Thanks for working on this! I have noticed that tests run for new PRs and force-pushed commits, but if a test fails due to a flake I am unable to re-run it (ex: "Run Java PreCommit"). PR that has this issue: https://github.com/apache/beam/pull/10369. Is this intended behaviour?
- Kirill On Tue, Jan 14, 2020 at 3:20 PM Luke Cwik <[email protected]> wrote: > Does the approval list live beyond the lifetime of the jenkins machine (my > initial impression is that the approval list disappears on Jenkins machine > restart)? > > Also, I imagine that ASF wants an explicit way to see who is approved and > who is denied which the plugin doesn't seem to allow. > > On Tue, Jan 14, 2020 at 3:11 PM Pablo Estrada <[email protected]> wrote: > >> I've merged https://github.com/apache/beam/pull/10582 to unblock >> existing contributors that are having trouble getting their PRs tested >> without committer help. We can discuss Kai's suggestion. >> >> Looking at https://github.com/jenkinsci/ghprb-plugin, it seems like the >> 'add to whitelist' comment adds contributors permanently to a whitelist. >> This would have more immediate results than the .asf.yaml file. It would be >> harder to track who has the privilege, but it doesn't sound like that >> concerns us, right? >> >> Thoughts from others? >> -P. >> >> On Tue, Jan 14, 2020 at 1:43 PM Kai Jiang <[email protected]> wrote: >> >>> Nice! I took a look at Beam Jenkins job properties ( >>> CommonJobProperties.groovy#L108-L111 >>> <https://github.com/apache/beam/blob/54e610809c87bdfba6ab94a8462e513ae17936e3/.test-infra/jenkins/CommonJobProperties.groovy#L108-L111>) >>> and it uses jenkinsci/ghprb-plugin >>> <https://github.com/jenkinsci/ghprb-plugin>. >>> It should support the feature of comment add to whitelist from >>> committer on PR for adding new contributors to whitelist. >>> Adding github account to asf yaml might be a little heavy if this >>> approach works. Could we also test on this method? >>> >>> Best, >>> Kai >>> >>> >>> On Tue, Jan 14, 2020 at 1:16 PM Pablo Estrada <[email protected]> >>> wrote: >>> >>>> I've added all the PR authors for the last 1000 merged PRs. I will >>>> merge in a few minutes. I'll have a follow up change to document this on >>>> the website. >>>> >>>> On Tue, Jan 14, 2020 at 11:29 AM Luke Cwik <[email protected]> wrote: >>>> >>>>> Should we scrape all past contributors and add them to the file? >>>>> >>>>> On Tue, Jan 14, 2020 at 11:18 AM Kenneth Knowles <[email protected]> >>>>> wrote: >>>>> >>>>>> Nice! This will help at least temporarily. We can see if it grows too >>>>>> unwieldy. It is still unfriendly to newcomers. >>>>>> >>>>>> Kenn >>>>>> >>>>>> On Tue, Jan 14, 2020 at 11:06 AM Pablo Estrada <[email protected]> >>>>>> wrote: >>>>>> >>>>>>> Hi all, >>>>>>> ASF INFRA gave us a middle-ground sort of workaround for this by >>>>>>> using .asf.yaml files. Here's a change to implement it[1], and >>>>>>> documentation for the .asf.yaml file[2], as well as the relevant section >>>>>>> for our case[3]. >>>>>>> >>>>>>> I'll check the docs in [2] well before pushing to merge, just to be >>>>>>> sure we're not breaking anything. >>>>>>> >>>>>>> [1] https://github.com/apache/beam/pull/10582 >>>>>>> [2] >>>>>>> https://cwiki.apache.org/confluence/display/INFRA/.asf.yaml+features+for+git+repositories >>>>>>> >>>>>>> [3] >>>>>>> https://cwiki.apache.org/confluence/display/INFRA/.asf.yaml+features+for+git+repositories#id-.asf.yamlfeaturesforgitrepositories-JenkinsPRWhitelisting >>>>>>> >>>>>>> On Mon, Jan 13, 2020 at 3:29 PM Luke Cwik <[email protected]> wrote: >>>>>>> >>>>>>>> I'm for going back to the status quo where anyone's PR ran the >>>>>>>> tests automatically or to the suggestion where users marked as >>>>>>>> contributors >>>>>>>> had their tests run automatically (with the documentation update about >>>>>>>> how >>>>>>>> link your github/jira accounts). >>>>>>>> >>>>>>>> On Mon, Jan 13, 2020 at 2:45 AM Michał Walenia < >>>>>>>> [email protected]> wrote: >>>>>>>> >>>>>>>>> Hi, >>>>>>>>> I wanted to decouple the conversation about solutions to the issue >>>>>>>>> from job execution requests. >>>>>>>>> We have 131 open PRs right now and 64 committers with job running >>>>>>>>> privileges. From what I counted, more than 80 of those PRs are not >>>>>>>>> authored >>>>>>>>> by committers. >>>>>>>>> I think that having committers answer testing and retesting >>>>>>>>> requests is a temporary solution and a permanent one should be >>>>>>>>> decided upon >>>>>>>>> soon. While it's an inconvenience for contributors familiar with the >>>>>>>>> workings of the project and the community, newcomers might be put off >>>>>>>>> by >>>>>>>>> the fact that the tests don't run automatically on their pull requests >>>>>>>>> (this is an industry standard which IMO should be upheld also in >>>>>>>>> Beam). The >>>>>>>>> barrier of finding one of committers who is active and willing to >>>>>>>>> trigger >>>>>>>>> their tests can make the entry to the project more difficult. >>>>>>>>> >>>>>>>>> I believe that the solution proposed by Kenneth in the Jira thread >>>>>>>>> https://issues.apache.org/jira/browse/INFRA-19670 is viable. The >>>>>>>>> questions are: do we want to implement this idea and what needs to be >>>>>>>>> done >>>>>>>>> to do it? >>>>>>>>> >>>>>>>>> Regards >>>>>>>>> Michał >>>>>>>>> >>>>>>>>> -- >>>>>>>>> >>>>>>>>> Michał Walenia >>>>>>>>> Polidea <https://www.polidea.com/> | Software Engineer >>>>>>>>> >>>>>>>>> M: +48 791 432 002 <+48791432002> >>>>>>>>> E: [email protected] >>>>>>>>> >>>>>>>>> Unique Tech >>>>>>>>> Check out our projects! <https://www.polidea.com/our-work> >>>>>>>>> >>>>>>>>
