| *** rlandy is now known as rlandy|bbl | 00:01 | |
| *** ianychoi has joined #zuul | 00:31 | |
| dmsimard | I don't want to get into the implementation details but TripleO is doing some pretty cool things with ara.. they're exporting data from ara to influxdb and then they have graphs for things that interest them. I wonder what it would take to do this kind of thing at scale in Zuul. | 01:14 |
|---|---|---|
| *** ianychoi has quit IRC | 01:16 | |
| dmsimard | Host and task granular metrics about all tasks in every jobs ? | 01:16 |
| *** ianychoi has joined #zuul | 01:16 | |
| dmsimard | In case you're curious what it looks like, it's open source :D https://github.com/openstack/tripleo-quickstart-extras/blob/master/roles/collect-logs/library/ara_influxdb.py | 01:17 |
| dmsimard | http://38.145.35.97/d/poOr-d0mk/ansible-exploration?orgId=1 | 01:17 |
| tristanC | it seems like ansible/awx deleted a tag, and zuul-executor are now unable to start a job because of https://ansible.softwarefactory-project.io/paste/show/21/ | 01:27 |
| tristanC | according to zuul/merger/merger.py it seems like the only solution is to restart merger and executors? | 01:29 |
| tristanC | well, closing and re-opening the PR seems to also be a solution | 01:32 |
| *** openstackgerrit has quit IRC | 01:37 | |
| *** tristanC has quit IRC | 01:37 | |
| *** mugsie has quit IRC | 01:37 | |
| *** dkehn has quit IRC | 01:37 | |
| *** logan- has quit IRC | 01:37 | |
| *** tflink has quit IRC | 01:37 | |
| *** odyssey4me has quit IRC | 01:37 | |
| *** dmellado has quit IRC | 01:37 | |
| *** rlandy|bbl has quit IRC | 01:37 | |
| *** D0han has quit IRC | 01:37 | |
| *** Diabelko has quit IRC | 01:37 | |
| *** jhesketh has quit IRC | 01:37 | |
| *** che-arne has quit IRC | 01:37 | |
| *** gundalow has quit IRC | 01:37 | |
| *** eumel8 has quit IRC | 01:37 | |
| *** maxamillion has quit IRC | 01:37 | |
| *** snapiri has quit IRC | 01:37 | |
| *** jiapei has quit IRC | 01:37 | |
| *** AJaeger has quit IRC | 01:37 | |
| *** spsurya has quit IRC | 01:37 | |
| *** sdake has quit IRC | 01:37 | |
| *** clarkb has quit IRC | 01:37 | |
| *** snk has quit IRC | 01:37 | |
| *** sdoran has quit IRC | 01:37 | |
| *** mgagne has quit IRC | 01:37 | |
| *** jimi|ansible has quit IRC | 01:37 | |
| *** smyers has quit IRC | 01:37 | |
| *** persia has quit IRC | 01:37 | |
| *** gouthamr has quit IRC | 01:37 | |
| *** ssbarnea_ has quit IRC | 01:37 | |
| *** Shrews has quit IRC | 01:37 | |
| *** pleia2 has quit IRC | 01:37 | |
| *** corvus has quit IRC | 01:37 | |
| *** jamielennox has quit IRC | 01:37 | |
| *** sshnaidm|afk has quit IRC | 01:37 | |
| *** jpena|off has quit IRC | 01:37 | |
| *** _ari_ has quit IRC | 01:37 | |
| *** hogepodge has quit IRC | 01:37 | |
| *** jbryce has quit IRC | 01:37 | |
| *** kmalloc has quit IRC | 01:37 | |
| *** mrhillsman has quit IRC | 01:37 | |
| *** cmurphy has quit IRC | 01:37 | |
| *** EmilienM has quit IRC | 01:37 | |
| *** fdegir has quit IRC | 01:37 | |
| *** mattclay has quit IRC | 01:37 | |
| *** panda has quit IRC | 01:37 | |
| *** SpamapS has quit IRC | 01:37 | |
| *** pabelanger has quit IRC | 01:37 | |
| *** jlk has quit IRC | 01:37 | |
| *** eventingmonkey has quit IRC | 01:37 | |
| *** lennyb has quit IRC | 01:37 | |
| *** adam_g has quit IRC | 01:37 | |
| *** bramwelt has quit IRC | 01:37 | |
| *** samccann has quit IRC | 01:37 | |
| *** fungi has quit IRC | 01:37 | |
| *** ttx has quit IRC | 01:37 | |
| *** timrc has quit IRC | 01:37 | |
| *** dcastellani has quit IRC | 01:37 | |
| *** bstinson has quit IRC | 01:37 | |
| *** gothicmindfood has quit IRC | 01:37 | |
| *** mnaser has quit IRC | 01:37 | |
| *** TheJulia has quit IRC | 01:37 | |
| *** mordred has quit IRC | 01:37 | |
| *** tobiash has quit IRC | 01:37 | |
| *** fbo has quit IRC | 01:37 | |
| *** swest has quit IRC | 01:37 | |
| *** rcarrillocruz has quit IRC | 01:37 | |
| *** evrardjp has quit IRC | 01:37 | |
| *** ssbarnea has quit IRC | 01:37 | |
| *** mmedvede has quit IRC | 01:37 | |
| *** anticw has quit IRC | 01:37 | |
| *** timburke has quit IRC | 01:37 | |
| *** ianychoi has quit IRC | 01:37 | |
| *** quiquell|off has quit IRC | 01:37 | |
| *** dmsimard has quit IRC | 01:37 | |
| *** zigo has quit IRC | 01:37 | |
| *** leifmadsen has quit IRC | 01:37 | |
| *** mhu has quit IRC | 01:37 | |
| *** j^2 has quit IRC | 01:37 | |
| *** jesusaur has quit IRC | 01:37 | |
| *** chandankumar has quit IRC | 01:37 | |
| *** ianw has quit IRC | 01:37 | |
| *** jlvillal has quit IRC | 01:37 | |
| *** aluria has quit IRC | 01:37 | |
| *** mestery has quit IRC | 01:37 | |
| *** andreaf has quit IRC | 01:37 | |
| *** robcresswell has quit IRC | 01:37 | |
| *** SotK has quit IRC | 01:37 | |
| *** ChanServ has quit IRC | 01:37 | |
| *** andreaf has joined #zuul | 01:43 | |
| *** mestery has joined #zuul | 01:43 | |
| *** aluria has joined #zuul | 01:43 | |
| *** jlvillal has joined #zuul | 01:43 | |
| *** ianw has joined #zuul | 01:43 | |
| *** jesusaur has joined #zuul | 01:43 | |
| *** chandankumar has joined #zuul | 01:43 | |
| *** j^2 has joined #zuul | 01:43 | |
| *** odyssey4me has joined #zuul | 01:43 | |
| *** tflink has joined #zuul | 01:43 | |
| *** logan- has joined #zuul | 01:43 | |
| *** dkehn has joined #zuul | 01:43 | |
| *** mugsie has joined #zuul | 01:43 | |
| *** mhu has joined #zuul | 01:43 | |
| *** leifmadsen has joined #zuul | 01:43 | |
| *** dmsimard has joined #zuul | 01:43 | |
| *** zigo has joined #zuul | 01:43 | |
| *** quiquell|off has joined #zuul | 01:43 | |
| *** ianychoi has joined #zuul | 01:43 | |
| *** timburke has joined #zuul | 01:43 | |
| *** mmedvede has joined #zuul | 01:43 | |
| *** ssbarnea has joined #zuul | 01:43 | |
| *** anticw has joined #zuul | 01:43 | |
| *** evrardjp has joined #zuul | 01:43 | |
| *** rcarrillocruz has joined #zuul | 01:43 | |
| *** swest has joined #zuul | 01:43 | |
| *** jimi|ansible has joined #zuul | 01:44 | |
| *** smyers has joined #zuul | 01:44 | |
| *** persia has joined #zuul | 01:44 | |
| *** samccann has joined #zuul | 01:44 | |
| *** fungi has joined #zuul | 01:44 | |
| *** ttx has joined #zuul | 01:44 | |
| *** gouthamr has joined #zuul | 01:44 | |
| *** ssbarnea_ has joined #zuul | 01:44 | |
| *** Shrews has joined #zuul | 01:44 | |
| *** corvus has joined #zuul | 01:44 | |
| *** pleia2 has joined #zuul | 01:44 | |
| *** jamielennox has joined #zuul | 01:44 | |
| *** openstackgerrit has joined #zuul | 01:45 | |
| *** robcresswell has joined #zuul | 01:45 | |
| *** SotK has joined #zuul | 01:45 | |
| *** sshnaidm|afk has joined #zuul | 01:45 | |
| *** mrhillsman has joined #zuul | 01:45 | |
| *** _ari_ has joined #zuul | 01:45 | |
| *** jpena|off has joined #zuul | 01:45 | |
| *** hogepodge has joined #zuul | 01:45 | |
| *** jbryce has joined #zuul | 01:45 | |
| *** kmalloc has joined #zuul | 01:45 | |
| *** cmurphy has joined #zuul | 01:45 | |
| *** EmilienM has joined #zuul | 01:45 | |
| *** fdegir has joined #zuul | 01:45 | |
| *** mattclay has joined #zuul | 01:45 | |
| *** che-arne has joined #zuul | 01:45 | |
| *** eumel8 has joined #zuul | 01:45 | |
| *** maxamillion has joined #zuul | 01:45 | |
| *** gundalow has joined #zuul | 01:45 | |
| *** dmellado has joined #zuul | 01:46 | |
| *** rlandy|bbl has joined #zuul | 01:46 | |
| *** D0han has joined #zuul | 01:46 | |
| *** Diabelko has joined #zuul | 01:46 | |
| *** jhesketh has joined #zuul | 01:46 | |
| *** gothicmindfood has joined #zuul | 01:46 | |
| *** bstinson has joined #zuul | 01:46 | |
| *** mnaser has joined #zuul | 01:46 | |
| *** TheJulia has joined #zuul | 01:46 | |
| *** mordred has joined #zuul | 01:46 | |
