*** yassine has quit IRC | 01:33 | |
*** yassine has joined #softwarefactory | 01:35 | |
*** caowei has joined #softwarefactory | 02:09 | |
*** maeca has joined #softwarefactory | 02:36 | |
tristanC | pabelanger: what kind of github support should be added to SF tooling? | 04:03 |
---|---|---|
tristanC | it seems like what we should do is uses a tenant_config_script that would automatically add all the project from a github org, like that user doesn't have to change anything in the scheduler config | 04:05 |
*** maeca has quit IRC | 04:35 | |
sfbender | Tristan de Cacqueray created software-factory/sf-ci master: zuul: use new routes https://softwarefactory-project.io/r/11719 | 04:57 |
sfbender | Tristan de Cacqueray created software-factory/sf-docs master: zuul: fix secrets url https://softwarefactory-project.io/r/11720 | 05:01 |
sfbender | Tristan de Cacqueray created software-factory/sf-ci master: zuul: temporary disable status.html checks https://softwarefactory-project.io/r/11722 | 05:16 |
*** flepied__ has quit IRC | 06:19 | |
sfbender | Merged software-factory/sf-ci master: zuul: temporary disable status.html checks https://softwarefactory-project.io/r/11722 | 06:46 |
sfbender | Merged software-factory/sf-config master: zuul: add support for new api routes https://softwarefactory-project.io/r/11701 | 06:50 |
*** flepied__ has joined #softwarefactory | 06:53 | |
*** flepied_ has joined #softwarefactory | 08:00 | |
*** flepied__ has quit IRC | 08:01 | |
*** flepied__ has joined #softwarefactory | 08:02 | |
*** flepied_ has quit IRC | 08:04 | |
*** flepied_ has joined #softwarefactory | 08:08 | |
*** flepied__ has quit IRC | 08:10 | |
*** flepied__ has joined #softwarefactory | 08:14 | |
sfbender | Merged software-factory/sf-ci master: zuul: use new routes https://softwarefactory-project.io/r/11719 | 08:15 |
*** flepied_ has quit IRC | 08:16 | |
sfbender | Merged software-factory/sf-config master: zuul: remove support for old route https://softwarefactory-project.io/r/11703 | 08:17 |
sfbender | Tristan de Cacqueray created software-factory/sf-config master: DNM: test commit https://softwarefactory-project.io/r/11725 | 08:17 |
*** flepied__ has quit IRC | 08:32 | |
*** apevec has joined #softwarefactory | 08:53 | |
sfbender | Merged software-factory/sf-docs master: zuul: fix secrets url https://softwarefactory-project.io/r/11720 | 09:01 |
tristanC | note, we are rolling out 3.0.0 release on the prod, expect a short interruption of zuul-web to apply the new routes | 09:35 |
sfbender | Merged www.softwarefactory-project.io master: Add sprint 2018-12 to sf blog https://softwarefactory-project.io/r/11706 | 09:39 |
sfbender | Merged www.softwarefactory-project.io master: Add 3.0 release https://softwarefactory-project.io/r/11358 | 09:46 |
apevec | congrats to upstream Zuul and SF teams, this is a major milestone! | 09:51 |
tristanC | gundalow: we temporarly removed ansible-network/network-engine-zuul from the scheduler because the "call-ansible-test-sanity" branch set an undefined job | 10:10 |
tristanC | apevec: thanks, sf-3.0 announcements are on their ways :-) | 10:11 |
*** caowei has quit IRC | 10:18 | |
gundalow | tristanC: Thanks for letting me know. | 10:47 |
*** flepied has joined #softwarefactory | 10:48 | |
tristanC | gundalow: also, the zuul-3.0.0 changed the api, the info endpoint is now: https://ansible.softwarefactory-project.io/zuul/api/info | 10:48 |
gundalow | above is just for debug? Just checking I don't need to change any configuration | 10:49 |
tristanC | yes, later it will be used in the web interface directly, until then loading_errors are exposed here | 10:50 |
tristanC | gundalow: right now you need to remove the ansible-test-sanity job from https://github.com/ansible-network/network-engine-zuul/blob/call-ansible-test-sanity/.zuul.yaml | 10:51 |
tristanC | then we'll reload the scheduler with the project included | 10:51 |
gundalow | tristanC: done | 10:54 |
tristanC | gundalow: all good, scheduler reloaded successfully | 10:55 |
gundalow | ansible-test-sanity added back, thanks | 10:56 |
tristanC | gundalow: actually, you can't merge a zuul.yaml that reference a job that doesn't exists | 11:02 |
tristanC | gundalow: you need to only keep the tox-linters job merged, and merge the network-engine-zuul/pull/1 instead when it's working | 11:03 |
