I am done with my research , we have a skip ci phrase in the jenkins and it works as expected. eg https://github.com/apache/pulsar/pull/5244.
The issue is github is configured to expect a response always, we can consider changing that. But this will open the repo to potential problem. As there will no enforcement of the ci check. I looked into adding a pipeline step to use git diff to check if a pr should proceed with the ci job but I am unable to make it work as expected in jenkins. The best course of action is simply to move the site folder to a new repo. -Ali On Thu, Sep 19, 2019 at 11:38 AM Matteo Merli <matteo.me...@gmail.com> wrote: > Another viable option could be to move the site to a separate > "pulsar-site" git repo. > That would again require some work, but it would be a cleaner > long-term approach. > > -- > Matteo Merli > <matteo.me...@gmail.com> > > On Wed, Sep 18, 2019 at 4:38 PM Sijie Guo <guosi...@gmail.com> wrote: > > > > +1 > > > > The later approach seems to be better to me. Please remember syncing the > > jenkins job configuration back to the jenkins job configurations. > > https://github.com/apache/pulsar/tree/master/.test-infra/jenkins > > And make sure all the changes made to Jenkins are under proper review. > > > > Thanks, > > Sijie > > > > On Thu, Sep 19, 2019 at 5:17 AM Ali Ahmed <alahmed...@gmail.com> wrote: > > > > > Currently there has been a growth in PR for documentation changes, but > it's > > > being slowed by the test run failures, we can relieve the issue, with > some > > > small changes. > > > > > > either not run the test if the git can't find a code change in files > > > git diff --name-only HEAD master | grep -v site2 > > > > > > or change the jenkins job to only look at specific folders for changes > > > steps to guard against > > > unnecessary test runs. > > > > > > git diff --name-only HEAD master | grep -v site2stages { > > > stage('CI Run') { > > > when { changeset "src/*"} > > > steps { > > > sh "make build-nginx" > > > sh "make start-nginx" > > > } > > > } > > > > > > We only to continue to improve test stability in parallel. > > > >
