| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: executor: properly format error exception https://review.openstack.org/630928 | 03:52 |
|---|---|---|
| *** bhavikdbavishi has joined #zuul | 04:15 | |
| *** bhavikdbavishi has quit IRC | 04:20 | |
| *** bhavikdbavishi has joined #zuul | 04:24 | |
| *** bhavikdbavishi has quit IRC | 04:59 | |
| *** bramwelt has quit IRC | 05:19 | |
| *** bhavikdbavishi has joined #zuul | 05:23 | |
| *** bhavikdbavishi has quit IRC | 05:28 | |
| *** bhavikdbavishi has joined #zuul | 05:35 | |
| *** spsurya has joined #zuul | 05:41 | |
| *** bhavikdbavishi has quit IRC | 06:20 | |
| *** mrhillsman has quit IRC | 06:42 | |
| *** corvus has quit IRC | 06:42 | |
| *** gundalow has quit IRC | 06:42 | |
| *** tacco has quit IRC | 06:42 | |
| *** Shrews has quit IRC | 06:42 | |
| *** Shrews has joined #zuul | 06:42 | |
| *** gundalow has joined #zuul | 06:42 | |
| *** mrhillsman has joined #zuul | 06:42 | |
| *** quiquell|off is now known as quiquell | 06:44 | |
| *** saneax has joined #zuul | 06:55 | |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: WIP: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 07:20 |
| *** themroc has joined #zuul | 07:23 | |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 07:32 |
| *** AJaeger has quit IRC | 07:33 | |
| quiquell | clarkb, tobiash: Is this ok now ? https://review.openstack.org/#/c/623294/ | 07:34 |
| quiquell | fungi, mordred: ^ | 07:34 |
| *** AJaeger has joined #zuul | 07:36 | |
| quiquell | AJaeger, tristanC: https://review.openstack.org/#/c/623294/ zuul-jobs multi-node-bridge role with public ips | 07:38 |
| AJaeger | quiquell: how was it tested? | 07:39 |
| AJaeger | quiquell: best answer on review... | 07:42 |
| quiquell | AJaeger: Ci here https://review.rdoproject.org/r/#/c/18121 uses static driver to connect to libvirt started up nodes | 07:47 |
| quiquell | https://review.rdoproject.org/r/#/c/18121 | 07:47 |
| quiquell | The libvirt nodes will paste it in the review | 07:48 |
| quiquell | libvirt jobs I mean | 07:48 |
| quiquell | AJaeger: It's running now https://softwarefactory-project.io/zuul/t/rdoproject.org/stream/e21cb9f29e4e419297beaacc9a9a22d0?logfile=console.log | 07:49 |
| quiquell | AJaeger: This is enough or do We have to do more testing ? | 07:49 |
| quiquell | Humm well, we really need the job that runs inside this | 07:50 |
| quiquell | This jobs are only installing zuul and starting up but not running the jobs inside that | 07:50 |
| quiquell | I have another preview for this | 07:50 |
| quiquell | s/preview/review/g | 07:50 |
| *** avass has joined #zuul | 07:56 | |
| AJaeger | quiquell: did you see http://logs.openstack.org/94/623294/9/check/openstack-infra-multinode-integration-centos-7/265896c/job-output.txt.gz#_2019-01-21_07_38_36_272242 - missing ")" | 08:07 |
| AJaeger | So, I really like to see testing of the version that you submitted to not break *all* jobs suddenly... | 08:08 |
| quiquell | AJaeger: ack, will look, I don't break to break stuff either | 08:24 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 08:26 |
| quiquell | AJaeger: damn very lame from my part... :-/ | 08:26 |
| quiquell | AJaeger: I see there is no linting at the checks | 08:26 |
| quiquell | Ahh wrong is there | 08:27 |
| quiquell | It's passing :-/ | 08:27 |
| *** gtema has joined #zuul | 08:32 | |
| *** electrofelix has joined #zuul | 08:43 | |
| *** saneax is now known as saneax|mtg | 08:59 | |
| *** jpena|off is now known as jpena | 09:03 | |
| *** SotK_ has joined #zuul | 09:10 | |
| *** hashar has joined #zuul | 09:16 | |
| *** panda|off is now known as panda | 09:29 | |
| *** themroc has quit IRC | 09:55 | |
| *** themroc has joined #zuul | 09:57 | |
| quiquell | AJaeger: this could be realted ? http://logs.openstack.org/94/623294/10/check/openstack-infra-multinode-integration-centos-7/5c616d9/job-output.txt.gz#_2019-01-21_08_32_12_073541 | 10:06 |
| *** fbo has joined #zuul | 10:07 | |
| quiquell | AJaeger: Ahh yes na forget about it | 10:07 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 10:16 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 10:16 |
| *** openstackgerrit has quit IRC | 10:21 | |
| *** openstackgerrit has joined #zuul | 10:42 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: WIP: runner: add support for depends-on https://review.openstack.org/632064 | 10:42 |
| tristanC | jhesketh: with 632064, prep-workspace and local job exec can use depends on given on the command line | 10:43 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 11:24 |
| *** avass has quit IRC | 11:24 | |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 11:34 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 11:48 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 11:50 |
| *** panda is now known as panda|lunch | 12:14 | |
| *** bhavikdbavishi has joined #zuul | 12:21 | |
| quiquell | AJaeger: Fixes centos-7 https://review.openstack.org/#/c/623294/ | 12:23 |
| *** hashar has quit IRC | 12:23 | |
| quiquell | AJaeger: I supose the ubuntu are related ? | 12:23 |
| *** hjensas has joined #zuul | 12:27 | |
| hjensas | I am trying a patch that depends-on two changes. https://review.openstack.org/576873 - It looks like only the first Depends-On is included in the CI run. Is the syntax not correct? Or is it not possible to depend on multiple changes? | 12:29 |
