| tristanC | pabelanger: have you seen https://review.openstack.org/#/c/532084 ? | 00:38 |
|---|---|---|
| *** rlandy has quit IRC | 00:58 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: WIP: Test using install-kubernetes role https://review.openstack.org/605803 | 01:00 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: WIP: Test using install-kubernetes role https://review.openstack.org/605803 | 01:08 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: WIP: Test using install-kubernetes role https://review.openstack.org/605803 | 01:14 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: WIP: Test using install-kubernetes role https://review.openstack.org/605803 | 01:27 |
| *** jiapei has joined #zuul | 01:39 | |
| tristanC | Shrews: here, nodepool created 2 nodes on minikube, one namespace and one pod: http://logs.openstack.org/03/605803/15/check/nodepool-functional-k8s/9b8cbed/ubuntu-xenial/nodepool/launcher.log | 01:42 |
| *** Shrews has quit IRC | 02:32 | |
| pabelanger | tristanC: no, didn't see that one. Quick look, it is a little difference. Opted to just use the zuul-executor, since we got ansible-galaxy for free. playbook: https://github.com/ansible-network/zuul-config/blob/master/playbooks/publish/galaxy.yaml job: https://github.com/ansible-network/zuul-config/blob/master/zuul.d/jobs.yaml#L70 | 02:32 |
| *** Shrews has joined #zuul | 02:32 | |
| pabelanger | tristanC: however, i didn't consider using zuul.tag for --branch, will need to confirm with team that will work | 02:32 |
| pabelanger | if so, happy to switch to that also | 02:32 |
| pabelanger | tristanC: https://object-storage-ca-ymq-1.vexxhost.net/v1/a0b4156a37f9453eb4ec7db5422272df/logs/8c/8c62c1f69a17ba41f1d1bf81485dff20f1ab69ff/post/release-ansible-role/c210f12/job-output.html is example job from zuul-executor | 02:36 |
| pabelanger | will catch up in the morning, EOD | 02:37 |
| *** rcarrillocruz has quit IRC | 02:51 | |
| tristanC | pabelanger: why couldn't you use 532084 on the executor? | 03:02 |
| *** rcarrillocruz has joined #zuul | 04:16 | |
| *** nilashishc has joined #zuul | 04:20 | |
| *** rcarrillocruz has quit IRC | 04:21 | |
| *** rcarrillocruz has joined #zuul | 04:25 | |
| *** nilashishc has quit IRC | 04:29 | |
| *** nilashishc has joined #zuul | 04:34 | |
| *** quiquell|off is now known as quiquell | 05:46 | |
| *** snapiri has quit IRC | 06:12 | |
| *** swest has quit IRC | 06:13 | |
| *** swest has joined #zuul | 06:15 | |
| *** klep has quit IRC | 06:24 | |
| *** quiquell is now known as quiquell|brb | 06:37 | |
| *** quiquell|brb is now known as quiquell | 07:09 | |
| *** electrofelix has joined #zuul | 07:20 | |
| *** goern has joined #zuul | 07:29 | |
| *** nilashishc has quit IRC | 07:34 | |
| *** nilashishc has joined #zuul | 07:36 | |
| *** chandankumar has joined #zuul | 07:40 | |
| *** jpena|off is now known as jpena | 07:44 | |
| *** snapiri has joined #zuul | 08:19 | |
| *** chandankumar has quit IRC | 08:51 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add install and deploy openshift roles. https://review.openstack.org/608610 | 08:52 |
| *** sshnaidm is now known as sshnaidm|lnch | 09:40 | |
| *** bhavikdbavishi has joined #zuul | 09:47 | |
| *** bhavikdbavishi has quit IRC | 09:52 | |
| *** sshnaidm|lnch is now known as sshnaidm | 09:59 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 10:06 |
| *** jpena is now known as jpena|off | 10:07 | |
| *** nilashishc has quit IRC | 10:14 | |
| *** nilashishc has joined #zuul | 10:15 | |
| *** quiquell is now known as quiquell|brb | 10:20 | |
| *** quiquell|brb is now known as quiquell | 10:21 | |
| *** ssbarnea_ has quit IRC | 10:24 | |
| *** quiquell is now known as quiquell|brb | 10:53 | |
| *** rlandy has joined #zuul | 12:04 | |
| *** quiquell|brb is now known as quiquell | 12:20 | |
| *** samccann has joined #zuul | 12:46 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add install and deploy openshift roles. https://review.openstack.org/608610 | 12:52 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 13:05 |