gundalow | hum, confused | 11:06 |
*** dsariel has quit IRC | 11:06 | |
gundalow | network-engine-zuul/pull/1 is WIP which defineds roles/ansible-test-sanity | 11:06 |
gundalow | network-engine-zuul contains what I'm testing (ie the Galaxy role) | 11:07 |
tristanC | yes it's fine, it runs the job on the project | 11:08 |
fbo_ | where is defined ansible-test-sanity job ? | 11:08 |
tristanC | the base principle is to only merge change that works, in this case, doing it by the review ensure that 1) the job exists and 2) it actually works on this project | 11:09 |
gundalow | https://github.com/ansible-network/network-engine-zuul/pull/1 adds `jobs: ansible-test-sanity` and has Depends-On: ansible-network/ansible-zuul-jobs#4 | 11:09 |
gundalow | AFAIK what's commited (ie not in PR) network-engine-zuul and ansible-zuul-jobs should pass zuul | 11:10 |
gundalow | the WIP PRs may alternate during passing & failing during their development | 11:11 |
tristanC | gundalow: it can't pass zuul because the ansible-test-sanity job is not merged yet, it's only defined in the jobs/puml/4 | 11:11 |
gundalow | tristanC: it in this context being? | 11:12 |
tristanC | the job* | 11:12 |
gundalow | hum | 11:13 |
gundalow | So how should I structure my PRs then? | 11:13 |
tristanC | the way you did is fine, have the job created in ansible-zuul-job/pull/4 and add it to the project in network-engine-zuul/pull/1 using a depends-on the first pr | 11:14 |
tristanC | but no need merge a zuul.yaml file change before, all can be done through pr | 11:15 |
gundalow | Don't think I've merged anything, ansible-test-sanity only exists/referenced by PRs | 11:18 |
tristanC | gundalow: this https://github.com/ansible-network/network-engine-zuul/blob/call-ansible-test-sanity/.zuul.yaml reference a job that is not merged yet | 11:19 |
gundalow | tristanC: that's a PR | 11:19 |
gundalow | oh, PR in ansible-network, not my fork | 11:20 |
tristanC | gundalow: the branch seems merged :) | 11:20 |
gundalow | yup, user error *cough* | 11:20 |
tristanC | heh that's fine :-) | 11:20 |
*** dsariel has joined #softwarefactory | 11:20 | |
*** dsariel has quit IRC | 11:26 | |
*** dsariel has joined #softwarefactory | 11:26 | |
pabelanger | tristanC: gundalow: yah, I know I've mentioned it a few times, but really think the way to stop broken code from landing, is enable branch protection and remove the ability for users to push code directly to projects zuul is running on. Can we do that on network-engine-zuul for testing? This will avoid the issue for zuul getting wedged | 12:00 |
*** maeca has joined #softwarefactory | 12:01 | |
tristanC | pabelanger: sure, but we also said a few times that zuul shouldn't break when a wrong change is merged | 12:04 |
tristanC | we identified the issue with fbo broken config protection and we'll improve the mechanism where configuration errors will be displayed on the ui instead | 12:04 |
*** maeca has quit IRC | 12:05 | |
fbo_ | it seems that branch protection won't prevent creating a new branch. It is about protecing a specific existing branch | 12:06 |
fbo_ | https://help.github.com/articles/defining-the-mergeability-of-pull-requests/#about-protected-branches | 12:06 |
gundalow | In GitHub I can set that CI must pass, though that's a per branch option. I don't see anyway of making that apply to all current & future branches | 12:06 |
pabelanger | gundalow: why do you need to pass CI? Trying to understand the user case, as zuul is usually used to gate all commits | 12:10 |
pabelanger | fbo_: right, creating of new branches should be done with zuul and automation, part of the discussion from yesterday. Again, to stop things from happening behind the back of zuul | 12:11 |
tristanC | pabelanger: there doesn't seems to be an option in github to restrict branch creation | 12:12 |
pabelanger | tristanC: yah, i think you need to remove the users from admin group. Like we do in gerrit, then add back in when force merging is needed, remove | 12:13 |
pabelanger | gundalow: rcarrillocruz: do you think we can merge https://github.com/ansible-network/ansible-zuul-jobs/pull/5 to start the process of helping increase test coverage on ansible-zuul-jobs? | 12:58 |