| tobiash | hjensas: the syntax is correct and the status ui correctly shows both dependencies | 12:39 |
| hjensas | tobiash: hm, I only see the python-ironicclient checked out in logs. http://logs.openstack.org/73/576873/9/check/networking-baremetal-multitenant-vlans-python2/6fd4b74/job-output.txt.gz#_2019-01-20_02_02_27_921057 | 12:40 |
| hjensas | tobiash: Wonder if it can be that the projects don't share a change queue. | 12:41 |
| fungi | hjensas: all three changes are showing in the zuul items list at http://logs.openstack.org/73/576873/9/check/networking-baremetal-multitenant-vlans-python2/6fd4b74/zuul-info/inventory.yaml | 12:42 |
| fungi | so all three should be checked out at those commits in the src tree | 12:43 |
| quiquell | tobiash: do you know if we can use ssh keys with password with gerrit connection ? | 12:43 |
| quiquell | fungi: ^ Â? | 12:43 |
| tobiash | quiquell: maybe if they are pre-loaded in the ssh agent | 12:44 |
| tobiash | quiquell: (if paramiko makes use of it by default) | 12:44 |
| fungi | agreed, i am not aware of any zuul configuration option to set an ssh connection password, but if there was a way to not prompt for it | 12:44 |
| quiquell | tobiash: ack, will try | 12:44 |
| tobiash | if ssh-agent doesn't work, then no | 12:45 |
| *** bhavikdbavishi has quit IRC | 12:46 | |
| hjensas | tobiash: fungi: Thanks, it does look like Zuul is doing what I want indeed. | 12:47 |
| *** badboy has joined #zuul | 12:53 | |
| badboy | hi guys | 12:53 |
| badboy | Zuul related question: | 12:53 |
| badboy | is it possible to to access the variable containing the changeid or gerrit url in the trigger playbook? | 12:54 |
| *** dkehn has quit IRC | 12:55 | |
| AJaeger | The ansible variable zuul.change is example, see https://zuul-ci.org/docs/zuul/user/jobs.html#zuul-variables | 12:57 |
| AJaeger | badboy: zuul.change is bad - but see link above for all variables ^ | 12:58 |
| *** hashar has joined #zuul | 13:00 | |
| badboy | AJaeger: thank you | 13:09 |
| quiquell | AJaeger: Do you know how to do kind of "tail" at command line from something like finger://5e24c349d9a5/bc718fe4778145e1bd8e356a83ff3574 at zuul job execution ? | 13:11 |
| *** bhavikdbavishi has joined #zuul | 13:29 | |
| *** rlandy has joined #zuul | 13:32 | |
| quiquell | AJaeger: Fixed https://review.openstack.org/#/c/623294/ | 13:33 |
| quiquell | AJaeger: ubuntu is failing though | 13:33 |
| mordred | quiquell: yoo should be able to just do "finger 5e24c349d9a5/bc718fe4778145e1bd8e356a83ff3574@zuul.example.com | tail -f" | 13:34 |
| quiquell | mordred: was not working with a local zuul, will try again | 13:35 |
| quiquell | mordred: thanks | 13:35 |
| quiquell | mordred: maybe you can help | 13:35 |
| mordred | quiquell: was teh finger command not working at all? | 13:35 |
| quiquell | mordred: was empty, but maybe the job was in a weird state will check again | 13:35 |
| mordred | nod | 13:35 |
| quiquell | mordred: so I don't find why ubuntu is failing here https://review.openstack.org/#/c/623294/ | 13:36 |
| quiquell | mordred: after fixing centos-7 failures | 13:36 |
| mordred | quiquell: looking | 13:36 |
| quiquell | mordred: Thanks so much | 13:36 |
| *** gtema has quit IRC | 13:38 | |
| mordred | oh. uhm | 13:38 |
| *** panda|lunch is now known as panda | 13:38 | |
| *** bhavikdbavishi has quit IRC | 13:42 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Only move directory contents once in merge-output-to-logs https://review.openstack.org/632092 | 13:47 |
| mordred | quiquell: I think may help ^^ | 13:47 |
| *** badboy has quit IRC | 13:48 | |
| quiquell | mordred: The moving was the error ? was somewhere hide before it :-) | 13:50 |
| mordred | *maybe* - that's certainly broken in that job log - maybe something else is broken too | 13:50 |
| tobiash | mordred: re dockerfiles, we should test if unreleased versions in requirements.txt work in this case (sometimes they are needed) | 13:51 |
| mordred | tobiash: agree | 13:51 |
| tobiash | mordred: because we accidentally upgraded our ansible to 2.7.x because of https://bugs.launchpad.net/pbr/+bug/1677243 | 13:51 |
| openstack | Launchpad bug 1677243 in PBR "Pbr cannot include unreleased dependencies from requirements" [Medium,Confirmed] | 13:51 |
| tobiash | we had in our dockerfile pip install <zuul-src> which triggered this bug | 13:52 |
| mordred | ah - lovely | 13:53 |
| mordred | the existing patch also has a bug in it related to how pip calculates transitive deps and how it installs things specified in the extras entries | 13:54 |
| mordred | but I've got a half-finished fix locally - I'll finish it once I finish this coffee | 13:55 |
| tobiash | cool | 13:55 |
| *** hashar has quit IRC | 13:55 | |
| *** hashar has joined #zuul | 13:56 | |
| tobiash | mordred: I don't really understand the py36 failure on https://review.openstack.org/631930 | 13:56 |
| tobiash | mordred: it looks like pip has some problems with the mirror infrastructure? | 13:56 |
| mordred | :( | 13:57 |
| tobiash | http://logs.openstack.org/30/631930/9/check/tox-py36/4e0da07/job-output.txt.gz#_2019-01-20_11_26_33_887168 | 13:57 |