| *** dcastellani has joined #zuul | 01:46 | |
| *** timrc has joined #zuul | 01:46 | |
| *** panda has joined #zuul | 01:46 | |
| *** adam_g has joined #zuul | 01:46 | |
| *** pabelanger has joined #zuul | 01:46 | |
| *** SpamapS has joined #zuul | 01:46 | |
| *** jlk has joined #zuul | 01:46 | |
| *** eventingmonkey has joined #zuul | 01:46 | |
| *** bramwelt has joined #zuul | 01:46 | |
| *** lennyb has joined #zuul | 01:46 | |
| *** fbo has joined #zuul | 01:46 | |
| *** tobiash has joined #zuul | 01:46 | |
| *** rlandy|bbl has quit IRC | 01:47 | |
| *** ChanServ has joined #zuul | 01:47 | |
| *** card.freenode.net sets mode: +o ChanServ | 01:47 | |
| *** tristanC has joined #zuul | 01:48 | |
| *** mgagne has joined #zuul | 01:49 | |
| *** snapiri has joined #zuul | 01:49 | |
| *** rlandy has joined #zuul | 01:49 | |
| dmsimard | pabelanger: what was the name of the fedora packaging test thing that was sort of similar to zuul ? we heard about it sometime thursday | 02:04 |
| dmsimard | https://pagure.io/standard-test-roles/ I think | 02:10 |
| *** rlandy has quit IRC | 02:21 | |
| pabelanger | dmsimard: that looks right, I didn't really dig into it | 02:32 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix multiple builds query parameters for a single column https://review.openstack.org/593605 | 02:33 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: prevent filter being cleaned on status reload https://review.openstack.org/609190 | 02:44 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: prevent filter being cleaned on status reload https://review.openstack.org/609190 | 03:05 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix zuul-build-dashboard homepage setting https://review.openstack.org/609200 | 03:05 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix support for comma/space separated status filter https://review.openstack.org/609201 | 03:11 |
| tristanC | 609190 and 609201 should fix the two reported issues about the new dashboard | 03:12 |
| dmsimard | tristanC: the other one that came up was the hamburger menu that didn't work properly (i.e, on mobile or when resizing the browser window to be very small) | 03:26 |
| tristanC | oh right, let me fix that now | 03:26 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix hamburger menu collapse https://review.openstack.org/609205 | 03:47 |
| tristanC | dmsimard: there ^ | 03:47 |
| *** bhavikdbavishi has joined #zuul | 03:58 | |
| *** nilashishc has joined #zuul | 03:59 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix support for comma/space separated status filter https://review.openstack.org/609201 | 04:02 |
| *** AJaeger has joined #zuul | 04:25 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add install and deploy openshift roles. https://review.openstack.org/608610 | 04:30 |
| openstackgerrit | Merged openstack-infra/zuul master: web: fix zuul-build-dashboard homepage setting https://review.openstack.org/609200 | 04:37 |
| openstackgerrit | Merged openstack-infra/zuul master: web: prevent filter being cleaned on status reload https://review.openstack.org/609190 | 04:38 |
| *** spsurya has joined #zuul | 04:52 | |
| *** quiquell|off is now known as quiquell | 05:22 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix support for comma/space separated status filter https://review.openstack.org/609201 | 05:35 |
| tobiash | tristanC: what's the filter expected to match? | 05:41 |
| tobiash | tristanC: when I filter for 'nova,neutron' in that change ^ then I also get a tempest item | 05:42 |
| tobiash | does it also filter for job names? | 05:42 |
| tristanC | tobiash: it's because that tempest is a in a queue shared by neutron | 05:44 |
| tobiash | ah ok | 05:44 |
| tristanC | well we could change that behavior, but that's how it used to work iiuc | 05:45 |
| tobiash | if it was like that before, it's fine, just wondered | 05:46 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement a Kubernetes driver https://review.openstack.org/535557 | 05:47 |
| *** swest has quit IRC | 05:49 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 05:52 |
| *** bhavikdbavishi1 has joined #zuul | 06:00 | |
| *** bhavikdbavishi has quit IRC | 06:01 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 06:01 | |
| *** quiquell is now known as quiquell|brb | 06:12 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add install and deploy openshift roles. https://review.openstack.org/608610 | 06:26 |
| *** pcaruana has joined #zuul | 06:50 | |
| *** quiquell|brb is now known as quiquell | 06:54 | |
| *** odyssey4me has quit IRC | 07:07 | |
| *** odyssey4me has joined #zuul | 07:08 | |
| *** mugsie has quit IRC | 07:08 | |
| *** mgagne has quit IRC | 07:09 | |
| *** logan- has quit IRC | 07:09 | |
| *** tflink has quit IRC | 07:09 | |
| *** mgagne has joined #zuul | 07:10 | |
| *** tflink has joined #zuul | 07:10 | |
| *** jiapei has joined #zuul | 07:11 | |
| *** logan- has joined #zuul | 07:11 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: Add install and deploy openshift roles. https://review.openstack.org/608610 | 07:13 |
| *** panda has quit IRC | 07:30 | |
| *** panda has joined #zuul | 07:32 | |
| *** electrofelix has joined #zuul | 07:33 | |
| *** mgagne has quit IRC | 07:34 | |
| *** mgagne has joined #zuul | 07:36 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 07:55 |
| *** swest has joined #zuul | 07:59 | |
| *** swest has quit IRC | 08:05 | |
| *** nilashishc_ has joined #zuul | 08:13 | |
| *** sshnaidm|afk is now known as sshnaidm | 08:22 | |
| *** swest has joined #zuul | 08:22 | |
| *** logan- has quit IRC | 08:23 | |
| *** logan- has joined #zuul | 08:27 | |
| *** rcarrillocruz has quit IRC | 08:30 | |
| *** swest has quit IRC | 08:34 | |
| *** nilashishc has quit IRC | 08:45 | |
| *** nilashishc_ is now known as nilashishc | 08:45 | |
| *** nilashishc has quit IRC | 08:47 | |
| *** nilashishc has joined #zuul | 08:48 | |
| *** swest has joined #zuul | 08:51 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 08:53 |
| *** bhavikdbavishi has quit IRC | 08:54 | |
| *** themroc has joined #zuul | 09:04 | |
| *** bhavikdbavishi has joined #zuul | 09:38 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 09:48 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: expand status panels only when the header is clicked https://review.openstack.org/609348 | 10:15 |
| openstackgerrit | Merged openstack-infra/zuul master: web: fix hamburger menu collapse https://review.openstack.org/609205 | 10:17 |
| *** _KaszpiR_ has joined #zuul | 10:29 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 10:41 |
| openstackgerrit | Merged openstack-infra/zuul master: web: fix support for comma/space separated status filter https://review.openstack.org/609201 | 10:53 |
| *** dkehn has quit IRC | 10:55 | |
| *** jpena|off has quit IRC | 10:58 | |
| *** swest has quit IRC | 10:59 | |
| *** sshnaidm is now known as sshnaidm|afk | 11:04 | |
| quiquell | Hello looks like running jobs are returning wrong stream url in their json | 11:10 |
| *** dkehn has joined #zuul | 11:10 | |
| *** dkehn has quit IRC | 11:15 | |
| *** swest has joined #zuul | 11:22 | |
| tobiash | quiquell: do you have a link? | 11:23 |
| *** sshnaidm|afk is now known as sshnaidm | 11:24 | |
| quiquell | tobiash: http://zuul.openstack.org/api/status | 11:25 |
| quiquell | stream.html?uuid=e82d0b12b9784cee82cb2ebe1bd44f42&logfile=console.log | 11:26 |
| quiquell | does not work | 11:26 |
| openstackgerrit | Simon Westphahl proposed openstack-infra/zuul master: Include enqueue and report timestamps in MQTT msg https://review.openstack.org/609355 | 11:26 |