| Shrews | tristanC: cool! | 13:10 |
| tristanC | Shrews: 570667 should get a similar result with openshift | 13:13 |
| Shrews | i'm not certain that a role is going to end up being the best way to test a new driver in the long run, but at least it's something | 13:13 |
| Shrews | i wasn't thinking something along the lines of a new tox environment (e.g, tox -ek8s) so we can do deep testing like we do with the openstack driver, but i haven't given much thought as to what that would require | 13:15 |
| tristanC | Shrews: i don't mind either way, though i still think we should test zuul, zuul-jobs and zuul-base-jobs too for completeness | 13:17 |
| Shrews | s/wasn't thinking/was thinking/ | 13:17 |
| Shrews | tristanC: oh yeah, totally | 13:17 |
| Shrews | i'll remove the WIP comment from 605823 since that seems to work | 13:20 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul-jobs master: Add role to install kubernetes https://review.openstack.org/605823 | 13:21 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/nodepool master: Ignore removed provider in _cleanupLeakedInstances https://review.openstack.org/608670 | 13:22 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add install and deploy openshift roles. https://review.openstack.org/608610 | 13:56 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 13:56 |
| Shrews | corvus: mordred: how would you feel about removing the tox-cover job from nodepool? AFAIA, none of us ever really check the output, so it seems wasteful | 14:37 |
| corvus | Shrews: wfm | 14:37 |
| corvus | easy to add back if/when we care | 14:38 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Remove tox-cover job https://review.openstack.org/608688 | 14:38 |
| clarkb | fwiw I added the job back when nodepool was flaky and the coverage was useful to understand what was undertested. less of a problem today | 14:39 |
| tobiash | Shrews: I looked at it sometimes (but also can just run that locally) | 14:41 |
| corvus | we can also add it to 'experimental' | 14:41 |
| tobiash | I'm fine with removing it | 14:41 |
| tobiash | just wanted to mention that it's not completely useless :) | 14:42 |
| *** quiquell is now known as quiquell|off | 14:44 | |
| pabelanger | +1 for experimental | 14:44 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/nodepool master: WIP: Second try to fix single cloud config reload https://review.openstack.org/608695 | 14:59 |
| tobiash | corvus, Shrews: this would be another approach to fix the single cloud reconfiguration use case ^ | 15:00 |
| tobiash | this further doesn't rely on the taskmanager's stoppedexception | 15:00 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Add tox-py36 job https://review.openstack.org/608700 | 15:14 |
| *** nilashishc has quit IRC | 15:20 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Add tox-py36 job https://review.openstack.org/608700 | 15:21 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 15:21 |
| *** panda is now known as panda|off | 15:23 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Create ansible-galaxy-import role https://review.openstack.org/608472 | 15:34 |
| corvus | pabelanger: should we just start working on support for maser instead? or do you have immediate use for current galaxy? | 15:36 |
| pabelanger | corvus: yah, imeediate use for current galaxy, ansible-network team releases every 2 weeks. I am also working on mazer role, but think it only works today against galaxy-qa server. | 15:37 |
| pabelanger | immediate* | 15:37 |
| pabelanger | corvus: I also don't mind running that role downstream until mazer is live | 15:38 |
| corvus | pabelanger: i think that role is fine, was just curious | 15:38 |
| corvus | pabelanger: i was thinking in zuul itself we could probably skip supporting galaxy and only support mazer (for having zuul install roles for running a job) | 15:39 |
| corvus | (but again, a role for current galaxy in zuul-jobs seems like a good idea) | 15:39 |
| pabelanger | Yah, I think both ideas are fine | 15:40 |
