| tristanC | matburt: to change your email setting, you need to go through the "Settings" on the top right here: https://softwarefactory-project.io/sf/user_settings.html | 02:10 |
|---|---|---|
| tristanC | matburt: there you need to uncheck "Synchronize user data with the Identity Provider" and set another mail | 02:11 |
| matburt | The email finally came through an hour ago | 02:12 |
| tristanC | matburt: i would recommend you start by not setting any nodeset and default to https://github.com/ansible-network/zuul-config/blob/master/zuul.d/jobs.yaml#L42-L45 | 02:13 |
| tristanC | that is because the non container node uses rdo-cloud capacity and jobs may be queued longer | 02:14 |
| matburt | Gotcha. I'm definitely going to need py2 and py3 | 02:14 |
| tristanC | and for the first job, you may want to begin with something simple like codestyle or unit tests | 02:15 |
| matburt | Yah I'm going to run the python linter against it | 02:17 |
| matburt | I have a question about gating actually | 02:17 |
| tristanC | feel free to ask any questions about zuul here or in #zuul | 02:18 |
| matburt | ideally I think `check` will run the tests and `gate` would be approvals on the github PR? | 02:19 |
| matburt | does that equate to the noop job on the gate? How does that work with multi-project gating? If I just need to go read some more I'm happy to do that also ;) | 02:19 |
| tristanC | matburt: gate usualy describe tests that are being run before the merge, using the PR + the current HEAD of the branch | 02:20 |
| tristanC | matburt: and to be succesfull, Zuul needs to be in charge of the PR merge, actually Zuul needs to be the only one in charge of the PR merge | 02:21 |
| matburt | gotcha, so check might equate to linters where gate would run the full suite? | 02:21 |
| matburt | Yep that's my idea behind it | 02:21 |
| matburt | and then it would be gate+approvals for it to exit the gate part of the pipeline? | 02:22 |
| tristanC | matburt: what matter is to run gate job that can catch issue caused by other PR merged after check but before gate | 02:23 |
| matburt | I'm starting here with Ansible Runner, with the idea to transition AWX integration testing over to zuul and have an AWX/Runner gate | 02:23 |
| matburt | gotcha so the check runs against just the branch and the gate runs against the branch+merge? | 02:23 |
| matburt | (Runner being a component of AWX in the next release of AWX) | 02:24 |
| tristanC | matburt: usually you want check to have all the jobs so that you get an accurate status automatically, when PR are created/updated | 02:24 |
| tristanC | matburt: then you may want to run a subset of the check in the gate to speed things up, but i think it's better to re-run the same jobs, just to be sure HEAD is not broken | 02:25 |
| tristanC | matburt: check runs with HEAD+PR on pr-created/updated event (or when a 'recheck' comment is added) | 02:26 |
| matburt | that makes sense... is there batching with other outstanding branches pending review in the same pipelines? | 02:26 |
| matburt | or is that configurable? | 02:26 |
| tristanC | matburt: gate runs with current HEAD+PR in gate queue+PR on custom condition, at the moment it's when check succeed + 'mergeit' label applied | 02:27 |
| tristanC | matburt: jobs can be assigned to a queue so that PR from multiple projects share the same flow | 02:27 |
| matburt | gotcha, that makes sense | 02:28 |
| tristanC | matburt: basically for AWX/Runner, you want to read that guide: https://review.openstack.org/#/c/571420/3/doc/source/user/howtos/pti.rst | 02:28 |
| tristanC | matburt: since there is no 'ansible/zuul-jobs', you can use 'ansible/AWX' or 'ansible/runner' in place of the howto 'org-jobs' project | 02:30 |
| matburt | I'll submit this to gerrit in the morning, this is the current runner configuration (I'll move the linter into the gate also) https://github.com/ansible/ansible-runner/blob/master/.zuul.yaml along with the job configuration here (I'll change the nodeset configuration): https://github.com/ansible/runner-zuul-jobs/blob/master/.zuul.d/jobs.yaml | 02:30 |
