| mordred | yup. makes sense to me | 23:44 |
| fungi | mgagne: no argument1 | 23:44 |
| jeblair | k, will leave comment to that effect | 23:44 |
| mordred | mgagne: if you're a god, you could update cgit ui to have a slightly better click on line to get highligted line link - and also add support for shift-click ranges | 23:45 |
| *** vipul is now known as vipul-away | 23:45 | |
| mordred | but you're doing a greatjob so far | 23:45 |
| mgagne | mordred: my C skills are a little rusty =) | 23:45 |
| mordred | mgagne: bah. I'm sure that's html | 23:45 |
| mgagne | mordred: you would be surprised | 23:45 |
| clarkb | jeblair: we can do it that way too | 23:46 |
| *** zehicle has joined #openstack-infra | 23:47 | |
| jeblair | jenkins01 is idle; upgrading plugin there | 23:47 |
| mordred | jeblair, fungi: I'm trying a new method for keeping track of reviews I need to do that clarkb and I talked about yesterday | 23:50 |
| mordred | I think it's super helpful so far | 23:50 |
| *** dina_belova has joined #openstack-infra | 23:50 | |
| jeblair | mordred: good. have you added nodepool to your watched list? :) | 23:51 |
| *** pcrews has quit IRC | 23:51 | |
| mordred | yes I have - and it's also in my starred list now | 23:51 |
| clarkb | jeblair: for backing up everything, basicaly keep ${name} everywhere it is currently used but remove it from the mysqldump command and use the option to backup everything there? | 23:51 |
| mordred | new method: use this: https://review.openstack.org/#/q/watchedby:mordred%2540inaugust.com+-label:CodeReview%253C%253D-1+-label:Verified%253C%253D-1+-label:Approved%253E%253D1++-status:workinprogress+-status:draft+-is:starred+-owner:mordred%2540inaugust.com,n,z | 23:51 |
| fungi | mordred: using the pretty sparkly stars? | 23:51 |
| mordred | to browse list of things I need to review and star them | 23:51 |
| fungi | indeed | 23:51 |
| mordred | then use https://review.openstack.org/#/q/is:starred+-label:CodeReview%253C%253D-1+-label:Verified%253C%253D-1,n,z | 23:52 |
| fungi | stars >> ponies | 23:52 |
| mordred | to go through the things I need to review, and I can unstar them when I'm done with them | 23:52 |
| fungi | i intend to try the same workflow | 23:52 |
| mordred | so that once my star list is depleted, I can re-assess the first list | 23:52 |
| mordred | helps my ADD not get overwhelmed | 23:52 |
| fungi | first i intend to finish catching up on e-mail and then fix the other things i broke. how did it get dark already? | 23:53 |
| jeblair | mordred: cool, i have taken notes on that and will try it. | 23:53 |
| jeblair | mordred: (though so far, important changes is still mostly working for me) | 23:53 |
| mordred | jeblair: you may want to replace the email address | 23:53 |
| mordred | jeblair: great! | 23:53 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Switch etherpad_lite backups to mysql_backup. https://review.openstack.org/41791 | 23:53 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Add new mysql_backup module. https://review.openstack.org/41790 | 23:54 |
| clarkb | jeblair: ^ that what you have in mind? | 23:54 |
| jeblair | jenkins01 is back up | 23:54 |
| fungi | yeah, i mostly stick to the important changes dashboard, but also my review throughput it very, very terrible | 23:54 |
| fungi | so anything to improve efficiency there is worth trying | 23:54 |
| *** tjones has joined #openstack-infra | 23:55 | |
| *** dina_belova has quit IRC | 23:55 | |
| jeblair | clarkb: yeah | 23:56 |
| tjones | is there any way to programmatically write to etherpad or pastebin or wiki? I want to write a cron job to generate a report which is publicly available. I could put it in my google drive but thought i'd check here 1st… | 23:56 |
| clarkb | tjones: yes for pastebin. Not sure about wiki or etherpad | 23:57 |
| fungi | tjones: mediawiki has an api, etherpad has an import capability though may require pretending to be a browser a little | 23:57 |
| fungi | tjones: and pastebinit can deal with paste.o.o | 23:57 |
| mordred | clarkb: I +2d the second change - but left it un approved so you can sheperd it in | 23:57 |
| jeblair | tjones: i've used the mediawiki api for this page: https://wiki.openstack.org/wiki/Infrastructure_Status | 23:57 |
| clarkb | mordred: thanks | 23:58 |
| tjones | thanks folks! | 23:58 |
| fungi | tjones: any time! | 23:58 |
| clarkb | it will run the backups on etherpad dev shortly and if that looks good I will apply it to etherpad.o.o tomorrow most likely | 23:58 |
| jeblair | tjones: i think you would need to talk to Ryan_Lane to get an account for the wiki | 23:58 |
| jeblair | clarkb, fungi, mordred: i'm about to run, but with the new plugins in place (thanks for your help!), i hope to run nodepool for real tomorrow. | 23:59 |
| mordred | jeblair: awesome | 23:59 |
| clarkb | ++ | 23:59 |
| tjones | jeblair: thx. I'll ask him | 23:59 |
| fungi | jeblair: i hope to review that still. thanks for all the great work! | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!