| mordred | tobiash: yeah - that looks like an issue with the wheel mirror | 13:59 |
| tobiash | mordred: how does the wheel mirror work? | 13:59 |
| tobiash | does it have a distinct set of wheels or is it a pull through cache? | 13:59 |
| mordred | distinct set - we build binary wheels for everythign in openstack's openstack/requirements/global-requirements.txt | 14:00 |
| mordred | and put them into a distro and arch specific pip mirror that we configure our jobs to use in addition to the normal caching mirror | 14:00 |
| mordred | to cut down on building lxml from source 100k a day :) | 14:00 |
| tobiash | hrm, but that failure looks like the wheel cache is the only mirror then? | 14:01 |
| mordred | oh - interesting | 14:02 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 14:03 |
| quiquell | mordred: Ok rebased to yours, let's see, https://review.openstack.org/#/c/623294/ | 14:04 |
| mordred | tobiash: http://logs.openstack.org/30/631930/9/check/tox-py36/4e0da07/tox/py36-2.log | 14:04 |
| quiquell | mordred: thanks | 14:04 |
| mordred | tobiash: it shows both mirrors | 14:04 |
| tobiash | hrm, does tox and the thing running inside have different pip configs? | 14:05 |
| tobiash | or is tox explicitly adding further indexes? | 14:05 |
| mordred | they shouldn't - we set the pip config in /etc/pip.conf | 14:05 |
| mordred | I agree with you though - it certainly seems like it is, for some reason, only using the wheel mirror | 14:06 |
| tobiash | maybe the initial pip version of the venv is too old | 14:06 |
| tobiash | and cannot upgrade itself | 14:06 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: DNM: test https://review.openstack.org/631970 | 14:10 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Fix concurrency and syntax issue in merge-output-to-logs https://review.openstack.org/632092 | 14:16 |
| mordred | quiquell: frickler spotted one more issue - just pushed up another rev to that patch | 14:17 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Fix concurrency and syntax issue in merge-output-to-logs https://review.openstack.org/632092 | 14:17 |
| quiquell | rlandy: ^ | 14:19 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: DNM: test https://review.openstack.org/631970 | 14:25 |
| tobiash | mordred: tox uses pip 18.1, while the fresh venv created for ansible uses 9.0.1 | 14:34 |
| tobiash | now the question is how does tox upgrade the pip in the venv... | 14:34 |
| tobiash | mordred: found out, the virtualenv tooling (used by tox) always installs the latest version while the venv module (used by me) doesn't install the latest but the one in the system | 14:39 |
| *** kmalloc has joined #zuul | 14:44 | |
| quiquell | mordred: Everything passing now https://review.openstack.org/#/c/623294/ | 14:45 |
| quiquell | mordred: thanks | 14:45 |
| kmalloc | Morning all! | 14:46 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul-jobs master: Default private_ipv4 to use public_ipv4 address when null https://review.openstack.org/623294 | 14:46 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: DNM: test https://review.openstack.org/631970 | 14:48 |
| hashar | tobiash: yes virtualenv defaults to adding setuptools, pip and wheel in the virtualenv and it also defaults to getting the latest version from pypi | 14:49 |
| rlandy | quiquell: thanks for completing that | 14:49 |
| hashar | tobiash: but you can instead use what is available on the host by using: virtualenv --no-download (and I think that can be passed using an env variable | 14:49 |
| hashar | tobiash: twist: one can also disable installation of those three packages with --no-setuptools --no-pip and/or --no-wheel | 14:50 |
| tobiash | hashar: thanks, I indeed need a recent one | 14:50 |
| tobiash | and venv doesn't seem to offer me a recent one, so trying virtualenv now | 14:50 |
| tobiash | hashar: do you know if generally venv should be favored over virtualenv or vice versa? | 14:51 |
| hashar | I just use tox :-] | 14:52 |
| tobiash | tox uses virtualenv :) | 14:52 |
| hashar | yeah ;) | 14:52 |
| hashar | theorically one could get the dependencies installed by the distribution (something like apt-get install $(bindep)) then find a trick to have tox to not install anypackage but only use system packages | 14:53 |
| hashar | which would then run the tests using the distro packages instead of whatever is on pypi. But I am probably overthinking | 14:53 |
| pabelanger | part of me thinks, we should just add a new step into install docs, that have the operator create the virtualenvs for ansible, then update zuul.conf to point to it | 14:56 |
| pabelanger | or do what we do today, but you only get x version of ansible | 14:56 |
| tobiash | pabelanger: that's exactly what this does, the ansible installation is a pre step in tox | 14:57 |
| pabelanger | tobiash: ah, I see now. manage-ansible expected to be called by operator | 14:59 |
| tobiash | yes | 14:59 |
| tobiash | and potentially optionally at executor startup, but I'd default to the operator | 15:00 |
| pabelanger | I use a role locally to prepare ansible, so nested ansible. Works okay. I know there have been talks at ansiblefest so also create a simple way of preparing ansible | 15:02 |
