> On Oct 25, 2015, at 12:05 AM, jun aoki <[email protected]> wrote:
>
> Hi Yetus,
>
> I'm trying to run test-patch for Geode patches on builds.a.o.
> On my ubuntu, this is how I layout the project directories
> .
> ├── incubator-geode
> └── yetus
This is exactly how I test and develop on my laptop. :D
> On builds.a.o, I'm thinking to setup Jenkins' Multiple SCMs plugin so that
> I can have the same directory layout, and the same command in Execute Shell
> should work.
Andrew Bayer suggested against doing multiple SCMs due to the load,
especially against the apache.org servers.
> Or, as you guys have been discussing, I'm speculating we'd have a release
> artifact of yetus like rpm, and publicly available and so that we can yum
> install it, thus I don't have to git clone for yetus....
>
> You guys have some recommendations?
I think we’ll end up with a tar ball relatively pretty soon. Sean is
blasting through those issues, especially thanks to the ‘nobuild’ plug-in.
Things will still be in a state of flux, but at least we’ll have a common
starting point.
FWIW, right now for Hadoop, I’m doing:
1. Let Jenkins extract Hadoop.
2. In the patch process directory, download and extract a yetus tarball
from github.
3. Pass in parameters as necessary
=========
PATCHPROCESS=${WORKSPACE}/patchprocess
mkdir -p ${PATCHPROCESS}
### Download Yetus
cd ${PATCHPROCESS}
curl -L https://api.github.com/repos/apache/yetus/tarball/YETUS-83 >
yetus.tar.gz
tar xvpf yetus.tar.gz
cd ${WORKSPACE}
/bin/bash ${WORKSPACE}/patchprocess/apache-yetus-*/dev-support/test-patch.sh \
--jenkins \
--patch-dir=${PATCHPROCESS} \
--basedir=${WORKSPACE} \
--project=hadoop \
--skip-dir=dev-support \
--mv-patch-dir \
--jira-password="${JIRA_PASSWORD}" \
--jira-user=hadoopqa \
--findbugs-strict-precheck \
--findbugs-home=/opt/findbugs \
--skip-dirs=dev-support \
--mvn-custom-repos \
--docker \
--multijdkdirs="/usr/lib/jvm/java-8-oracle" \
HADOOP-${ISSUE_NUM}
===
… which reminds me, I need to fix some of the more ridiculous Docker
bugs soon …