| quiquell | zuul.openstack.org/stream.html?uuid=e82d0b12b9784cee82cb2ebe1bd44f42&logfile=console.log | 11:26 |
| tobiash | tristanC, mordred, corvus: I got the feedback from my users that the fonts in the new web ui are a little bit too small (I also have the impression), would you agree with that? | 11:26 |
| quiquell | tobiash: I think the stream url has change but the API is still composing it wrong | 11:26 |
| tobiash | quiquell: do you know if that changed (I think there was no change to this related to the new web ui)? | 11:27 |
| quiquell | tobiash: Yep just after the new web ui | 11:27 |
| quiquell | tobiash: was working fine before | 11:27 |
| quiquell | tobiash: I can open a bug for it | 11:27 |
| tobiash | so it's an api problem as the ui is working right? | 11:28 |
| quiquell | tobiash: If I click on the ui it's working fine | 11:28 |
| tobiash | ah, the stream url is relative now | 11:28 |
| quiquell | tobiash: It's like this now http://zuul.openstack.org/stream/0f34a011be214e5ba3796694dc6e33d1?logfile=console.log | 11:28 |
| tobiash | quiquell: hrm, I don't see something in the react change that could explain that this changed: https://review.openstack.org/#/c/607479/4 | 11:30 |
| tristanC | tobiash: i have no opinions related to css ;) | 11:34 |
| tristanC | Shrews: here, launcher logs of openshift project and pod creation: http://logs.openstack.org/67/570667/22/check/nodepool-functional-openshift/cb06572/launcher/nodepool/launcher.log | 11:34 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: Implement an OpenShift resource provider https://review.openstack.org/570667 | 11:37 |
| tristanC | corvus: would be nice to move on the openshift driver, please let me know what else is needed ^ | 11:37 |
| tristanC | corvus: about adding a zuul/nodepool/zuul-jobs integration test, do you think we could use the new quickstart? | 11:42 |
| mordred | tristanC: I'll see if I can figure it out - but if you look at the status page on mobile, the check box for "expand by default" strangely shifts down to the next line | 11:52 |
| Shrews | tristanC: corvus: neat! I still have concerns that testing with playbooks isn't going to allow for more substantial testing of the driver. I think all that you have is great, except maybe we should replace the small portion where you wait for nodes, show the nodes/pods, with a tox role that runs functional tests | 11:52 |
| mordred | tristanC: I'm jetlagged and my brain doesn't work - but I do think it would be great to have some CI on the quickstart so that we make sure it still works - I think corvus has some thoughts on that | 11:53 |
| tristanC | Shrews: you would mind bootstraping a tox env and test class, i'm not sure to understand how you want to do this... | 11:53 |
| tristanC | mordred: about css, i don't really understand how all this stuff work tbh... | 11:54 |
| tristanC | mordred: corvus: i guess the trick is to be able to build the zuul/nodepool image before running the docker-compose, and make sure the quickstart can use speculative content | 11:55 |
| mordred | tristanC: me either :) - but I'll see what I can figure out | 11:56 |
| mordred | quiquell: I just tried a stream from the web and it worked for me - http://zuul.openstack.org/stream/58381873fcab468896a02ce02b1bf5b3?logfile=console.log ... so I agree, the value in the status.json itself is wrong | 11:57 |
| mordred | quiquell: are you using that value from something? | 11:57 |
| Shrews | tristanC: i just mean having your testing playbook include this role (http://git.openstack.org/cgit/openstack-infra/zuul-jobs/tree/roles/tox/README.rst) which would run a 'tox -ek8s' suite of tests | 11:58 |
| Shrews | wouldn't that give a developer more ways to test the driver? | 11:59 |
| tristanC | Shrews: what would be the k8s content of tox.ini? | 12:00 |
| tristanC | mordred: quiquell: yes, the interface is cheating to accomodate with the legacy links returned from status api, now that it is deployed we can fix that | 12:00 |
| quiquell | mordred: Yep, the greasy monky stuff that show you the progress of gerrit | 12:01 |
| dmsimard | There's some funky rewrite rules too, maybe it's that | 12:01 |
| dmsimard | ¯\_(ツ)_/¯ | 12:01 |
| quiquell | mordred: it's taking the link of running from the json | 12:01 |
| tristanC | the cheat logic is: http://git.zuul-ci.org/cgit/zuul/tree/web/src/containers/status/ChangePanel.jsx#n226 | 12:02 |
| quiquell | tristanC: Thanks so much, do I have to write a bug or something to track progrress ? | 12:02 |
| Shrews | tristanC: something along the lines of: stestr --test-path ./nodepool/tests/k8s run | 12:02 |
| Shrews | tristanC: Consider this only a suggestion. You know k8s much better than I and would likely be the lead on fixing any bugs. If you think ansible playbooks give you the flexibility you need, then maybe my suggestion can be ignored. | 12:04 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix stream url from status json https://review.openstack.org/609365 | 12:07 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: remove legacy stream.html url support https://review.openstack.org/609366 | 12:07 |
| tristanC | mordred: quiquell: ^ those should make status build url consistent | 12:08 |
| quiquell | tristanC: thanks ! | 12:08 |
| quiquell | so fast | 12:08 |
| tristanC | quiquell: it has been quite sometime i refactored that web interface, there may be some more left-over cleanup to do :-) | 12:09 |
| *** jesusaur has quit IRC | 12:10 | |
| quiquell | tristanC: cool | 12:11 |
| *** quiquell is now known as quiquell|lunch | 12:11 | |
| dmsimard | tristanC: great work! | 12:11 |
| tristanC | dmsimard: thanks! | 12:12 |
| dmsimard | I need someone to help me out with the ARA 1.0 interface if you get bored :p | 12:12 |
| tristanC | Shrews: shouldn't we need a new BaseTestCase class to setup zk but not start launcher thread, or should we also drop the "command: nodepool-launcher" and run the service from tox? | 12:13 |
| tristanC | Shrews: could we start doing that with tox -eopenstack ? | 12:15 |
| Shrews | tristanC: there would have to be a new base class, yes | 12:15 |
| Shrews | tristanC: we could do it for the openstack-driver specific tests, but that would be a lot of work to redo the tests | 12:16 |
| Shrews | this is new territory, so i'm not exactly sure what the trade offs are | 12:17 |
| *** jesusaur has joined #zuul | 12:17 | |
| mordred | tristanC, tobiash: first patch looks great- I think we should wait for $time to land the second one, since it would mean someone updating web app without having their scheduler updated would be broken | 12:18 |
| mordred | tristanC, tobiash: so I'm going to drop a -2 on it just so it doesn't accidentally land - but I'm actually +2 on it :) | 12:18 |
| tobiash | mordred: so $time is like 2 days? | 12:18 |
| Shrews | tristanC: but if the new base class is created generically enough, it could be reused for other in-flight drivers. so maybe it could assume that the launcher is already started with the correct driver setup in nodepool.yaml | 12:19 |
| mordred | tobiash: I'm thinking we need to wait until we've cut a new tag | 12:19 |
| tristanC | tobiash: scheduler needs to be restarted though | 12:19 |
| EmilienM | tristanC: good work on http://zuul.openstack.org UI - it's crazy fast, thanks a lot | 12:20 |
| tobiash | mordred: that long? I think with a new tag you should restart all services anyway | 12:20 |
| tobiash | and everyone following master should know what he's doing ;) | 12:20 |
| mordred | so - honestly, landing the second patch isn't as important as the first one - it's just a cleanup -so we could even wait for a few weeks or a month without negative impact - other than forgetting about it :) | 12:20 |