| pabelanger | let me look to see where that ended up | 15:02 |
| pabelanger | I also think mazer is going to support something like that too, but not 100% | 15:02 |
| tobiash | after switching from venv to virtualenv it works :) | 15:03 |
| *** avass has joined #zuul | 15:03 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: WIP: Manage ansible installation within zuul https://review.openstack.org/631930 | 15:04 |
| *** themroc has quit IRC | 15:12 | |
| *** themroc has joined #zuul | 15:12 | |
| tobiash | pabelanger: responded on 631930 | 15:17 |
| pabelanger | thanks, will review in a bit | 15:18 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: DNM: test https://review.openstack.org/631970 | 15:28 |
| *** dkehn has joined #zuul | 15:41 | |
| *** bhavikdbavishi has joined #zuul | 15:45 | |
| *** quiquell is now known as quiquell|off | 15:58 | |
| *** saneax|mtg has quit IRC | 16:02 | |
| *** bhavikdbavishi has quit IRC | 16:03 | |
| *** themroc has quit IRC | 16:08 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Fix concurrency and syntax issue in merge-output-to-logs https://review.openstack.org/632092 | 16:14 |
| *** bhavikdbavishi has joined #zuul | 16:17 | |
| *** corvus has joined #zuul | 16:42 | |
| *** avass has quit IRC | 16:46 | |
| *** bhavikdbavishi has quit IRC | 16:50 | |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/zuul-jobs master: Add a slack-notify role https://review.openstack.org/623594 | 16:59 |
| *** hashar has quit IRC | 17:04 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Revert "Fix concurrency and syntax issue in merge-output-to-logs" https://review.openstack.org/632135 | 17:14 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Use find instead of ls for determining dir emptiness https://review.openstack.org/632136 | 17:18 |
| pabelanger | Hmm, I am getting a traceback from fingergw service, because I don't have DNS setup: http://paste.openstack.org/show/743060/ | 17:24 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Use find instead of ls for determining dir emptiness https://review.openstack.org/632136 | 17:25 |
| pabelanger | I remember having this issue before, but cannot remember how I fixed it (aside from setting up dns) | 17:25 |
| *** bhavikdbavishi has joined #zuul | 17:26 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Use find instead of ls for determining dir emptiness https://review.openstack.org/632136 | 17:33 |
| tobiash | pabelanger: the executor name must be resolvable or an ip address | 17:40 |
| tobiash | https://zuul-ci.org/docs/zuul/admin/components.html#attr-executor.hostname | 17:41 |
| pabelanger | tobiash: ah, thanks. I knew we had a setting some place | 17:42 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Improve debug logging for fingerclient https://review.openstack.org/632140 | 17:43 |
| pabelanger | tobiash: some extra debug to help in the future^ | 17:43 |
| corvus | pabelanger, tobiash: see my note on https://review.openstack.org/632140 | 17:48 |
| pabelanger | corvus: sure! | 17:49 |
| tobiash | Oh yes, that makes sense | 17:49 |
| *** logan- has quit IRC | 17:52 | |
| *** logan_ has joined #zuul | 17:53 | |
| *** logan_ is now known as logan- | 17:53 | |
| *** bhavikdbavishi has quit IRC | 18:00 | |
| *** electrofelix has quit IRC | 18:05 | |
| *** openstackgerrit has quit IRC | 18:07 | |
| *** openstackgerrit has joined #zuul | 18:11 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Improve exception handling of fingerclient https://review.openstack.org/632140 | 18:11 |
| pabelanger | tobiash: corvus: ^updated as requested | 18:11 |
| *** jpena is now known as jpena|off | 18:13 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Build zuul containers with dockerfile not pbrx https://review.openstack.org/631840 | 18:14 |
| mordred | corvus, tobiash: ^^ that (And its depends-on in system-config) deal with our optional depends expressed in extras a bit better I think | 18:14 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: docker: add ability to restrict repository names https://review.openstack.org/631848 | 18:15 |
| corvus | mordred: thanks, i'm working through the system-config change now | 18:15 |
| mordred | corvus: woot | 18:16 |
| corvus | mordred: what version of debian is python:slim? | 18:22 |
| tobiash | corvus: it's stretch | 18:27 |
| tobiash | according to https://hub.docker.com/_/python/ it's python 3.7.2 on stretch | 18:28 |
| corvus | it looks like we could get bubblewrap from debian via stretch-backports. it's also in buster, but i don't think python has a buster image | 18:28 |
| corvus | https://packages.debian.org/stretch-backports/bubblewrap | 18:28 |
| *** saneax|mtg has joined #zuul | 18:28 | |
| corvus | (it is also in stretch without backports, but it's very old) | 18:28 |
| corvus | but perhaps we could enable the stretch-backports repository and install it that way? | 18:29 |
| mordred | corvus: that's a good point ... | 18:31 |
| mordred | corvus: although I think the stretch-backports is there you just have to explicitly pin to it ... let me poke at doing that | 18:31 |
| mordred | woudl be WAY less lines than that chunk | 18:31 |
| corvus | yeah, and gets us out of the business of maintaining that ppa :) | 18:32 |