fbo_ | pabelanger: finally I remembered that I saw something about branches/and github in the zuul doc and it was here: https://zuul-ci.org/docs/zuul/admin/tenants.html#attr-tenant.exclude-unprotected-branches ... | 13:03 |
fbo_ | so it we wiil need to set known branch as protected and set this option and even PRs that are sourced from the local repo (not the user fork) should not cause trouble | 13:05 |
pabelanger | yes, this is why we want protected branches. Can we do that now? | 13:06 |
fbo_ | we can set that but first it seems that gundalow need to set all ansible-network repos with protected branch | 13:08 |
pabelanger | yah, i think that is an important check we should add when we import a project into zuul main.yaml, check if protected, fail patchset if not | 13:10 |
pabelanger | fbo_: just confirmed with tobiash, we don't see all projects, just ansible-network-zuul for now, it is per project setting | 13:14 |
pabelanger | https://zuul-ci.org/docs/zuul/admin/tenants.html#attr-tenant.untrusted-projects.<project>.exclude-unprotected-branches | 13:15 |
gundalow | pabelanger: I was think about branch protection in the context of stopping bad branched. I don't personally believe we have a need for anyhing more than what Zuul will already offer | 13:16 |
pabelanger | gundalow: do you see a downside of enabling branch protection? | 13:17 |
gundalow | What exactly would you like protecting? | 13:18 |
gundalow | I'm happy or us to test anything on network-engine-zuul, that's what it's there for | 13:18 |
pabelanger | Can't force push to the branch / Can't delete the branch. Basically the items outlined in https://help.github.com/articles/about-protected-branches/ | 13:19 |
pabelanger | that will prevent changes from getting passed zuul and leading to unknown repo states | 13:19 |
pabelanger | but ya, network-engine-zuul only right now | 13:20 |
pabelanger | maybe zuul-config / ansible-zuul-jobs, as they are just zuul jobs for the project | 13:20 |
fbo_ | https://softwarefactory-project.io/r/#/c/11729 | 13:22 |
gundalow | network-engine-zuul & zuul-config updated so that: 1) can't force push 2) can't delete branch | 13:24 |
pabelanger | perfect | 13:24 |
pabelanger | I'd be interested to see how that affects your normal workflow | 13:24 |
pabelanger | and if does, see how we could adapt zuul to it | 13:25 |
gundalow | Shouldn't do, that's what we have on ansible/ansible | 13:26 |
pabelanger | ack | 13:26 |
fbo_ | ok so we are ready to merge the 11729 right ? | 13:28 |
*** flepied_ has joined #softwarefactory | 13:28 | |
fbo_ | gundalow: pabelanger ^ ? | 13:30 |
pabelanger | ideally there would be a way to query the API to check, but ya. think so | 13:31 |
*** flepied has quit IRC | 13:31 | |
gundalow | fbo_: What branches will those check/gates apply to? I want to ensure that it will only affect network-engine-zuul & ansible-zuul-jobs. I think zuul/ansible_networking.yaml needs updating to remove ansible-network/network-config and ansible-network/network-cli | 13:33 |
pabelanger | for 11729 it is just ansible-engine-zuul atm | 13:38 |
fbo_ | gundalow: repos listed in ansible_networking.yaml are loaded by zuul but the new setting only affect ansible-engine-zuul yes | 13:39 |
gundalow | cool | 13:39 |
gundalow | in that case I'll give your PR a shipit and zuul will merge | 13:39 |
fbo_ | yes you can +1 and I'll +2 to have it merged | 13:40 |
gundalow | I've deleted the call-ansible-test-sanity branch, and will create that as a PR from my fork. Doing a recheck on ansible-zuul-jobs/pull/5 | 13:48 |
pabelanger | ack | 13:51 |
gundalow | green and zuul merged, great | 13:51 |
pabelanger | yah | 13:53 |
fbo_ | nice | 13:54 |
pabelanger | fbo_: updated comment on 11729 since ansible-zuul-jobs has branch protection too | 13:56 |
pabelanger | lets also enable it for zuul | 13:56 |
fbo_ | pabelanger: zuul-config you mean ? | 14:01 |
pabelanger | no, I think gundalow only did ansible-zuul-jobs and ansible-engine-zuul | 14:02 |
*** yassine has quit IRC | 14:04 | |
gundalow | I can do the same for zuul-config if you'd like | 14:04 |
*** yassine has joined #softwarefactory | 14:05 | |
pabelanger | lets do these 2 first, confirm then do zuul-config | 14:05 |
*** flepied__ has joined #softwarefactory | 14:07 | |
fbo_ | pabelanger: ok I just +2 let wait for the reconfiguration to happen | 14:09 |
*** flepied_ has quit IRC | 14:09 | |
fbo_ | done | 14:13 |
