hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From 张铎(Duo Zhang) <palomino...@gmail.com>
Subject Re: [DISCUSS] GitHub PRs without JIRA number
Date Tue, 10 Sep 2019 14:20:58 GMT
Actually the job for testing PR is here...

https://builds.apache.org/job/hadoop-multibranch/

I've added the 'Change requests' option(seems I have the permission...),
and then clicked the 'Scan Now', the job for PR-1404 has been scheduled

https://builds.apache.org/job/hadoop-multibranch/view/change-requests/job/PR-1404/


The infra team has shutdown the github webhook on scheduling builds
automatically when there are new PRs or new updates to existing PRs,
because this violate the rule that only committers can trigger jenkins
jobs, so we need to click the 'Scan Now' button to trigger builds. And it
is possible to schedule a periodical job to do the scan, for HBase the
interval is 10 minutes, and for hadoop, there are too many branches and
PRs, the job is still running while I'm writing this email, so I think
maybe we should use a greater interval, maybe 20 or 30 minutes?

Thanks.

Steve Loughran <stevel@cloudera.com> 于2019年9月10日周二 下午7:36写道:

>
>
> On Tue, Sep 10, 2019 at 9:07 AM 张铎(Duo Zhang) <palomino219@gmail.com>
> wrote:
>
>> Nits: You can change the jenkins job config to not trigger pre commit
>> build
>> for stale PRs if only the base branch is changed. By default, either the
>> PR
>> itself changed, or the base branch is changed, the branch sources plugin
>> will both trigger a build. You can add a Change requests option, and
>> select
>> 'Ignore rebuilding merge branches when only the target branch changed'. So
>> the stale PRs will not waste jenkins build resources any more.
>>
>> +1 for this. If old PRs want to be tested by their creator, they can
> rebase. Having a way to ask jenkins to do it on the PR lets other.
>
>
>> And on retesting a PR, just go to the pipeline jenkins job, find the
>> related job for the PR, and click build manually.
>>
>>
>>
>
>    1. I like the command approach. Spark has this, and some gerrit
>    pipelines do.
>    2. I've just tried to do this on a PR (
>    https://github.com/apache/hadoop/pull/1404) but can't see how to. The
>    precommit hdfs job only takes the JIRA number and looks for an attachment,
>    judging by the way it is failing to merge things in
>    https://builds.apache.org/view/PreCommit%20Builds/job/PreCommit-HDFS-Build/27830/
>
> Is there another build or some trick I should use?
> Ni
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message