| *** saneax|mtg has quit IRC | 18:33 | |
| corvus | mordred: one more comment on that change | 18:33 |
| tobiash | wow, the default bubblewrap version in stretch is really old... | 18:33 |
| *** saneax|mtg has joined #zuul | 18:33 | |
| tobiash | mordred: how do you get the intermediate images? | 18:34 |
| tobiash | docker doesn't tag them by default | 18:34 |
| tobiash | ah, found it, using --target during docker build | 18:37 |
| tobiash | mordred: in the zuul change I think you're missing the targets in the zuul.yaml (as you have them in the system-config part) | 18:40 |
| tobiash | mordred: and a minimal dockerignore file would be helpful, I don't want to copy my whole tox dir into the context during local builds ;) | 18:42 |
| tobiash | mordred: feel free to squash https://review.openstack.org/631929 into your change | 18:42 |
| mordred | tobiash: yes - totally missed the target lines - and also, dockerfile would be great! | 18:49 |
| tobiash | hrm, I cannot build these zuul images, it somehow refuses to use our proxy :/ | 18:50 |
| tobiash | hm, installation of sudo works, after that it seems to unset the proxy vars | 18:52 |
| mordred | tobiash: :( - how do you normally use the proxy from inside the image? is there something I'm doing that's stupid? | 18:52 |
| mordred | *weird* | 18:52 |
| tobiash | mordred: I'm using build-arg | 18:52 |
| tobiash | mordred: like this: docker build --build-arg http_proxy=$http_proxy --build-arg https_proxy=$http_proxy --build-arg no_proxy=$no_proxy -t zuul-base --target zuul-base . | 18:53 |
| tobiash | that forwards my local proxy vars into the build context | 18:53 |
| tobiash | and build-args won't persist into the image | 18:53 |
| mordred | tobiash: I wonder if those vars aren't exported - so they assemble script isn't seeing them? no - that doesn't make sense, you said the sudo install worked and that's in a subshell | 18:54 |
| tobiash | mordred: but either the installation of sudo or the source of the js stuff could break it | 18:54 |
| *** saneax|mtg is now known as saneax | 18:54 | |
| mordred | tobiash: oh - maybe sourcing the js stuff is the culprit - and maybe we should just run that like a normal script instead of with source? | 18:54 |
| tobiash | probably | 18:55 |
| tobiash | what's the reason for sourcing it? | 18:56 |
| mordred | me overthinking things - no good reason | 18:57 |
| tobiash | hrm, not sourcing it doesn't fix it | 18:57 |
| tobiash | ah I know the culprit | 18:57 |
| mordred | ok. well - I'm going to remove that anyway because it's weird | 18:57 |
| tobiash | sudo kills the proxy vars | 18:57 |
| mordred | AHA | 18:57 |
| mordred | we need sudo -E in that script | 18:57 |
| mordred | tobiash: I've got that updated in my latest patch | 19:00 |
| pabelanger | tobiash: thanks for config setting again, fixed my issue | 19:00 |
| mordred | tobiash: also - there is an issue with transitive deps I'm still poking at locally | 19:01 |
| tobiash | pabelanger: you're welcome | 19:01 |
| pabelanger | that, and opening some firewall ports :) | 19:01 |
| tobiash | mordred: I wonder if we instead check for EUID==0 and use sudo only it we're not root | 19:01 |
| tobiash | however sudo -E still makes sense | 19:02 |
| tobiash | mordred: sudo -E fixes the downloads | 19:02 |
| mordred | tobiash: that would be nicer than the wrapper script | 19:02 |
| tobiash | oh now I see, docker-install-js-tools is just a wrapper script because of sudo | 19:03 |
| mordred | yeah | 19:03 |
| mordred | didn't want to update that script too much directly without wider discussion | 19:03 |
| tobiash | I think conditionally setting a SUDO var and use "${SUDO} foo" is the correct thing to do here | 19:05 |
| mordred | ++ | 19:07 |
| *** dkehn has quit IRC | 19:09 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Add a minimal dockerignore file https://review.openstack.org/631929 | 19:10 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Build zuul containers with dockerfile not pbrx https://review.openstack.org/631840 | 19:10 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Build zuul containers with dockerfile not pbrx https://review.openstack.org/631840 | 19:13 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Use sudo -E in install-js-tools only if not root https://review.openstack.org/632148 | 19:13 |
| mordred | tobiash: I split the install-js-tools change into its own patch with that last thing ^^ (realized it could be landed by itself as an overall improvement) | 19:13 |
| tobiash | mordred: commented on 632148 | 19:17 |
| -openstackstatus- NOTICE: The error causing post failures on jobs has been corrected. It is safe to recheck these jobs. | 19:19 | |
| mordred | tobiash: hrm. echo $EUID shows that - but env | grep UID does not ... I have been foiled! | 19:19 |
| tobiash | that's a very special variable, env maybe doesn't list it | 19:20 |
| tobiash | mordred: wait, that's internal to bash | 19:21 |
| tobiash | mordred: ah ok, the shebang lists /bin/bash | 19:22 |
| tobiash | so the variable can be used | 19:22 |
| mordred | tobiash: yah - we should be ok there using it | 19:22 |
| mordred | *phew* | 19:22 |
| tobiash | mordred: ok, both should work, not sure if my comment is -1 worthy though | 19:24 |