| pabelanger | as I under stand, once mazer is live, galaxy import will no longer work | 15:41 |
| corvus | i'm going to go into a corner and work on https://review.openstack.org/608344 for a bit | 15:46 |
| Shrews | nobody puts corvus in a corner | 15:47 |
| * Shrews apologizes for 80's movies reference | 15:48 | |
| *** sshnaidm is now known as sshnaidm|afk | 15:51 | |
| *** ssbarnea|bkp2 has joined #zuul | 15:54 | |
| openstackgerrit | Merged openstack-infra/nodepool master: Remove tox-cover job https://review.openstack.org/608688 | 16:07 |
| *** panda|off has quit IRC | 16:22 | |
| *** panda has joined #zuul | 16:23 | |
| clarkb | I've approved the additon of a test workspace setup role to zuul jobs so I can test a change to the actual role for that in base-test | 16:46 |
| clarkb | will approve the base test update once that goes in | 16:47 |
| *** sshnaidm|afk is now known as sshnaidm | 16:48 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Add test workspace setup role https://review.openstack.org/608342 | 17:03 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: docker-compose quickstart example https://review.openstack.org/608344 | 17:22 |
| corvus | \o/ i just ran a job on that ^ | 17:22 |
| corvus | i think i've got the sequencing worked out, so 'docker-compose up' will bring everything up. i worked around the ssh host key issue for now too, so we don't have to wait on that. | 17:23 |
| corvus | next up is log publication, then writing some more words about setting up jobs :) | 17:24 |
| Shrews | corvus: i left a few minor comments on the previous PS for that | 17:24 |
| Shrews | i'm curious how it worked, given my comments | 17:25 |
| corvus | Shrews: oh cool i'll take a look | 17:28 |
| Shrews | if the only difference in the containers is the default command, perhaps we should reconsider how pbrx works | 17:28 |
| *** ssbarnea_ has joined #zuul | 17:29 | |
| *** ssbarnea|bkp2 has quit IRC | 17:30 | |
| corvus | Shrews: ah yes! zuul daemonizes by default, and that's not ideal in containers, so we override to use "-d". the only difference in the containers is the command, so since we're not using that anyway, i just used zuul/zuul. i think the best solution to this is to change the default behavior of zuul and nodepool to run in the foreground by default, and switch '-d' to mean daemonize. when we do that, we can | 17:32 |
| corvus | switch to using the different pbrx images. | 17:32 |
| *** sshnaidm is now known as sshnaidm|afk | 17:32 | |
| corvus | (also some systemd folks may prefer zuul to run in the foreground) | 17:33 |
| corvus | and finally, we should separate the daemonization flag from the debug log flag. | 17:34 |
| Shrews | corvus: If that's the plan (which I like), then perhaps we should start off the document using the containers we intend to use (to prevent confusing new users) | 17:34 |
| Shrews | or maybe they won't even bother looking at the docker compose file, in which case, it's irrelevant | 17:35 |
| corvus | Shrews: good point. i'll switch them. in case they do look. :) | 17:35 |
| corvus | (early on, it was helpful for me to have a shorter docker image list, but i'm past that now) | 17:35 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul-jobs master: Do no merge testing with base-test https://review.openstack.org/608732 | 17:36 |
| corvus | does anyone happen to know if mordred's change to fix the javascript in the container merged and if we published a new container image with that? | 17:50 |
| clarkb | corvus: https://review.openstack.org/#/c/608411/ seems to have merged which should have ended up in a new docker image | 17:52 |
| clarkb | unsure if that needs a release though https://review.openstack.org/#/c/607638/1/.zuul.yaml | 17:52 |
| corvus | clarkb: looks like openstackzuul-pbrx-push-container-images is the job and runs in post | 17:55 |
| corvus | so yeah, it should be there. but i'm still getting 404s | 17:59 |
| corvus | if anyone feels like poking at that, that's a good parallel task that will help with the quickstart | 18:03 |
| Shrews | corvus: you get 404s on your running container? | 18:04 |
