Github user JoshRosen commented on a diff in the pull request:
https://github.com/apache/spark/pull/3121#discussion_r20400528
--- Diff: pom.xml ---
@@ -978,6 +978,7 @@
<spark.testing>1</spark.testing>
<spark.ui.enabled>false</spark.ui.enabled>
<spark.executor.extraClassPath>${test_classpath}</spark.executor.extraClassPath>
+
<spark.driver.allowMultipleContexts>true</spark.driver.allowMultipleContexts>
--- End diff --
Yeah, I think that we'd eventually want to have this be `false`, but there
are a few test suites for which this might be difficult (such as SparkSQL) and
I didn't want to undertake a huge refactoring in this PR. If you SSH into
Jenkins and look at the `unit-tests.log` files, you'll notice that this PR
introduces a very small number of warning log messages in the tests, since most
suites properly clean up their contexts. I did manage to fix a few of the
obvious cases that I found, so this PR does contain a bit of test refactoring.
I'll open a JIRA to finish this cleanup later.
By the way, if you look in `SparkContextSuite.scala`, there's a test
fixture that changes system properties in individual unit tests (and resets
them after the test); I use this to ensure that the property / configuration
option works as expected.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
