poi-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dominik Stadler <dominik.stad...@gmx.at>
Subject Re: Builds changes
Date Sun, 27 Nov 2016 21:42:58 GMT
Hi,

I started with defining the Jenkins jobs as part of the source via the
Jenkins Job DSL plugin. This is a very nice way to better control the
Jenkins jobs as well as let's us re-use things across multiple jobs. Until
now we often had to manually adjust every job for something, e.g. the list
of "usable" slaves. Now you can describe the Jenkins Setup in a nice
groovy-based language.

See http://svn.apache.org/repos/asf/poi/trunk/jenkins for the script and
https://github.com/jenkinsci/job-dsl-plugin/wiki for documentation of the
Job.

For now I just added one test-job, but I would like to replace all current
Jenkins jobs with generated ones, just keeping the "seed" job manually set
up (but rarely changed at all!)


Dominik.


On Mon, Nov 14, 2016 at 8:23 PM, Dominik Stadler <dominik.stadler@gmx.at>
wrote:

> I saw the mails, but did not see anything affecting our jobs a lot
>
> I would also like to experiment with the job dsl plugin, it sounds much
> better to have this config in the SCM, especially with the constant missing
> JDKs on the slaves and other changes that affect all our jobs.
>
> Dominik
>
> On Nov 14, 2016 5:29 PM, "Javen O'Neal" <onealj@apache.org> wrote:
>
>> For developers not subscribed to the builds mailing list, there is a good
>> discussion going on that affects POI.[1]
>>
>> Jobs timeouts have been reduced to 120 minutes. We can revert this if
>> needed (like the slow executors).
>>
>> Job history has been reduced to 5. If we have a streak of broken builds,
>> it's helpful to know the last successful build. I don't know if the new
>> pruning rules will keep that kind of information.
>>
>> Someone suggested managing our jenkins config in a svn-controlled file. I
>> like this idea and think POI should adopt it.
>>
>> Please check out the other changes that we "made with a hammer" to see if
>> we need to revert or compromise.
>>
>> [1]
>> https://lists.apache.org/thread.html/f0f38645fed334b0858e764
>> 76a8076fe3fae86355e14d162edd4268f@
>> <builds.apache.org>
>>
>

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