| tristanC | EmilienM: you're welcome, it can still be improved if needed, the change indents is still done globally | 12:21 |
| tobiash | ok, I'm just concerned about the latter ;) | 12:21 |
| mordred | tobiash: indeed - but it's fairly safe to run completely CD for the web app and still be careful about scheduler restarts | 12:21 |
| mordred | tobiash: ++ | 12:21 |
| mordred | tobiash: maybe we should make a topic we use for things like this - like 'merge-after-release' | 12:21 |
| tobiash | mordred: ok, from that perspective fine so let's wait until the next tag | 12:21 |
| mordred | tobiash: and then when we cut tags we can go look at anything we've marked? | 12:21 |
| tobiash | that's a great idea | 12:22 |
| mordred | k. I updated the topic | 12:22 |
| tobiash | maybe add a zuul job in the release pipeline that sends a notification about this ;) | 12:22 |
| Shrews | tristanC: I wouldn't change what you have yet though. Let's get input from corvus first. | 12:24 |
| Shrews | i mean, the openstack functional tests don't do more than what you already have, so maybe that's enough :) | 12:25 |
| tristanC | Shrews: I still think the ultimate test is to run zuul/nodepool/zuul-jobs, and to do that, it seems easier to do go with ansible | 12:26 |
| Shrews | now that i think about it, the tox tests are more about testing the launcher code itself | 12:26 |
| tristanC | Shrews: ftr, here is the sf-ci test for openshift integration: https://softwarefactory-project.io/cgit/software-factory/sf-ci/tree/health-check/zuul-openshift.yaml | 12:27 |
| Shrews | tristanC: that's fine. i defer to your expertise here :) | 12:28 |
| tristanC | imo, we should make that playbook work with the quickstart deployment | 12:28 |
| *** bhavikdbavishi has quit IRC | 12:30 | |
| *** quiquell|lunch is now known as quiquell | 12:30 | |
| tristanC | Shrews: well there is value to use tox/python client to test the code, it maybe easier to do fault-injection | 12:30 |
| tristanC | but i would be more confortable if there is a test that demonstrate a zuul-job can run openshift as well as one that build an image and test it too. | 12:31 |
| tristanC | and now that there will be a gerrit deployment process in quickstart doc, it doesn't seems overly complicated to do that | 12:32 |
| tristanC | but i'll defer to corvus regarding to what is expected/needed to get that driver moving :) | 12:33 |
| Shrews | maybe we need both? maybe create a FakeKubernetesDriver for the tox tests | 12:33 |
| Shrews | and just piggy back on top of the py35 tox env | 12:35 |
| *** rlandy has joined #zuul | 12:39 | |
| *** bhavikdbavishi has joined #zuul | 12:41 | |
| *** bhavikdbavishi has quit IRC | 12:41 | |
| *** bhavikdbavishi1 has joined #zuul | 12:41 | |
| *** bhavikdbavishi1 has quit IRC | 12:46 | |
| mordred | tristanC, Shrews: it also might make sense to make another deployment similar to the quickstart but more tuned for being usable for testing features - for feature testing we might need/want a different set of presets that would make the quickstart more complex | 12:54 |
| mordred | but I'm also just saying random words | 12:54 |
| Shrews | corvus: i didn't give https://review.openstack.org/608344 a second run through, but the new contents lgtm so +2'd. Didn't approve in case someone else wants to give this version a try | 13:07 |
| *** swest has quit IRC | 13:17 | |
| *** quiquell is now known as quiquell|off | 14:14 | |
| *** che-arne has quit IRC | 14:18 | |
| Shrews | https://review.openstack.org/608700 <-- enable py36 testing on nodepool... can haz reevuuu? | 14:21 |
| tristanC | https://review.openstack.org/593605 <-- multi filters on build page can haz review too? | 14:25 |
| *** SotK has quit IRC | 14:27 | |
| mordred | Shrews, tristanC: +2 from me | 14:28 |
| *** clarkb has joined #zuul | 14:42 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Reorganize admin manual https://review.openstack.org/609432 | 14:43 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-website master: Switch the "Get Started" button to quick-start https://review.openstack.org/609434 | 14:46 |
| *** nilashishc has quit IRC | 14:48 | |
| *** nilashishc has joined #zuul | 14:48 | |
| corvus | Shrews, tobiash, mordred, clarkb, tristanC, j^2: https://review.openstack.org/#/q/topic:zuul-quickstart has 3 changes which i think are ready to review/land. | 14:48 |
| corvus | the build output from 609432 should show the admin manual in a coherent state | 14:49 |
| corvus | tristanC, Shrews: do i understand correctly that earlier you were discussing something like using the docker-compose file to set up a system for functional testing that would replace part of what is in 605803 (the nodepool k8s functional test change)? | 14:53 |
| *** chandankumar is now known as chkumar|off | 14:55 | |
| Shrews | corvus: no. i think mordred suggested that as something to ponder | 15:06 |
| clarkb | I think that would work fine for the existing tests fwiw | 15:07 |
| clarkb | though we might want some way of disabling subsets of services (like does nodepool functional testing need a gerrit or even a zuul?) | 15:07 |
| Shrews | i think we just need a consensus on what type of testing is required for new drivers | 15:07 |
| Shrews | is what tristanC has in his ansible playbooks enough? or do we also want deeper, unit testing via tox? or both? | 15:08 |
| Shrews | or the 3rd thing | 15:08 |
| clarkb | Shrews: I think for many drivers we won't be able to do functional testing because it isn't practical to require we have working azure credentials for example | 15:09 |
| clarkb | Shrews: that probably means unit testing via tox is a minimum we can rely on using faked services | 15:09 |
| clarkb | then if we can functionally test it do our best within reason (openstack, k8s, maybe a generic container driver if that happens) | 15:10 |
| *** bhavikdbavishi has joined #zuul | 15:11 | |
| *** panda has quit IRC | 15:13 | |
| *** panda has joined #zuul | 15:14 | |
| clarkb | I do think the functional testing that was added way back when for openstack helped significantly to ensure both the driver and nodepool were workign as expected | 15:14 |
| clarkb | so having as much functional testing as possible is a win imo | 15:15 |
| corvus | Shrews, tristanC: i think the functional testing needs to happen with tox. it's hard to simulate the full lifecycle of a request with just ansible and min-ready. i think we can use the playbooks/roles in 605803 to do the setup, but then we should run 'tox -e functional_k8s' to run tests which submit node requests, and return nodes, so that we can test deletion too. | 15:24 |
| *** SotK has joined #zuul | 15:25 | |
| *** techraving has joined #zuul | 15:28 | |
| Shrews | corvus: that's sort of what i was pushing for | 15:33 |
| Shrews | s/sort of/pretty much exactly/ | 15:33 |
| Shrews | i'll try to play around with the BaseTestCase class to see if I can come up with something generic tristanC can use | 15:34 |
| *** bhavikdbavishi1 has joined #zuul | 15:41 | |
| *** bhavikdbavishi1 has quit IRC | 15:41 | |
| *** bhavikdbavishi has quit IRC | 15:42 | |
| *** themroc has quit IRC | 15:48 | |
| *** bhavikdbavishi has joined #zuul | 15:50 | |
| *** sshnaidm has quit IRC | 15:58 | |
| *** sshnaidm has joined #zuul | 15:59 | |
| *** caphrim007_ has joined #zuul | 16:00 | |
| *** techraving has quit IRC | 16:25 | |
| clarkb | corvus: or maybe even tox -e functional and depending on the setup you get k8s or openstack or $otherthing | 16:34 |