| corvus | Shrews: yep, localhost:9000 | 18:04 |
| *** nilashishc has joined #zuul | 18:04 | |
| Shrews | did you expose the port? | 18:04 |
| corvus | Shrews: yes, the 404 is from cherrypy | 18:04 |
| *** electrofelix has quit IRC | 18:05 | |
| corvus | i'm assuming that since i switched the image to zuul-web, docker will have pulled the latest; i guess it's possible it didn't though. i should figure out how to double check that | 18:05 |
| corvus | zuul/zuul-web latest 6e629f6d6a98 8 days ago 236 MB | 18:06 |
| corvus | that looks rather old actually | 18:06 |
| corvus | i'll go delete stuff and try again | 18:06 |
| *** j^2 has joined #zuul | 18:07 | |
| tobiash | this one will fix the disk accountant in the zuul docker images: https://review.openstack.org/606314 | 18:10 |
| Shrews | corvus: i'm running a shell on a zuul-web container... i don't see any zuul static files | 18:11 |
| Shrews | trying to figure out where they're supposed to be | 18:11 |
| *** nilashishc has quit IRC | 18:13 | |
| tobiash | Shrews: I think I have a hack in my images for this too (in the dockerfile) | 18:13 |
| tobiash | Shrews: I think they need to be in <installroot>/pythonx/site-packages/zuul/web/static/ | 18:14 |
| Shrews | doesn't seem to me that the pbr setup_hook has been run at all. | 18:14 |
| pabelanger | currently doing a budget for Berlin (zuul forum), which hotel have people booked with? Seems all are a bit from the conference | 18:15 |
| Shrews | tobiash: yeah, they're not there | 18:15 |
| Shrews | # ls /usr/local/lib/python3.7/site-packages/zuul/web | 18:15 |
| Shrews | __init__.py __pycache__ handler.py | 18:15 |
| tobiash | definitely missing | 18:16 |
| tobiash | is the source dir still in the image? | 18:16 |
| Shrews | oh | 18:16 |
| tobiash | I think my hack was to move it from the source dir (I clone zuul in the dockerfile) to the correct dir | 18:16 |
| Shrews | i wonder if this is it: | 18:16 |
| Shrews | def _build_javascript(): | 18:16 |
| Shrews | if subprocess.call(['which', 'yarn']) != 0: | 18:16 |
| Shrews | return | 18:16 |
| Shrews | 'which yarn' returns 1 (not found) | 18:16 |
| tobiash | makes sense | 18:17 |
| Shrews | so the code below it to setup the static files doesn't execute | 18:17 |
| corvus | Shrews: my understanding is the build container is supposed to have yarn installed | 18:17 |
| corvus | (but, if it doesn't, that would certainly be a problem) | 18:17 |
| corvus | the final zuul-web container shouldn't have it though | 18:17 |
| tobiash | is pbrx a two step process? | 18:17 |
| Shrews | corvus: maybe the build container does, but zuul-web doesn't. i'm not sure when this setup part gets executed | 18:18 |
| corvus | Shrews: it's supposed to happen at build time | 18:18 |
| tobiash | in this case the bug is probably that the js stuff is just not installed and thrown away with the source | 18:18 |
| clarkb | tobiash: yes there is a build container that is thrown away, and all of the artifacts from that container are installed cleanly into the target containers with wheels so that you don't have build artifacts left in the image | 18:18 |
| clarkb | the change above https://review.openstack.org/#/c/608411/ was supposed to be the fix for that I think | 18:18 |
| tobiash | ok, so that makes my hack impossible ;) | 18:18 |
| clarkb | maybe check if building a local sdist/wheel works at all if yarn is installed | 18:19 |
| tobiash | btw, the react rewrite fixes the installation of the js stuff | 18:19 |
| Shrews | yarn is not installed on zuul-base either | 18:19 |
| Shrews | :( | 18:19 |
| corvus | Shrews: it's not supposed to be | 18:20 |
| corvus | the container with yarn installed is discarded; it should not be in any of our published images | 18:20 |
| corvus | (or, i should say, the container which is *supposed* to have yarn installed, i don't know that it does) | 18:21 |
| Shrews | corvus: all of the images are based off of the *-base image | 18:21 |
| Shrews | corvus: so your statement confuses me | 18:21 |