| tristanC | matburt: that looks correct | 02:35 |
| matburt | the review will just be adding those two repos | 02:35 |
| matburt | I really appreciate your help on this... zuul takes a little getting used to | 02:36 |
| matburt | I spent a day or so trying to get zuul set up on my own and.... well, it's not easy | 02:36 |
| tristanC | excellent, feel free to add AWX too so that you can toy with the PTI guide | 02:36 |
| matburt | Yep looking forward to that. | 02:37 |
| tristanC | you're welcome, we are working hard trying to reduce the learning curve steepness... | 02:39 |
| tristanC | there are more guide in the progress to explain more concepts like how to export job logs, do pypi publish on tag, etc. | 02:41 |
| matburt | that's good to hear, that'll be excellent for Runner | 02:42 |
| matburt | Right now I build those by hand | 02:42 |
| gundalow | matburt: gh/ansible has the SF Zuul app installed. You will need to add the repo to the list in the settings | 05:10 |
| gundalow | matburt: https://github.com/apps/softwarefactory-project-zuul/installations/234854 think that's the right url | 05:12 |
| *** jpena|off is now known as jpena | 07:54 | |
| sfbender | Tristan de Cacqueray created www.softwarefactory-project.io master: Add gertty blog post https://softwarefactory-project.io/r/13534 | 08:07 |
| tristanC | gundalow: matburt: i think the link to install the GitHub app is: https://github.com/apps/softwarefactory-project-zuul | 08:18 |
| gundalow | Ah, thanks | 08:18 |
| gundalow | Apps installed, just need to request permissions for individual repos | 08:19 |
| gundalow | matburt: you will also need to enable squash commits, and other conf as listed in https://github.com/ansible/community/blob/master/group-network/roles_development_process.rst#new-role | 08:20 |
| sfbender | Fabien Boucher created www.softwarefactory-project.io master: Update blog-zuul-01-setup-sandbox.rst according to last proposed changes https://softwarefactory-project.io/r/13536 | 10:50 |
| sfbender | Fabien Boucher created software-factory/sf-docs master: Update sf-doc copyright year notice https://softwarefactory-project.io/r/13537 | 10:57 |
| *** jpena is now known as jpena|lunch | 11:27 | |
| *** sshnaidm is now known as sshnaidm|afk | 11:28 | |
| *** sfbender has quit IRC | 11:50 | |
| *** jpena|lunch is now known as jpena|off | 12:31 | |
| *** jpena|off is now known as jpena | 12:32 | |
| matburt | having trouble accessing the web services this morning, is the service down? | 13:11 |
| spredzy | matburt: the infrastructure on which SF relies is down at the moment | 13:13 |
| matburt | gotcha... what infrastructure is it on? | 13:13 |
| spredzy | An OpenStack Cloud called RDO-Cloud | 13:13 |
| spredzy | Managed internally by Red Hat ops | 13:14 |
| matburt | gotcha | 13:15 |
| rcarrillocruz | i thikn rdo is down yeah | 13:34 |
| rcarrillocruz | :/ | 13:34 |
| rcarrillocruz | oh look, spredzy is here! | 13:34 |
| rcarrillocruz | :P | 13:34 |
| rcarrillocruz | i hope rdo is back in a couple hours, i have a cloud vpn demo and i was hoping to show openstack provisioner | 13:35 |
| rcarrillocruz | sigh | 13:35 |
| rcarrillocruz | i need to get a vexxhost account | 13:35 |
| spredzy | rcarrillocruz: o/ | 13:36 |
| spredzy | I am a #softwarefactory lurker | 13:37 |
| spredzy | ;) | 13:37 |
| *** sshnaidm|afk has quit IRC | 14:10 | |
| *** sshnaidm|afk has joined #softwarefactory | 14:19 | |
| *** sshnaidm|afk is now known as sshnaidm | 14:20 | |
| pabelanger | rcarrillocruz: yah, rdocloud is down again it seems | 14:26 |
| pabelanger | ovs | 14:26 |
| dmsimard | This cloud needs to be branded something else than RDO cloud because it's not RDO's fault :( | 14:27 |
| rcarrillocruz | haha | 14:27 |
| rcarrillocruz | is the RDO outage still ongoing? | 15:30 |
| rcarrillocruz | bah | 15:30 |
| rcarrillocruz | confirmed | 15:30 |
| rcarrillocruz | i can't even get to the dashboard | 15:30 |