| clarkb | but maybe you need enough job specific setup that there needs to be different envs | 16:34 |
| corvus | clarkb: yeah, i'm thinking there may be enough variations within a driver that just one label might not be enough for testing. in other words, the functional tests for each driver may need a unique set of labels and things to test. especially if nodepool grows more complexity. | 16:36 |
| *** bhavikdbavishi has quit IRC | 16:36 | |
| clarkb | corvus: brainstorming topic: talked to tripleo folks a bit this morning about how to communicate and work together with openstack infra better. One of the topics that came up was Zuul isn't always very clear on why it took actions (or didn't take action). We might want to prioritize stuff like user feedback (cycle detection reporting for example) in the near future? | 16:37 |
| clarkb | I think that fits into the get bootstrapped and running work you are doing as it is in the name of user experience | 16:37 |
| corvus | clarkb: any examples other than cycle reporting? | 16:39 |
| clarkb | corvus: when the gate resets we've seen a lot of confusion around why jobs are restarting that have already passed (it is because zuul has a new future state and the old one is invalid) | 16:41 |
| clarkb | corvus: I'm not sure how to communicate that through the dashboard but that was another one | 16:41 |
| clarkb | another is when changes are inactive due to being outside of the window | 16:41 |
| corvus | clarkb: hrm, those last two sound less like a zuul ui problem and more like some people don't understand what zuul is doing. i'm not sure there's a tweak we can make to the status page that can actually explain the theory behind gating. perhaps we can suggest that users who wish to understand what zuul is doing should read https://zuul-ci.org/docs/zuul/user/gating.html ? | 16:45 |
| clarkb | corvus: ya one idea was to make links to how gating works more conspicuous in the UI | 16:46 |
| clarkb | not necessarily solve that problem directly, but have appropriate breadcrumbs to the docs | 16:46 |
| corvus | that's a good idea... i don't think there are any links to the docs in the web ui now :) | 16:47 |
| chkumar|off | clarkb: corvus Hello | 16:47 |
| corvus | oh, er, i stand corrected, there is now that the new dashboard merged :) | 16:47 |
| clarkb | corvus: oh neat | 16:47 |
| chkumar|off | clarkb: corvus I was in PyCon India 2018 last week talking about zuul | 16:48 |
| chkumar|off | they basically like zuul depends on feature when I was talking with people | 16:48 |
| corvus | chkumar|off: thanks! that's great to hear :) | 16:49 |
| chkumar|off | but on doc side, they find it is very hard to understand like how it is different from jenkins | 16:49 |
| chkumar|off | like a feature comparison page with various CI tools would be great to have | 16:49 |
| corvus | chkumar|off: yeah, people at ansiblefest were asking similar things | 16:50 |
| corvus | considering that there are no other project gating systems, it can be tricky to do a comparison. :) | 16:50 |
| chkumar|off | corvus: it would be great if we get a something quickstart like stuff which can be easily installable and ready to use for demo purpose | 16:51 |
| corvus | chkumar|off: we should have one within a few hours i think: https://review.openstack.org/608344 | 16:51 |
| chkumar|off | just running a single shell script or one or two command everything will be ready | 16:51 |
| corvus | chkumar|off: yep, that's what that does -- it does it with docker, and you can have a full system with 'docker-compose up' | 16:52 |
| chkumar|off | corvus: I will try and let you know :-) | 16:52 |
| corvus | chkumar|off: you can see the doc preview rendered here: http://logs.openstack.org/44/608344/13/check/tox-docs/931664e/html/admin/tutorial.html | 16:53 |
| corvus | clarkb: did you want to look at that, or should we go ahead and +W it? | 16:53 |
| clarkb | corvus: I'm ok with yall +W'ing it. I still plan to run it and see how it goes for me and can provide followup changes or feedback if necessary | 16:53 |
| chkumar|off | corvus: can we change this tutorial with github as people are not much fan of gerrit? | 16:54 |
| clarkb | I'm dealing with post conference administrivia for multiple conferences right now so want to get that done (so I can get paid for expenses :) ) | 16:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Reorganize admin manual https://review.openstack.org/609432 | 16:54 |
| corvus | chkumar|off: no -- there is no way to install a copy of github in a container since it's not free software. | 16:54 |
| chkumar|off | corvus: hehe, may be we need some one to write integration for gitlab then which will deploy gitlab ce and do the integration :-) | 16:55 |
| corvus | chkumar|off: that would be an option. someone at ansiblefest volunteered to write a gitlab driver; hopefully they will, and once that's done we could do the same with gitlab. | 16:56 |
| chkumar|off | corvus: great :-) | 16:56 |
| corvus | chkumar|off: but as the docs say -- the idea is to teach folks about zuul -- the concepts apply equally to github | 16:56 |
| corvus | chkumar|off: and if we want, we can extend the tutorial for how to connect to github (you can still use the same docker-compose file and modify zuul.conf -- the config files are bind-mounted into the containers) | 16:57 |
| corvus | chkumar|off: the main thing is that i wanted something completely self-contained where someone can get started quickly. | 16:59 |
| clarkb | corvus: looking at the new ui maybe link to gating docs in the gate pipeline description? I think that is actually configuration right? I'll push that change up now | 16:59 |
| corvus | clarkb: yeah... i'm not sure if we can do html in that though | 16:59 |
| chkumar|off | corvus: I will try the tutorail tomorrow and have the feedback :-) | 16:59 |
| corvus | chkumar|off: great, thanks! we might merge the change before then, so just let us know in here if you have comments after it merges. | 17:00 |
| clarkb | corvus: https://review.openstack.org/609497 something like that maybe | 17:03 |
| corvus | clarkb: yeah, i'm not sure what we need to do to make it a hyperlink though | 17:10 |
| clarkb | ya that will had to be edited int he context of the surrounding html? maybe an href would just work (tm) though? | 17:11 |
| corvus | i'd try it in my sandbox, but i don't think the web stuff is working in the container images yet | 17:11 |
| clarkb | I was going to say maybe that is more motivation to run the sandbox sooner than later :P | 17:12 |
| corvus | let me double check that | 17:12 |
| corvus | yeah, localhost:9000 is still 404 on / | 17:15 |
| corvus | tristanC, Shrews, mordred: ^ if one of you has a chance to look at the current status of javascript in the container images, that would be great :) | 17:15 |
| clarkb | did shrews change for the hook merge? | 17:15 |
| SpamapS | off topic: I'm in search of laptop recommendations. Any suggestions are appreciated. | 17:16 |
| clarkb | SpamapS: I like the modern lenovo x200 series. The latest one isn't super happy with linux though so might have to go to previous gen (x270 iirc). There are also the x210 and x330 lcdfans fabricobbled older but new thinkpads | 17:17 |
| corvus | cause right now, the container tutorial curiously omits any mention that there is a web interface :( | 17:18 |
| SpamapS | clarkb: no offense, but that.. just added confusion. thanks? ;) | 17:25 |
| clarkb | SpamapS: there is at least one person that takes old thinkpads and puts modern hardware in them so that you can have old school keyboard with new cpu and stuff | 17:28 |
