Ergonomic, hermetic, user-friendly software build system for Python, Go, Java, Kotlin, Protobuf, Scala, Shell. Pants lets you fearlessly scale up your codebase!

Pants

So, in regards to the Jenkins failure - Travis passed and that test passes locally. Have you seen that be a flaky test <@U04S45AHA>? ...it doesn't look flaky...

travis: <https://travis-ci.org/pantsbuild/pants/builds/129236704>
jenkins: <http://jenkins.pantsbuild.org/job/pantsbuild/job/pants/branch/PR-3375/lastBuild/consoleFull>
test: tests/python/pants_test/core_tasks:prep_command_integration

okay...will dig then. I suspect jenkins config related though - this test puts sentinel files in /tmp

The node with the failure is here: <http://jenkins.pantsbuild.org/job/pantsbuild/job/pants/branch/PR-3375/lastBuild/execution/node/268/log/>

I don't think you have vault access, let me ssh in and see whats up with /tmp

... but backing up, until we're using jenkins for real, it probably doesn't make sense to block on it for real if there is good evidence its a jenkins-specific bork

Right - that node is dead to save us money

good - I was hoping someone would say that

Well, reading that test it is not sharding safe

Its broken for the case where you have 2 pants CI on the same machine, using pytest sharding

So I think this is a low-probability hit rate bug

<https://github.com/pantsbuild/pants/issues/3383> - I cc'd <@U053H6LQS>

thanks for looking into that for me! I am going to proceed with the 1.1.0-pre0 release then.

edited: Do I release to pypi first or push to master?

<@U04S45AHA>: Tempted to flip to Jenkins as the canonical CI at this point

Whenever we see a "jenkins only" bug at this point it tends to be a bug in our tests rather than Jenkins causing issues

git plugin issues still exist.  I see failed to fetch errors still,  like this: <http://jenkins.pantsbuild.org/job/pantsbuild/job/pants/branch/master/12/console>

Jenkins seems good but the git plugin has been tricky - there was a real crazy bug last night with non-existant SHAs

missing files which seemed to be attributed to git issues

<@U0607UFLH>: The non-existent SHAs I've seen as a result of force pushes to branches, I think, and also with the whole "merge master then run" behavior