pabelanger | k | 14:14 |
*** flepied_ has joined #softwarefactory | 14:30 | |
*** maeca has joined #softwarefactory | 14:33 | |
*** flepied__ has quit IRC | 14:33 | |
*** caowei has joined #softwarefactory | 15:08 | |
*** caowei has quit IRC | 15:45 | |
*** dsariel has quit IRC | 16:12 | |
*** flepied__ has joined #softwarefactory | 16:25 | |
*** flepied_ has quit IRC | 16:27 | |
*** flepied__ has quit IRC | 17:22 | |
*** flepied__ has joined #softwarefactory | 18:01 | |
*** flepied__ has quit IRC | 18:47 | |
pabelanger | gundalow: see dib-fedora-25 is a node for SF, that might get you python3 | 18:50 |
pabelanger | but since fedora-25 is EOL, we really should be using fedora-27 | 18:51 |
pabelanger | nhicher: fbo_: tristanC: any change we can bring a fedora-27 online? | 18:51 |
pabelanger | chance* | 18:51 |
gundalow | 2.7 and one of more of 3.{5,6,7} would be good | 18:52 |
gundalow | As I said, may end up longer term building our own images with tooling in to save on runtime install of libraries, though that's something to consider later on | 18:52 |
gundalow | Thanks :) | 18:58 |
gundalow | erm, how come | 19:01 |
gundalow | erm, how come https://github.com/ansible-network/ansible-zuul-jobs/pull/4 got merged, I don't see a +1/GH review | 19:01 |
*** flepied__ has joined #softwarefactory | 19:08 | |
gundalow | That's really confusing | 19:11 |
pabelanger | looking | 19:20 |
nhicher | pabelanger: I can prepare a fedora 27 image pabelanger | 19:20 |
pabelanger | nhicher: can you pull up the scheduler logs on pull/4 and see why gate ran?^ | 19:22 |
pabelanger | the github Ui makes it hard to see votes | 19:22 |
nhicher | pabelanger: I'm not sure ansible-network use image provides by sf, but an image rcarrillocruz provides few day ago | 19:23 |
rcarrillocruz | Is a WIP, more changes to that image are needed before putting int tests | 19:23 |
pabelanger | yah | 19:24 |
pabelanger | gundalow: mattclay: possible somebody hit approved on that PR? Like I said, I am not used to the UX for github | 19:24 |
nhicher | ok, thanks rcarrillocruz, I will prepare a fedora-27 | 19:24 |
pabelanger | need access to zuul logs | 19:25 |
pabelanger | https://github.com/ansible-network/zuul-config/blob/master/zuul.d/pipelines.yaml#L53 | 19:25 |
pabelanger | would mean approved is need to enqueue into gate | 19:25 |
mattclay | pabelanger: There are no approvals for that PR on GH. | 19:28 |
pabelanger | mattclay: thanks, so yah need to look at zuul logs now. | 19:29 |
nhicher | pabelanger: https://softwarefactory-project.io/paste/show/1014/ | 19:33 |
pabelanger | nhicher: is there anything above for 0x7f9168540f98 ? | 19:34 |
nhicher | pabelanger: https://softwarefactory-project.io/paste/show/1015/ | 19:36 |
nhicher | human log search, that's weird =) | 19:36 |
pabelanger | nhicher: should see something like | 19:36 |
pabelanger | zuul.Pipeline.ansible-network.gate: Adding change | 19:36 |
pabelanger | want to see the events / match logic above that | 19:37 |
pabelanger | Hmm, I think I see the issue | 19:39 |
pabelanger | https://github.com/ansible-network/zuul-config/blob/master/zuul.d/pipelines.yaml#L51 | 19:39 |
nhicher | https://softwarefactory-project.io/paste/show/1016/ | 19:39 |
pabelanger | I think we want: http://paste.openstack.org/show/717943/ | 19:39 |
pabelanger | which applys the approved label | 19:40 |
pabelanger | that is missing | 19:40 |
pabelanger | and other 2 conditions were true | 19:40 |
pabelanger | so jobs enqueued into gate | 19:40 |
pabelanger | mattclay: gundalow: can you propose a revert of PR/4 | 19:41 |
pabelanger | I'll push up a pipeline change | 19:41 |
mattclay | pabelanger: https://github.com/ansible-network/ansible-zuul-jobs/pull/7 | 19:52 |
pabelanger | thanks | 19:53 |
*** flepied__ has quit IRC | 20:21 | |
*** flepied has joined #softwarefactory | 20:22 | |
pabelanger | mattclay: gundalow: haven't given up, just working to enable more debug logs on zuul to see what is happening | 20:23 |
*** flepied has quit IRC | 21:04 | |
*** dsariel has joined #softwarefactory | 21:05 | |
*** apevec has quit IRC | 21:42 | |
*** dsariel has quit IRC | 22:14 | |
*** nijaba has joined #softwarefactory | 22:46 | |
*** nijaba has quit IRC | 22:54 | |
*** nijaba has joined #softwarefactory | 22:57 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!