| clarkb | otherwise if you find the keyboard acceptable on modern thinkpads I would get an x270 myself | 17:28 |
| pabelanger | SpamapS: so far X1 carbon has been nice, made the switch 6 weeks ago | 17:30 |
| pabelanger | using fedora-28, only finger scanner seems to no work | 17:31 |
| SpamapS | clarkb: ahh | 17:31 |
| SpamapS | pabelanger: don't want thin or big. I like the form factor I had with my x260... I'd prefer the keyboard of the x220 tho. ;) | 17:31 |
| pabelanger | ack | 17:32 |
| openstackgerrit | Merged openstack-infra/nodepool master: Add tox-py36 job https://review.openstack.org/608700 | 17:38 |
| SpamapS | pabelanger: looking closer at the X1 now.. 3 recommendations so far. ;) | 17:55 |
| *** pcaruana has quit IRC | 17:55 | |
| pabelanger | nice | 17:55 |
| dmsimard | SpamapS: I have a t480s which has basically the same specs as the x1 but it has pci-e nvme instead of ssd and has like one more usb port or something :p | 17:56 |
| dmsimard | If you can, get the high res display option | 17:56 |
| Shrews | I chose the X1 for my work laptop replacement, but I haven't received it yet | 17:58 |
| clarkb | my issue with the x1 is battery life. the x200 series has much better battery life (in part because the screen has fewer pixels) | 18:01 |
| *** techraving has joined #zuul | 18:02 | |
| SpamapS | Yeah the x260 that I had (or was it x250?) had the external bump battery and lasted like 9 - 10 hours for non-intensive stuff | 18:10 |
| *** pcaruana has joined #zuul | 18:12 | |
| *** panda has quit IRC | 18:20 | |
| *** panda has joined #zuul | 18:23 | |
| *** techraving has quit IRC | 18:27 | |
| *** electrofelix has quit IRC | 18:28 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add a quick-start test job https://review.openstack.org/609514 | 18:32 |
| corvus | Shrews, mordred, tristanC, clarkb: ^ that job is a bit of a milestone. :) | 18:33 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Add tox functional testing for drivers https://review.openstack.org/609515 | 18:34 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add a quick-start test job https://review.openstack.org/609514 | 18:34 |
| Shrews | I have no idea if something like 609515 will work, btw. Just seeing what happens when it runs | 18:35 |
| corvus | now we need to figure out how to build and use docker images from the change under test | 18:35 |
| corvus | Shrews: looks plausible :) | 18:35 |
| Shrews | i don't think much of our base test classes (if anything) needs to change | 18:36 |
| Shrews | maybe just add some new waitForTHING() calls | 18:36 |
| Shrews | corvus: oh, that's neat | 18:38 |
| Shrews | corvus: next we'll have zuul coding it's own bug fixes | 18:38 |
| corvus | Shrews: i really don't want the quick-start to break :) | 18:39 |
| corvus | the re-organized admin guide is at http://logs.openstack.org/32/609432/2/check/tox-docs/c8dd910/html/admin/index.html | 18:39 |
| Shrews | tricky part will be keeping the doc and playbook in sync | 18:39 |
| corvus | Shrews: yep. i didn't think auto-extracting commands would work. however, since most of the content is inlined files, just copying the files and committing them gets us a lot of functional testing. | 18:40 |
| Shrews | we could drop the nodepool-zuul-functional test (or replace it) with that | 18:43 |
| tobiash | tristanC, mordred: with the new web ui I get xss problems with our authentication provider (saml based) which works with redirects | 18:56 |
| *** caphrim007_ has quit IRC | 18:57 | |
| clarkb | tobiash: this is authentication to the zuul ui loading external resources? | 18:57 |
| tobiash | when the index.html is in the cache | 18:57 |
| tobiash | yes | 18:57 |
| clarkb | in that case you probably need to set CORS rules that allow your external resources. I doubt we make that configurable in zuul today though | 18:57 |
| tobiash | clarkb: our zuul api routes set the cors header | 18:58 |
| tobiash | but maybe not the static stuff | 18:58 |
| clarkb | tobiash: for openstack I see access control allow origin * for the cherrypy handled requests | 19:00 |
| clarkb | tobiash: but not for the js and css files. I wonder if that is the problem | 19:00 |
| tobiash | it fails when downloading the service-worker.js | 19:02 |
| tobiash | I'll try to exclude that from authentication | 19:02 |
| *** pcaruana has quit IRC | 19:03 | |
| clarkb | tobiash: or set the CORS headers in your proxy too? | 19:05 |
| tobiash | I tried that, but that seems to fail within a redirect chain in the auth provider :/ | 19:05 |
| *** ssbarnea has quit IRC | 19:29 | |
| *** pcaruana has joined #zuul | 19:55 | |
| *** pcaruana has quit IRC | 20:06 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Add tox functional testing for drivers https://review.openstack.org/609515 | 20:06 |
| corvus | tobiash: maybe you can add a cors header here: http://git.zuul-ci.org/cgit/zuul/tree/zuul/web/__init__.py#n455 ? | 20:17 |
| tobiash | corvus: that doesn't work (tried to inject cors header to everything in the proxy) | 20:18 |
| tobiash | what gets blocked is a redirect of a redirect within the other domain (which doesn't send cors headers) | 20:18 |
| tobiash | what I'm trying now is to enforce that index.html is never loaded from disk as the redirect is triggered on the first network call | 20:19 |
| tobiash | but cache headers don't work because the request for index.html is serviced by the service worker (didn't know yet that something like that exists) | 20:20 |
| tobiash | so my next try is to short-circuit the registration of the service worker: https://git.zuul-ci.org/cgit/zuul/tree/web/src/registerServiceWorker.js#n33 | 20:21 |
| tobiash | that will hopefully force index.html to be the first network call which gets redirected (and should work because it's no xss then) | 20:22 |
| corvus | Shrews: would you mind re-adding your +2 to https://review.openstack.org/609432 ? (i missed a git add earlier) | 20:27 |
| clarkb | corvus: I'm working through the docker compose doc and there is a missing step to add yourself to the docker group or run docker compose with sudo | 20:30 |
| * clarkb uses sudo because throwaway VM | 20:31 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add a quick-start test job https://review.openstack.org/609514 | 20:33 |
| pabelanger | clarkb: there is a pretty large warning about doing that in documentation: https://docs.docker.com/install/linux/linux-postinstall/#manage-docker-as-a-non-root-user we should add the same warning, if we do that | 20:34 |
| clarkb | corvus: my only other comment so far is making a new image just to run ansible seems like overkill? but maybe I like bash more than others | 20:34 |
| clarkb | pabelanger: I'd be fine with just adding sudo to the command in the doc too | 20:34 |
| clarkb | pabelanger: it works either way | 20:34 |
| corvus | clarkb: yeah, we should add a step like that. i assume the quick-start job i'm writing is about to run into that. | 20:35 |
| clarkb | pabelanger: just pointing out we don't document that step (either way) | 20:35 |
| corvus | clarkb: re ansible, you mean for gerritconfig? | 20:35 |
| clarkb | corvus: ya the setup playbook appears to be stuff that simple shell script could do then you don't need another image | 20:35 |
| clarkb | ssh keygen and file copies and such | 20:35 |
| pabelanger | clarkb: ack | 20:35 |
| corvus | clarkb: the uri module in ansible is *life changing*. take a closer look at the setup playbook. i think writing that kind of thing in ansible is amazing. | 20:36 |
| tobiash | corvus, tristanC: disabling the service worker fixed the issue for me ;) | 20:37 |