| pabelanger | yah, no change from what I see | 15:31 |
| jruzicka | I expected nothing but RDO Cloud still let me down. | 16:03 |
| jruzicka | yes, it shouldn't be called RDO Cloud. | 16:03 |
| jruzicka | In current state, it shouldn't even exist. | 16:03 |
| *** jpena is now known as jpena|away | 17:27 | |
| matburt | looks like things are back up? | 17:46 |
| matburt | https://softwarefactory-project.io/r/#/c/13538/ | 17:52 |
| matburt | looks like a connectivity problem? | 17:52 |
| pabelanger | matburt: we should look to create ansible/ansible-zuul-jobs, or ansible/zuul-jobs for common shared job repo for all ansible namespace projects | 17:57 |
| matburt | alrighty, let me see what I can do about that | 17:58 |
| pabelanger | however | 18:00 |
| pabelanger | for now | 18:00 |
| pabelanger | you don't really need that | 18:00 |
| matburt | pabelanger: though honestly, I did just want to put runner jobs there | 18:00 |
| pabelanger | matburt: you can place them in your project to start | 18:00 |
| matburt | ansible core and tower/awx/runner tend to develop separately | 18:00 |
| pabelanger | sure, but there will be some common things | 18:00 |
| matburt | well, I did just rename the repo so I think we can roll forward. I'll update this | 18:00 |
| pabelanger | and, like I say, nothing saying this can't be in tree of ansible-runner to start | 18:01 |
| matburt | yeah I'm still on the fence as to where I want the tests to live | 18:01 |
| pabelanger | right, I think that is a larger design discussion for sure | 18:01 |
| pabelanger | but, for POC, I think safe to start them there | 18:01 |
| pabelanger | or generic ansible/ansible-zuul-jobs repo | 18:02 |
| matburt | if it's okay with you... I'll land them in zuul-jobs | 18:02 |
| pabelanger | no | 18:02 |
| pabelanger | that doesn't really work | 18:02 |
| pabelanger | actually | 18:02 |
| pabelanger | where is zuul-jobs | 18:02 |
| pabelanger | matburt: yah, that is fine | 18:02 |
| pabelanger | sorry, thought you meant zuul-config, which was trusted | 18:03 |
| pabelanger | matburt: you should remove https://github.com/ansible/zuul-jobs/tree/master/.zuul.d first, since that would wedge zuul before the import | 18:03 |
| pabelanger | then, once project is loaded into zuul memory, you can add it back | 18:04 |
| matburt | ah interesting | 18:04 |
| pabelanger | ideally, the check job in sf.io would raise an error for that | 18:04 |
| matburt | alrighty, that's done | 18:06 |
| matburt | oh huh, the `git review` created another review | 18:08 |
| matburt | I'm a bit new to gerrit | 18:08 |
| matburt | should I close the other one? | 18:08 |
| pabelanger | yah, you need to same change-id in the commit message, for it to be updated | 18:09 |
| pabelanger | you can abandon via webui | 18:09 |
| pabelanger | matburt: also, since you have a tox.ini file, you shouldn't really need to create new jobs. We have tox jobs by default in zuul-jobs: https://zuul-ci.org/docs/zuul-jobs/jobs.html | 18:10 |
| pabelanger | so, you should be able to use tox-docs, tox-linters, etc to run on your entry points | 18:10 |
| pabelanger | then writing new ones | 18:10 |
| matburt | yep, I'm almost certain that's probably not going to work though... I wanted to get something light in and working | 18:10 |
| matburt | my tox configuration requires pipenv, a particular tox plugin that uses multiple pipenvs, and py2.7 and py3.6+ | 18:11 |
| pabelanger | as long as pip dependencies are installed when tox is run, should be fine | 18:12 |
| pabelanger | but if you need to setup things, you can use pre-run playbook then parent to the tox jobs | 18:13 |
| matburt | we'll have to see how things go, in the meantime... running the linter should be a good validation of it | 18:16 |
| pabelanger | +1 | 18:16 |
| pabelanger | matburt: ah, you need to use git commit --ammend with git-review | 18:19 |
| pabelanger | new commit create new change-id, which are new reviews in gerrit | 18:19 |
