hector-client/hector

Tests run: 286, Failures: 4, Errors: 1, Skipped: 2 :(

Opened this issue · 6 comments

The-Alchemist commented

I just downloaded the latest code, and I'm getting unpredictable failures during the maven test phase.

Results :

Failed tests:

Tests in error:

Tests run: 286, Failures: 4, Errors: 1, Skipped: 2

20:49:34,082  INFO CassandraDaemon:218 - Stop listening to thrift clients
20:49:34,087  INFO MessagingService:539 - Waiting for messaging service to quiesce
20:49:34,091  INFO MessagingService:695 - MessagingService shutting down server thread.
[INFO] ------------------------------------------------------------------------
[INFO] Reactor Summary:
[INFO]
[INFO] hector ............................................ SUCCESS [2.394s]
[INFO] test .............................................. SUCCESS [4.998s]
[INFO] hector-core ....................................... FAILURE [4:42.328s]
[INFO] hector-object-mapper .............................. SKIPPED
[INFO] ------------------------------------------------------------------------
[INFO] BUILD FAILURE
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 4:52.176s
[INFO] Finished at: Sat Nov 03 20:49:34 EDT 2012
[INFO] Final Memory: 30M/335M
[INFO] ------------------------------------------------------------------------

Even worse, I can re-run this and get more, or less, failures. (I just re-run this and got Failures: 3, Errors: 1, Skipped: 2).

More directly, I'm working on pull request 544 to remove log4j as a dependency and I'm trying to get all tests to pass.

I'm gonna attach the surefire reports, but here is the quickie summary:

  • testSimpleSelect(me.prettyprint.cassandra.model.CqlQueryTest)
  • testQuerySingleSubColumnExtractSuper(me.prettyprint.cassandra.service.template.SuperCfTemplateTest)
  • testTimeUUIDWithClockResolution(me.prettyprint.cassandra.utils.TimeUUIDUtilsTest)

Any ideas? Thanks, guys!

The-Alchemist commented

You can download the surefire-reports as a 7z archive or as a ZIP archive

The-Alchemist commented

I think I figured out why this happens: Java 7 doesn't give a predictable order to Class.get*Methods(), which can break JUnit tests that depend on running in a certain order.

This affects those test classes that query and modify the database in different methods without cleaning, like CqlQueryTest and SuperCfTemplateTest.

I'm willing to fix up the tests. I see a few possible approaches:

  1. Update the tests to clean up their own data after each method (in a @After method in BaseEmbededServerSetupTest, for example)
  2. Make the test order explicit via mechanisms like a custom JUnit-runner as described at http://intellijava.blogspot.com/2012/05/junit-and-java-7.html

What approach would most likely be accepted?

pauloricardomg commented

I'm still having this issue. Any fix for this? Maybe switching to java 6?

patricioe commented

It works for java 6. Sometime this weekend I can try to give java 7 a try
and see what test breaks.
On Jun 6, 2013 4:49 PM, "Paulo Ricardo Motta Gomes" <
notifications@github.com> wrote:

I'm still having this issue. Any fix for this? Maybe switching to java 6?

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/546#issuecomment-19081507
.

lewismc commented

This is the issue we are experiencing within current builds 1.1-5 a year on from when it was reported.
I am going to take this issue on tonight and see if I can get anywhere.
I suggest to take The-Alchemist's 1st proposal as IMHO cleaning up individual test output (or at least truncating the data) is good practice anyway.
I'll create a branch and hopefully get a pull request fired off soon.
Thanks
Lewis

zznate commented

SGTM - thanks again @lewismc