| mordred | tobiash: might as well clean it up while we're working on it | 19:24 |
| *** hashar has joined #zuul | 19:25 | |
| mordred | tobiash: building these images over and over again is totally driving me crazy | 19:32 |
| tobiash | that's fun ;) | 19:33 |
| corvus | mordred: after lunch, i plan on working on the pass-to-parent versions of these jobs | 19:33 |
| tobiash | at least if you have a decent network speed | 19:33 |
| mordred | tobiash: I'm having this issue where for some reason when the builder image goes through the 'build wheels for everything so we don't need dev tools in the final images' step - it's not building a wheel for netifaces which is needed by openstacksdk for the executor image | 19:33 |
| mordred | tobiash: but - you kind of have to go from the start and do all the steps to get there to see if a change fixes it | 19:34 |
| SpamapS | The wheel has worked for millenia. Build the wheel. | 19:34 |
| mordred | SpamapS: I feel like that should be the slogan for a new self-help corporate process cult | 19:34 |
| tobiash | mordred: oh these kind of round trips are driving me crazy too | 19:35 |
| SpamapS | I've taken to just putting mirrors closer to build resources rather than trying to plumb caches in. | 19:35 |
| mordred | SpamapS: well, for this it's not a cache - as much as it's a build artifact | 19:35 |
| SpamapS | Oh, why are you building them over and over? | 19:36 |
| mordred | SpamapS: the builder image in the multi-step build builds binary wheels - so that in the final image we can pip install the wheel without needing to have compilers and headers and whatnot installed into the final image | 19:36 |
| mordred | SpamapS: the step that's currently broken is one of the things that needs to get built is not getting built | 19:37 |
| mordred | so the final image stage is bombing out on not being able to pip install netifaces because of lack of gcc | 19:37 |
| mordred | I'm 100% sure it's a comma in the wrong place somewhere | 19:37 |
| mordred | the stack works quite well otherwise | 19:37 |
| tobiash | mordred: can you check for the missing wheel in a run step? | 19:37 |
| tobiash | that way you could catch the error earlier and iterate faster | 19:38 |
| mordred | tobiash: *maybe* - unfortunately the wheels are stored in the wheel dir by hashed directories | 19:38 |
| tobiash | ugh | 19:38 |
| mordred | tobiash: yeah | 19:38 |
| mordred | tobiash: but - if this most recent run doesn't do it - I'm going to start putting in some run statements to see what's going wrong | 19:38 |
| mordred | Running setup.py bdist_wheel for netifaces: started | 19:40 |
| mordred | Running setup.py bdist_wheel for netifaces: finished with status 'done' | 19:40 |
| mordred | Stored in directory: /output/wheels/wheels/23/8f/f3/7054578f04c904f70757c5c85a6e2823baa69d42365526e93d | 19:40 |
| mordred | ok. it at least built the wheel this time | 19:40 |
| mordred | corvus: I'm excited about the pass-to-parent versions of these jobs! | 19:42 |
| tobiash | corvus: if you're planning a zuul release soon, 613261 would be something that I think should go into the next release | 19:42 |
| * mordred shows everybody his excited face | 19:42 | |
| tobiash | yeah, pass-to-parent will make many things easier :) | 19:44 |
| pabelanger | also interested in testing it | 19:47 |
| mordred | tobiash: BWAHAHAHAHAHAHA - I found the issue | 19:47 |
| mordred | tobiash: it was ... I wasn't actually copying the wheel cache content in - so in the context of the problem image, there WAS no wheel cache | 19:47 |
| tobiash | mordred: that explains something ;) | 19:48 |
| mordred | yeah | 19:48 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Use sudo -E in install-js-tools only if not root https://review.openstack.org/632148 | 20:03 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Build zuul containers with dockerfile not pbrx https://review.openstack.org/631840 | 20:03 |
| mordred | tobiash: ^^ I think that latest version works now | 20:04 |
| tobiash | trying | 20:04 |
| *** saneax has quit IRC | 20:22 | |
| *** ianw is now known as ianw_pto | 20:26 | |
| *** studarus has joined #zuul | 20:28 | |
| tobiash | mordred: I've patched requirements.txt to pull in a patched ansible and it pulled in the latest version instead | 20:31 |
| mordred | tobiash: poo. ok. I'll poke at that next | 20:32 |
| tobiash | mordred: an easy solution is to not rely on pbr to install, but pip install the requirements first | 20:33 |
| mordred | oh - right - because of pbr not supporting -e syntax in the first place :) | 20:34 |
| tobiash | yes | 20:35 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul-jobs master: Propose some job writing guidelines https://review.openstack.org/631507 | 20:35 |
| mordred | tobiash: yes - I agree - that's not a bad idea. | 20:36 |
| *** studarus has quit IRC | 20:47 | |
| *** studarus has joined #zuul | 20:48 | |
| mordred | tobiash: hrm re: the issue with -e on the build_wheel line | 21:10 |
| tobiash | mordred: docker build still running | 21:11 |
| tobiash | but it looks like it works | 21:12 |
| mordred | oh good - I was just looking at your review comment on the python-builder assemble script bdist_wheel line | 21:12 |
| mordred | but I woudln't expect requiremnts to get involved there | 21:12 |