| matburt | yah, I abandoned the old one | 18:20 |
| pabelanger | so, if you git rebase, and squash your commits, you'll get them into the first patch | 18:20 |
| pabelanger | matburt: no, you need to rebase | 18:20 |
| matburt | is it possible to review multiple commits in a changeset? | 18:20 |
| pabelanger | no | 18:20 |
| matburt | interesting | 18:20 |
| pabelanger | 1 commit per change-id | 18:20 |
| pabelanger | so, if you restored your first | 18:21 |
| pabelanger | we'd can land both changes | 18:21 |
| pabelanger | but is 2 code reviews | 18:21 |
| pabelanger | which, is fine | 18:21 |
| matburt | okay done | 18:21 |
| pabelanger | but usually with gerrit workflow you ammend your patch, and push up the same change-id, so comments are in 1 place | 18:22 |
| matburt | gerrit is a bit new for me. my git usage went from emailed patches straight to github for the most part | 18:22 |
| pabelanger | yah, it takes a little change in workflow, but I find it much nicer then github | 18:22 |
| matburt | should we issue a `recheck` against that first one? looks like the 2nd one passed | 18:23 |
| pabelanger | you can | 18:23 |
| matburt | pabelanger: alrighty, that check went through | 18:37 |
| matburt | tristanC: two changesets (sorry): https://softwarefactory-project.io/r/13538 and https://softwarefactory-project.io/r/13539 | 19:11 |
| *** jpena|away is now known as jpena|off | 19:38 | |
| tristanC | matburt: hello, i'll squash them and approve them now | 22:20 |
| *** sshnaidm is now known as sshnaidm|off | 22:35 | |
| matburt | tristanC: thank ya kindly | 23:17 |
| matburt | Now I just need to flip the switch on the runner repo and let it start picking up the PRs | 23:17 |
| tristanC | matburt: as you can see, the projects are added to https://ansible.softwarefactory-project.io/zuul/projects.html | 23:18 |
| tristanC | matburt: however, there is config-error now, see: https://ansible.softwarefactory-project.io/zuul/api/config-errors | 23:18 |
| matburt | tristanC: yes, pabelanger told me to pull the job configuration out of zuul-jobs until after this got added in | 23:19 |
| matburt | otherwise it would cause problems with zuul (?) | 23:19 |
| dmsimard | matburt: hey you're getting the hang of this :D | 23:20 |
| matburt | dmsimard: yes! my mission was to have this all running on a basic level this week | 23:20 |
| matburt | https://github.com/ansible/zuul-jobs/commit/6f2ad0f42390c1d80731d4caae1ad4921cc18a96 | 23:20 |
| matburt | tristanC: I'm going to revert that one and get things moving | 23:21 |
| dmsimard | matburt: sweet :D | 23:21 |
| tristanC | matburt: let's try to use zuul gating on ansible/zuul-jobs | 23:22 |
| tristanC | matburt: so to get a PR merged, you need to add a review, e.g. in https://github.com/ansible/zuul-jobs/pull/2/files | 23:22 |
| tristanC | matburt: and then add the "mergeit" label | 23:22 |
| matburt | excellent, I also need to enable the zuul app on the repo so it'll successfully merge | 23:23 |
| tristanC | oh, but zuul-jobs doesn't have pipelines yet | 23:23 |
| matburt | it gets those from the other repo | 23:23 |
| matburt | https://github.com/ansible/zuul-config | 23:23 |
| matburt | I was thinking about redoing that but I don't know enough about the pipelines yet | 23:23 |
| matburt | For the moment, I need to go have some dinner with the fam... I'll be back later | 23:23 |
| matburt | tristanC: so if you can have a look at that and make some recommendations | 23:25 |
| tristanC | matburt: sure, so first, visit https://github.com/apps/softwarefactory-project-zuul and install it on zuul-jobs and ansible-runner | 23:26 |
| tristanC | matburt: then let's configure branch protection for zuul-jobs: https://softwarefactory-project.io/docs/user/zuul_user.html#configure-branch-protection | 23:27 |
| tristanC | the required status is "ansible/check" (not "local/test") | 23:27 |
| tristanC | once this is done, let's recheck: https://github.com/ansible/zuul-jobs/pull/3 | 23:28 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!