| corvus | clarkb: i'm especially a fan of the wait_for pattern, but also the bit at the bottom with the user creation rest api. the yaml->json stuff there is great. | 20:39 |
| clarkb | ya some bits would be more verbose in shell for sure. Just wondering if building an image just for that is worth it? I guess if it makes it easier to understand what is happening then probalby (intended to be instructional afterall) | 20:40 |
| corvus | clarkb: anyway, the world actually needs an ansible container image. but, lacking an upstream image, we should at least be re-using the centos image that gerrit is based off of, so the build shouldn't take long... | 20:40 |
| corvus | clarkb: if it is taking too long, maybe we should look into that | 20:41 |
| clarkb | in theory its only long the first time too | 20:41 |
| clarkb | unless your VM had a tiny disk to start and you had to start over then its the first two times >_> | 20:42 |
| corvus | yep. oh, i guess it's possible the ansible image is being created first, in which case the time would be allocated to it rather than gerrit... | 20:42 |
| corvus | and gerrit is more specific in version numbers, so if that happens we might end up with two different centos images | 20:42 |
| corvus | maybe we should see if we can use the ubuntu image instead | 20:43 |
| corvus | (which would be shared with the 'node' image) | 20:43 |
| pabelanger | if we are building on alphine for zuul, couldn't we also put ansible on that version and share the layer? | 20:44 |
| corvus | yes alpine would work too | 20:44 |
| pabelanger | I think the ubuntu idea too | 20:44 |
| pabelanger | err | 20:44 |
| pabelanger | I think the ubuntu idea is great too | 20:44 |
| corvus | i'll try alpine real quick | 20:46 |
| clarkb | oh and something I think mordred pointed out is each RUN entry gives you a new layer (which may or may not be desireable) I think figuring that stuff out now is probably the wrong order of operations though. Get it working reliably an dthen worry about too many layers later | 20:46 |
| pabelanger | I think our 2 step build process helps cut down on a lot of waste also | 20:48 |
| corvus | yes i tried to use && liberally, but can fix cases where i missed that in a followup | 20:48 |
| clarkb | for the zuul containers themselves yes | 20:48 |
| corvus | oh i know why we picked centos -- it already has all the python stuff installed (because of rpm/yum) | 20:52 |
| clarkb | ah right | 20:52 |
| corvus | (still working on alpine, just a bit more trial/error to get all the packages) | 20:53 |
| clarkb | corvus: gerritconfig_1 | fatal: [localhost]: FAILED! => {"changed": false, "cmd": "/usr/sbin/udevadm info --query property --name /dev/vda1", "msg": "Timer expired after 10 seconds", "rc": 257} | 20:55 |
| clarkb | this happened when gathering facts | 20:55 |
| clarkb | this prevented setup.yaml from running | 20:55 |
| clarkb | I think we may not need to gather facts in that playbook. But I wonder if that is a side effect of running ansible in a container? | 20:55 |
| corvus | weird... i have not run into that | 20:56 |
| clarkb | I'm running on ubuntu xenial because I had a VM available to run on that platform | 20:56 |
| corvus | so am i, because i haven't upgraded yet | 20:56 |
| corvus | clarkb: can you 'docker exec' into that container and try that out? | 20:57 |
| clarkb | hrm which says its /sbin/udevadm not /usr/sbin/udevadm | 20:58 |
| clarkb | oh irght its in a container derp | 20:58 |
| corvus | (i destroyed all my containers for the alpine thing so it's not easy for me right now) | 20:58 |
| clarkb | I stopped docker-compose but I should be able to rerun it? /me tries | 20:58 |
| corvus | yeah, the whole process should be idempotent (at least if it works), so you can keep stopping/starting the same containers | 20:59 |
| clarkb | looks like it happened again on another compose up so at least it is consistnet. Going to test directly now | 21:00 |
| corvus | oh, since it's gerritconfig, it's going to exit, so you may need to restart it with a command or something | 21:00 |
| clarkb | ya I'm going to use docker run to run commands on the same image | 21:01 |
| clarkb | ok there is no such device node when run directly. I'm guessing because I am not attaching the docker volumes /me will start the existing containerwith a different command | 21:03 |
| clarkb | ok I think that started it without volumes attached and then it worked? | 21:08 |
| corvus | clarkb: i changed the command in the compose file to "/bin/bash -c 'sleep 99999999999'" and then did 'docker exec /bin/bash' and i see the volumes but i don't see any devices like that | 21:10 |
| clarkb | corvus: ya my volumes have content in them implying they were present when the setup.yaml actually ran | 21:11 |
| clarkb | but I too saw no such volumes when manually execing into it | 21:11 |
| clarkb | the host node has a /dev/vda1 | 21:12 |
| clarkb | I wonder if the container is leaking somehow with ansible fact gathering | 21:12 |
| corvus | that's a really reassuring thought | 21:12 |
| clarkb | corvus: running `sudo docker run examples_gerritconfig ansible localhost -m setup` shows a vda device (but returns relatively quickly) that does not show up if you change the command to ls -l /dev | 21:16 |
| corvus | i'll try that | 21:16 |
| corvus | clarkb: i see my host system devices there | 21:17 |
| clarkb | corvus: ya, this is kind of trippy | 21:18 |
| clarkb | kernel says "nothing to see here" if you ls /dev, but ansible fact gathering finds it anyway | 21:18 |
| *** ssbarnea has joined #zuul | 21:18 | |
| clarkb | if we don't need facts maybe we start by disabling fact gathering? | 21:18 |
| clarkb | I'm going to delete things and try with playbook updated to not gather facts | 21:19 |
| corvus | clarkb: yeah, want to add 'gather_facts: no' to all the playbooks? | 21:19 |
| clarkb | corvus: ya I'll do that and test it in a moment | 21:19 |
| corvus | clarkb: you should probably add that to all the playbooks in examples/playbooks/*.yaml | 21:19 |
| corvus | scheduler and executor run similar playbooks before starting services to aid in sequencing | 21:20 |
| clarkb | yup I see them too | 21:20 |
| clarkb | examples_gerritconfig_1 exited with code 0 | 21:24 |
| corvus | clarkb: i'll incorporate that into this alpine change | 21:24 |
| clarkb | corvus: ok | 21:24 |
| clarkb | definitely seems happier with that chnage made | 21:24 |
| clarkb | I'll leave review notes in gerrit too just to make sure we don't miss anything | 21:26 |
| clarkb | but sounds like you are on top of it | 21:26 |
| corvus | clarkb: yeah, i'm doing the alpine image change, gather_facts, and some general tidying of the playbooks | 21:30 |
| clarkb | corvus: don't forget sudo to run docker-compose too | 21:30 |
| corvus | clarkb: ah yeah. so just 'sudo docker-compose up' is the only thing we need to change there? | 21:32 |
| clarkb | yup | 21:32 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Fixes for quick-start playbooks https://review.openstack.org/609551 | 21:36 |
| corvus | clarkb: omnibus fixes ^ | 21:36 |
| corvus | clarkb: and yes, i think my zuul job just ran into the sudo issue :) | 21:37 |
| corvus | 2018-10-10 20:57:52.850586 | ubuntu-xenial | Couldn't connect to Docker daemon. You might need to install Docker: | 21:37 |
| clarkb | yup thats the error | 21:38 |
| clarkb | I'll pull down the fixes change and rerun it shortly | 21:39 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add a quick-start test job https://review.openstack.org/609514 | 21:40 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Fixes for quick-start playbooks https://review.openstack.org/609551 | 21:40 |