| corvus | Shrews: building is a two step process: a "build container" is created with yarn installed. it builds a wheel. the wheel is installed into the final image that we publish. | 18:21 |
| Shrews | where is this process? pbrx certainly isn't doing that | 18:22 |
| Shrews | and pbrx is the only process i'm aware of that builds images that we publish | 18:23 |
| corvus | tobiash: mordred updated the react patch with the same fix for building/installing js that has already landed; it should work the same either way | 18:23 |
| tobiash | ah ok | 18:23 |
| Shrews | corvus: oh, maybe it's python-base container that has this. | 18:25 |
| Shrews | i forgot about that "other" intermediary container | 18:25 |
| corvus | Shrews: http://git.openstack.org/cgit/openstack/pbrx/tree/pbrx/container_images.py#n215 | 18:25 |
| corvus | yes that | 18:25 |
| *** panda has quit IRC | 18:26 | |
| corvus | and yarn [compile test platform:apk] | 18:26 |
| corvus | is in zuul's bindep which should cause it to be installed there when we do the bdist_wheel | 18:27 |
| *** panda has joined #zuul | 18:28 | |
| openstackgerrit | Merged openstack-infra/zuul master: Add coreutils to bindep on platform apk https://review.openstack.org/606314 | 18:42 |
| corvus | i've approved the react patch | 18:43 |
| pabelanger | Yay | 18:45 |
| Shrews | oh, hahah. corvus, i think i found the issue | 18:52 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul master: Fix pbr setup hook https://review.openstack.org/608753 | 18:53 |
| Shrews | corvus: ^^ | 18:53 |
| Shrews | let's wait to see the output to confirm it's getting called, but i think that's it | 18:54 |
| openstackgerrit | Merged openstack-infra/zuul master: Revert "Revert "web: rewrite interface in react"" https://review.openstack.org/607479 | 19:08 |
| *** samccann has quit IRC | 19:10 | |
| *** samccann has joined #zuul | 19:10 | |
| tobiash | \o/ | 19:31 |
| *** samccann_ has joined #zuul | 19:43 | |
| *** samccann_ has quit IRC | 19:43 | |
| tobiash | corvus: just tried the docker-compose but it doesn't work behind a mandatory proxy. I guess that's fine for the start? | 20:00 |
| corvus | tobiash: ah yeah... though if you know how to fix that, we should update it to support that :) | 20:09 |
| corvus | Shrews: wow nice | 20:09 |
| clarkb | tobiash: is it failing to dowload the images? everyting else should be local right? | 20:10 |
| tobiash | clarkb: some docker builds fail: http://paste.openstack.org/show/731712/ | 20:11 |
| corvus | we could build all the necessary images; it just seemed a little weird. maybe we can pass the necessary env variable to docker | 20:14 |
| tobiash | corvus: with a mandatory proxy we need to pass the proxy vars as build args to the docker build | 20:16 |
| tobiash | they will be available as env vars then inside the build | 20:16 |
| clarkb | corvus: or reuse the centos image that gerrit is already pulling in? | 20:16 |
| Shrews | corvus: right? i can't find anything in the output to indicate it actually ran though | 20:18 |
| tobiash | corvus: added a comment to 608344 that fixes the build of the node image for me behind the proxy | 20:18 |
| corvus | clarkb: can you elaborate? are you suggesting the centos image has everything we need for a build node already installed? | 20:20 |
| clarkb | corvus: yes because centos uses yum which requires python | 20:21 |
| clarkb | (however I don't know if it has an sshd in it) | 20:21 |
| corvus | tobiash: that looks reasonable | 20:21 |
| tobiash | corvus: same for gerrit | 20:21 |
| corvus | and gerritconfig? | 20:21 |
| tobiash | I'm waiting for the gerrit build to finish | 20:22 |
| tobiash | the proxy via vpn is not that superfast ;) | 20:22 |
| corvus | oh.. you know what, we might be able to use the upstream gerrit image now | 20:24 |
| corvus | i think the only change in there was the ssh-keygen and i don't think it's necessary any more | 20:25 |
| clarkb | password: secret ftw | 20:25 |