| tobiash | your followup worked | 21:13 |
| mordred | so I'm glad the other bits seem to be making it work | 21:13 |
| mordred | sweet | 21:13 |
| tobiash | it cloned the -e dep twice but it worked | 21:13 |
| mordred | yeah - there are few parts of this that aren't the most efficient at build time :) | 21:14 |
| tobiash | the important thing is that it works :) | 21:14 |
| *** ssbarnea|bkp2 has joined #zuul | 21:16 | |
| mordred | yes! | 21:17 |
| openstackgerrit | Tobias Urdin proposed openstack-infra/zuul-jobs master: Use become for packages in upload-puppetforge https://review.openstack.org/632163 | 21:17 |
| *** ssbarnea|rover has quit IRC | 21:18 | |
| tobiash | mordred: did you see the comment about bwrap on 631840? | 21:20 |
| mordred | tobiash: I did - I actually tried that locally in one iteration - but since bubblewrap is getting installed from bindep anyway it didn't make much of a difference in the end | 21:21 |
| tobiash | ah, then just ignore it | 21:22 |
| *** studarus has quit IRC | 21:24 | |
| mordred | tobiash: my nitpicky brain wants to invent something to make that better - but I don't think it'll actually wind up being better :) | 21:25 |
| tobiash | bwrap is that small that it's just wasted time | 21:25 |
| mordred | tobiash: yeah | 21:26 |
| mordred | the tox failures are a bit disturbing | 21:26 |
| mordred | corvus: I think https://review.openstack.org/#/q/topic:replace-pbrx is all ready for review | 21:27 |
| mordred | SpamapS: ^^ you might also care - or not care | 21:27 |
| tobiash | mordred: the zuul-build-image job seems to be unused | 21:27 |
| mordred | oh! so it is. one sec | 21:28 |
| tobiash | the tox failures don't use anything from that so looks like unrelated failures | 21:29 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Build zuul containers with dockerfile not pbrx https://review.openstack.org/631840 | 21:29 |
| mordred | tobiash: I totally forgot to add all three of the jobs :) | 21:30 |
| tobiash | :) | 21:30 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Use node v10 instead of node v8 https://review.openstack.org/632165 | 21:30 |
| tobiash | mordred: what's the official status of zuul-migrate? | 21:33 |
| tobiash | I thought it was more an openstack tooling for the switch to v3? | 21:33 |
| tobiash | do we really need a zuul-migrate image? | 21:34 |
| mordred | yeah - I think we left it as some example of how things could work? | 21:34 |
| mordred | probably not - I included it just because we _had_ been making it with pbrx | 21:34 |
| tobiash | ah ok | 21:34 |
| mordred | but it's maybe a better idea to just exclude it | 21:34 |
| tobiash | just thinking if we have it packaged as a docker image we make it somewhat an official tool | 21:34 |
| mordred | yah | 21:35 |
| mordred | on a similar but different note - is a docker image for zuul-bwrap useful? | 21:35 |
| mordred | I suppose possibly so? | 21:35 |
| tobiash | I was just about to ask the same ;) | 21:35 |
| tobiash | I doubt it | 21:35 |
| mordred | how about I push up a follow up patch removing those so it's easy to discuss | 21:36 |
| tobiash | fine for me | 21:36 |
| corvus | why wouldn't a docker image for zuul-bwrap be useful? | 21:36 |
| tobiash | corvus: what's the intentional of zuul-bwrap command? | 21:38 |
| corvus | if you find that you need to test something out with zuul-bwrap, and you typically use docker images for deployment, then that would be the time it would be useful, yeah? | 21:38 |
| corvus | tobiash: it's mostly a debugging/test tool | 21:38 |
| tobiash | corvus: is it for debugging executors? | 21:38 |
| corvus | tobiash: yes | 21:39 |
| tobiash | my first hunch would have been that I'd use it when I exec into the executor container | 21:39 |
| corvus | i guess it's advanced enough we could just say that ^ | 21:39 |
| corvus | so either do that, or override the command | 21:39 |
| corvus | okay, good talk. i'm on board with dropping it. :) | 21:40 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Remove zuul-migrate and zuul-bwrap images https://review.openstack.org/632167 | 21:40 |
| mordred | tobiash, corvus: ^^ put that up for discussion | 21:41 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul-jobs master: Propose some job writing guidelines https://review.openstack.org/631507 | 21:41 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Add docker image build jobs https://review.openstack.org/632172 | 21:47 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Switch to zuul-jobs docker jobs https://review.openstack.org/632173 | 21:48 |
| corvus | mordred, tobiash: ^ building on that | 21:48 |
| tobiash | Cool :) | 21:50 |
| corvus | mordred: bug in 631840 | 21:51 |
| *** spsurya has quit IRC | 22:09 | |
| fungi | tobias-urdin: https://review.openstack.org/632163 looks good to me. if some other sympathetic soul in zuul-jobs-core agrees, i can try reenqueuing the tag for your canary again | 22:11 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Build zuul containers with dockerfile not pbrx https://review.openstack.org/631840 | 22:18 |
| mordred | corvus: fixed - you want me to rebase your patch on it? | 22:18 |
| corvus | mordred: ++ | 22:18 |
