Dongjoon Hyun created SPARK-14765:
-------------------------------------
Summary: Jenkins should run tests only based on the PR contents
Key: SPARK-14765
URL: https://issues.apache.org/jira/browse/SPARK-14765
Project: Spark
Issue Type: Bug
Components: Project Infra
Reporter: Dongjoon Hyun
h5. Background
Yesterday, when I exposed a function to Python/R, only Python and R module are
changed. At the first successful build, it took 25 minutes, but the next one
took 125 minutes.
* [25
minutes|https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/56291/consoleFull]
{code}
[info] Found the following changed modules: pyspark-sql, sparkr
{code}
* [125
minutes|https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/56298/]
{code}
[info] Found the following changed modules: pyspark-sql, sparkr, mllib, hive,
sql, root
{code}
h5. Problem
`identify_changed_files_from_git_commits` function in `run-tests.py` simply
runs `git diff` to the master. It means **Newly updated master code** also
considered as a changed file in the PR.
h5. How to fix
Add `git rebase` command before running `git diff`.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