| corvus | yep. that's a day i'm never getting back. but i learned a lot about gerrit and docker. | 20:26 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: docker-compose quickstart example https://review.openstack.org/608344 | 20:28 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Support mandatory proxy in docker-compose example https://review.openstack.org/608764 | 20:30 |
| tobiash | corvus: that fixes the docker builds ^ | 20:31 |
| tobiash | feel free to squash or throw away | 20:31 |
| corvus | tobiash: thanks i'll squash that into my change now, you can go ahead and abandon | 20:31 |
| tobiash | ok | 20:31 |
| tobiash | corvus: scheduler startup fails on my laptop: http://paste.openstack.org/show/731713/ | 20:35 |
| corvus | tobiash: ah i think i understand | 20:37 |
| tobiash | I don't ;) | 20:37 |
| tobiash | but it's late | 20:37 |
| corvus | it's a misleading error which could mean any kind of authentication error | 20:37 |
| corvus | i don't think the user has been created yet; we probably need the scheduler to wait a bit more before starting | 20:38 |
| corvus | when i went through this the first time, i didn't have any projects in the tenant config. | 20:38 |
| corvus | so... there's two ways we could fix this: start with an empty tenant config and have the user add the necessary projects, or have the scheduler wait for the account to exist before starting. | 20:38 |
| corvus | tobiash: if you ctrl-c and then 'docker-compose up' again i think it should work the second time | 20:39 |
| tobiash | trying | 20:39 |
| tobiash | gerritconfig failed: http://paste.openstack.org/show/731714/ | 20:43 |
| tobiash | but that also could well be caused by the proxy | 20:43 |
| corvus | tobiash: which task was that? | 20:45 |
| tobiash | it couldn't resolve 'gerrit' | 20:45 |
| tobiash | add admin ssh key to gerrit | 20:46 |
| corvus | tobiash: hrm. docker is supposed to do some magic to do internal dns resolution | 20:47 |
| corvus | and gerritconfig "depends-on" gerrit so it should be up when it runs | 20:47 |
| tobiash | strance is that 'get first login screen' doesn't output an error | 20:47 |
| tobiash | s/strance/strange | 20:47 |
| corvus | it has ignore_errors set to true | 20:48 |
| corvus | (because the first time that url is hit, it normally is an error) | 20:49 |
| *** pcaruana has quit IRC | 20:49 | |
| tobiash | ah and it will be the same 404 | 20:49 |
| tobiash | I think that 404 comes from the docker proxy of my docker for mac | 20:50 |
| corvus | tobiash: is docker causing the request from the gerritconfig container to go through your corporate proxy on the way to the gerrit container? | 20:52 |
| tobiash | currently checking that | 20:52 |
| tobiash | maybe the proxy vars are injected into the containers (which is bad) | 20:52 |
| corvus | tobiash: i think maybe that's the case: https://docs.docker.com/docker-for-mac/#proxies | 20:52 |
| corvus | tobiash: so it might just be ansible honoring the proxy env var | 20:53 |
| tobiash | yeah, docker for mac and proxies guarantees various problems... | 21:00 |
| tobiash | cannot get it to work today... | 21:00 |
| corvus | tobiash: maybe you can set the proxy env variable for the gerritconfig container... it shouldn't need to use the network | 21:00 |
| corvus | the external network that is | 21:01 |
| tobiash | it shouldn't be set (build args are not persistent) | 21:02 |
| tobiash | and I also tried to explicitly unset it in the ansible task | 21:02 |
| clarkb | you should be able to se a noproxy var for specific ips or ramges | 21:03 |
| corvus | tobiash: i'm not 100% sure setting it in the task will help depending on how the uri module is implemented (it might only look at the env var that ansible-playbook was invoked with) | 21:04 |
| corvus | tobiash: i meant that docker-compose lets you set env variables for the containers too (not just the build) | 21:05 |
| corvus | tobiash, clarkb: maybe we can add "no_proxy=${no_proxy},gerrit,..." to the docker-compose file... | 21:05 |
| corvus | (though it seems really weird to me that we would need to explicitly add the vars in the build but not running but whatever docker) | 21:07 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: docker-compose quickstart example https://review.openstack.org/608344 | 21:08 |