| *** hashar has quit IRC | 22:18 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Switch to zuul-jobs docker jobs https://review.openstack.org/632173 | 22:19 |
| openstackgerrit | Merged openstack-infra/zuul master: Add a minimal dockerignore file https://review.openstack.org/631929 | 22:25 |
| openstackgerrit | Merged openstack-infra/zuul master: Use sudo -E in install-js-tools only if not root https://review.openstack.org/632148 | 22:29 |
| openstackgerrit | Merged openstack-infra/zuul master: Improve exception handling of fingerclient https://review.openstack.org/632140 | 22:30 |
| *** dkehn has joined #zuul | 22:40 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Build zuul containers with dockerfile not pbrx https://review.openstack.org/631840 | 22:46 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Remove zuul-migrate and zuul-bwrap images https://review.openstack.org/632167 | 22:46 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Switch to zuul-jobs docker jobs https://review.openstack.org/632173 | 22:46 |
| corvus | mordred: can you add openstackzuul to openstack/ ? | 22:51 |
| mordred | corvus: yes | 22:59 |
| mordred | corvus: with nodepool should we just add nodepool to the allowed-projects of zuul-build-image and friends and override the image variables in the nodepool repo? | 23:03 |
| corvus | mordred: i think so | 23:03 |
| mordred | corvus: ok. I'll send in some patches for that | 23:04 |
| corvus | (and eventually, we'd probably put those jobs in a zuul tenant config repo or something) | 23:05 |
| mordred | yeah | 23:06 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Allow nodepool to use zuul-*-image jobs https://review.openstack.org/632186 | 23:06 |
| openstackgerrit | Monty Taylor proposed openstack-infra/nodepool master: Build images with Dockerfile instead of pbrx https://review.openstack.org/632187 | 23:15 |
| mordred | corvus: ^^ I think taht should about do it | 23:15 |
| corvus | mordred: needs a depends-on | 23:15 |
| mordred | corvus: oh good point | 23:15 |
| corvus | zuul told me | 23:15 |
| corvus | mordred: why publish the -base images? | 23:16 |
| openstackgerrit | Monty Taylor proposed openstack-infra/nodepool master: Build images with Dockerfile instead of pbrx https://review.openstack.org/632187 | 23:16 |
| mordred | corvus: mostly was just copying what pbrx had done - we could totalluy remove that | 23:17 |
| mordred | corvus: if someone wants to build their own images using ours as base images, there's always the zuul and nodepool utility images that are straightforward | 23:17 |
| mordred | corvus: actually, come to think of it - we could ditch -base competely and just use zuul and nodepool for that same role ourselves | 23:18 |
| corvus | mordred: oh, i see, sorry i missed something earlier | 23:18 |
| corvus | so -base doesn't have a command, but 'nodepool' and 'zuul' do | 23:19 |
| mordred | yeah. but I think that's maybe fine... | 23:19 |
| corvus | i kinda like that, i think... | 23:19 |
| corvus | i dunno. don't make me decide today. | 23:20 |
| mordred | corvus: http://paste.openstack.org/show/743067/ is an alternative | 23:20 |
| mordred | corvus: how about if I push up a patch with that ^^ as a followup | 23:20 |
| mordred | and we can discuss on the patch | 23:20 |
| mordred | (I could go either way) | 23:20 |
| corvus | the advantage of the current system is there is a distinct image which is just the system, and other images which have commands. it feels nice that they are separate. | 23:21 |
| mordred | corvus: yah - that is a good point | 23:21 |
| corvus | i agree though, the downsides to using the 'zuul' and 'nodepool' images as such are not huge. :) | 23:22 |
| openstackgerrit | Monty Taylor proposed openstack-infra/nodepool master: Build images with Dockerfile instead of pbrx https://review.openstack.org/632187 | 23:23 |
| openstackgerrit | Monty Taylor proposed openstack-infra/nodepool master: Stop building an explicit nodepool-base image https://review.openstack.org/632188 | 23:23 |
| mordred | corvus: (the second patch made me realize there was a bug in the first patch) | 23:24 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Stop building an explicit zuul-base image https://review.openstack.org/632189 | 23:26 |
| mordred | corvus: and there's the equiv for zuul | 23:26 |
| SpamapS | equivs! wow | 23:27 |
| * SpamapS couldn't think of it yesterday | 23:28 | |
| mordred | SpamapS: yes. equivs | 23:28 |
| mordred | SpamapS: you might find https://review.openstack.org/#/c/631878/9/docker/python-builder/Dockerfile amusing :) | 23:28 |
| openstackgerrit | James E. Blair proposed openstack-infra/nodepool master: Build images with Dockerfile instead of pbrx https://review.openstack.org/632187 | 23:28 |
| corvus | mordred: ^ wrong depends-on | 23:28 |
| mordred | corvus: awesome | 23:29 |
| mordred | SpamapS: especially lines 18-25 | 23:29 |
| SpamapS | yeah | 23:30 |
| SpamapS | amazing | 23:30 |
| openstackgerrit | James E. Blair proposed openstack-infra/nodepool master: Stop building an explicit nodepool-base image https://review.openstack.org/632188 | 23:30 |
| corvus | mordred: rebase ^ | 23:30 |
| fungi | SpamapS: yes, equivs is mightily handy! | 23:59 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!