| clarkb | hrm does docker compose not notice when a dockerfile has changed? I guess not | 21:42 |
| clarkb | I cleaned up everything but the images themselves | 21:42 |
| *** ssbarnea has quit IRC | 21:45 | |
| corvus | clarkb: no, i think 'docker-compose build $imagename' is needed if you change an image | 21:50 |
| clarkb | ya I went and ran rmi on the images that were built recently and am rerunning | 21:50 |
| corvus | that works too | 21:51 |
| clarkb | also volume prune so that the full setup.yaml playbook runs | 21:51 |
| corvus | clarkb: any idea why docker-compose up still failed? http://logs.openstack.org/51/609551/2/check/zuul-quick-start/61b39f8/ara-report/ | 21:51 |
| corvus | clarkb: oh i guess i need to start the docker service? | 21:52 |
| corvus | oh wait i don't see docker itself being installed here: http://logs.openstack.org/51/609551/2/check/zuul-quick-start/61b39f8/ara-report/result/21e7ffab-7579-4a10-a7e9-d757e3c1bf7c/ | 21:52 |
| clarkb | corvus: on debuntu it should start automatically, at least it did for me using the package docker.io on ubuntu | 21:53 |
| clarkb | the ombibus of fixes seems to work for me | 21:53 |
| corvus | clarkb: did you already have it installed? | 21:53 |
| corvus | it looks like 'sudo apt-get install docker-compose' isn't enough | 21:53 |
| clarkb | corvus: that is what I did, it pulled in docker.io. Oh you know what I rebooted after package installs becaues I also did a dist-upgrade | 21:54 |
| clarkb | could be set to start on boot but not after package install | 21:54 |
| corvus | clarkb: but if you look in that last link, it just says docker.io is recommended, not required | 21:55 |
| clarkb | corvus: oh, by default ubuntu installs recommends but I think our test node images may disable that? | 21:55 |
| clarkb | that may also explain it :) | 21:55 |
| corvus | ooooh | 21:55 |
| corvus | okay, so i guess it's okay to omit that from the text instructions and just add it to the playbook | 21:55 |
| clarkb | yes, though shouldn't hurt to also add it to the text instructions | 21:56 |
| corvus | how do our images disable that? (i'm wondering if we can/should re-enable that considering what we're trying to test here) | 21:58 |
| clarkb | its an /etc/apt/$file config option iirc /me looks | 21:58 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add a quick-start test job https://review.openstack.org/609514 | 21:58 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Fixes for quick-start playbooks https://review.openstack.org/609551 | 21:58 |
| clarkb | corvus: /etc/apt/apt.conf.d/95disable-recommends is a file created by the dpkg element in dib that does it | 22:00 |
| corvus | clarkb: so just an rm of that should do it? | 22:01 |
| clarkb | sets APT::Install-Recommends "0"; and Apt::Install-Suggests "0"; my hunch is if you delete that file early in the test it will do what you want | 22:01 |
| clarkb | corvus: yup may need an apt-get update too? | 22:01 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Remove disable-recommends in quick-start test https://review.openstack.org/609553 | 22:04 |
| corvus | clarkb: ^ that's a try without the update | 22:04 |
| corvus | docker-compose is running in http://zuul.openstack.org/stream/48fde8a0be884622ab46cb0c8fde149c?logfile=console.log | 22:04 |
| corvus | i'm a dolt. the zuul image has ansible installed, we could just use it. | 22:05 |
| corvus | (well, that will be true at least until we get to the point where zuul is installing ansible internally) | 22:06 |
| corvus | installing ansible takes a long time. maybe it's worth it, and we'll deal with the fallout from that later (maybe by publishing our own ansible images or something) | 22:07 |
| fungi | installing via pip? | 22:07 |
| clarkb | fungi: ya it compiles things like paramiko bindings to openssl | 22:07 |
| fungi | i see. looks like they only publish an sdist to pypi, no wheels | 22:08 |
| clarkb | ya iirc mordred was going to try and track people down to make ansible wheels at ansiblefest | 22:08 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Quick-start: remove ansible Dockerfile https://review.openstack.org/609555 | 22:09 |
| corvus | hopefully we're not using any post-2.5 things | 22:09 |
| clarkb | corvus: I don't think so at least I didn't see anything that stood out to me | 22:10 |
| clarkb | also the centos version of ansible is probably older ? | 22:10 |
| corvus | clarkb: we pip installed on the centos image | 22:10 |
| clarkb | ah | 22:10 |
| Shrews | FYI, they are expecting lots of power outages in my area tomorrow due to yet *another* hurricane headed this way. If I don't respond, that would be the reason why | 22:15 |
| corvus | Shrews: stay safe! | 22:19 |
| clarkb | and dry! | 22:23 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add a quick-start test job https://review.openstack.org/609514 | 22:26 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Fixes for quick-start playbooks https://review.openstack.org/609551 | 22:26 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Remove disable-recommends in quick-start test https://review.openstack.org/609553 | 22:26 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Quick-start: remove ansible Dockerfile https://review.openstack.org/609555 | 22:26 |
| corvus | clarkb: can you review https://review.openstack.org/609432 ? | 22:28 |
| corvus | it looks like we're ready to land the first cut of the tutorial, but if we can land that with it it'd be great | 22:28 |
| clarkb | ya I'll look | 22:29 |
| clarkb | corvus: we may also want to point https://zuul-ci.org/start at the new quick start stuff | 22:30 |
| *** rlandy has quit IRC | 22:33 | |
| *** sshnaidm is now known as sshnaidm|afk | 22:33 | |
| clarkb | corvus: one comment on 609432, missed console -> shell conversion. I +2;d it bceause overall pretty minor but will let you decide if you want to approve at that case | 22:35 |
| clarkb | s/at/in/ | 22:36 |
| corvus | yeah, i'll get that in a followup, thanks. | 22:38 |
| corvus | clarkb: re /start -- https://review.openstack.org/609434 | 22:38 |
| clarkb | +2 | 22:39 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Add tox functional testing for drivers https://review.openstack.org/609515 | 22:41 |
| openstackgerrit | Merged openstack-infra/zuul master: Add docker-compose quickstart example https://review.openstack.org/608344 | 22:59 |
| openstackgerrit | Merged openstack-infra/zuul master: Reorganize admin manual https://review.openstack.org/609432 | 23:06 |
| clarkb | corvus: I approved the /start url because ^ wnt in | 23:06 |
| clarkb | I don't think that should cause any major issues since post has higher priority than gate now | 23:07 |
| clarkb | (so doc publishing should happen first) | 23:07 |
| openstackgerrit | Merged openstack-infra/zuul-website master: Switch the "Get Started" button to quick-start https://review.openstack.org/609434 | 23:09 |
| clarkb | corvus: left some comments on the test addition change. I think I figured out hwy git review is failing | 23:26 |
| Shrews | q/c 2 | 23:42 |
| Shrews | doh | 23:42 |
| Shrews | corvus: tristanC: ooh neat. the tox func test "works": http://logs.openstack.org/15/609515/3/check/nodepool-functional-k8s/3720d29/job-output.txt.gz#_2018-10-10_22_54_03_612777 | 23:44 |
| Shrews | by "works", i mean, the test that tests nothing other than stuff starts seems to not fall over dead | 23:45 |
| clarkb | Shrews: seems like a good start | 23:47 |
| Shrews | clarkb: i think so. we could conceivably replace the tools/check_devstack_plugin.sh script with this | 23:54 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!