| corvus | tobiash: ^ see what i did there for gerritconfig | 21:08 |
| tobiash | yes, that worked | 21:09 |
| tobiash | and scheduler is running | 21:10 |
| corvus | yay! i think my scheduler.yaml playbook still isn't right | 21:10 |
| corvus | i'm trying cycles of "docker rm examples_gerrit_1; docker volume prune" to verify | 21:11 |
| tobiash | corvus: but I would recomment a slightly different version: http://paste.openstack.org/show/731715/ | 21:11 |
| corvus | that's needed to remove all the persistent state in gerrit | 21:11 |
| tobiash | as I think completely unsetting these variables is cleaner and more robust | 21:11 |
| corvus | tobiash: ok... i figured the no_proxy thing might be safest, in case something does want to access the network later? | 21:12 |
| tobiash | corvus: well, if you want to add stuff later that wants to download things then yes | 21:13 |
| corvus | tobiash: it seems like the most generally correct thing to do -- if it works, it should solve the problem and be future proof (and it might get copy/pasted to something else), so if it works i'd like to use just no_proxy. | 21:15 |
| tobiash | ack | 21:15 |
| corvus | tobiash: should the 'https_proxy' line use the https_proxy variable? (it's currently <<https_proxy: "${http_proxy}">> ) | 21:16 |
| openstackgerrit | Merged openstack-infra/zuul master: Fix pbr setup hook https://review.openstack.org/608753 | 21:16 |
| tobiash | yes | 21:16 |
| tobiash | I was just lazy as it's the same here ;) | 21:17 |
| tobiash | I did a prune and docker-compose up starts up completely on docker on mac and with proxy | 21:18 |
| tobiash | I think that's the most challenging combination ;) | 21:18 |
| corvus | tobiash: yay! i think i just about have the scheduler playbook right, so my next patchset should work for your env and mine from scratch | 21:20 |
| tobiash | corvus: cool, will try it tomorrow | 21:21 |
| * tobiash calls it a day | 21:21 | |
| corvus | tobiash: thanks! | 21:22 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: docker-compose quickstart example https://review.openstack.org/608344 | 21:44 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Revert "Revert "Revert "web: rewrite interface in react""" https://review.openstack.org/608779 | 21:45 |
| corvus | mordred: when you're around, can we chat about the js tarball thing? there are two issues we should resolve: 1) it's an undocumented deployment method. 2) it's untested. openstack should, like everyone else, really be deploying zuul via a method that is both documented and tested. (cc tristanC, tobiash, fungi) | 21:50 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: docker-compose quickstart example https://review.openstack.org/608344 | 22:11 |
| mordred | corvus: I agree with taht premise - I'm afraid my in-flight wifi is potentially too spotty for me to be able to participate in a meaninful discussion on the correct fix ... but let me say a few thoughts and if we get lucky maybe it can even be close to a chat - but I might also disappear midthought | 22:53 |
| mordred | corvus: for our draft jobs to work, we need to deploy the static javascript code to logs.o.o - perhaps we should rework those jobs to extract the contents of the built tarball into place? | 22:54 |
| mordred | corvus: alternately we can keep the draft jobs as they are and change openstack's deploy to proxy the static content served by zuul-web instead of serving the content directly from apache | 22:55 |
| mordred | ok. that's not a ton of thoughts - but those are my main two thoughts | 22:55 |
| clarkb | apache is set to cache the data coming frkm zuul right? | 22:57 |
| clarkb | maybe thats good enough and zuul can serve the content | 22:57 |
| mordred | yah - it should be fine to do that | 22:59 |
| mordred | I think it should also make the rewrite rules much simpler | 23:00 |
| *** rlandy is now known as rlandy|bbl | 23:11 | |
| *** quiquell|off has quit IRC | 23:27 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!