excalibur-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Carsten Ziegeler <cziege...@apache.org>
Subject Re: [GUMP@vmgump]: Project excalibur-pool-impl-test (in module excalibur) failed
Date Tue, 29 Dec 2009 10:22:42 GMT
Leo Simons wrote:
> Hey folks,
> 
> On 12/23/09 7:24 AM, Gump wrote:
>> DEBUG - Put a org.apache.avalon.excalibur.pool.test.PoolableTestObject
>> back into the pool.
>> DEBUG - Put a org.apache.avalon.excalibur.pool.test.PoolableTestObject
>> back into the pool.
>> DEBUG - Got a org.apache.avalon.excalibur.pool.test.PoolableTestObject
>> from the pool.
>> Tests run: 1001, Failures: 0, Errors: 1, Skipped: 0, Time elapsed:
>> 3.718 sec<<<  FAILURE!
>> Running
>> org.apache.avalon.excalibur.pool.test.ResourceLimitingPoolMultithreadMaxStrictBlockTestCase
>>
>> TimedTest (WAITING): LoadTest (NON-ATOMIC): ThreadedTest:
>> testGetPut(org.apache.avalon.excalibur.pool.test.ResourceLimitingPoolMultithreadMaxStrictBlockTestCase)(repeated):
>> 1755 ms
>> Tests run: 100, Failures: 0, Errors: 0, Skipped: 0, Time elapsed:
>> 1.806 sec
>> Running
>> org.apache.avalon.excalibur.pool.test.ResourceLimitingPoolMultithreadMaxStrictTestCase
>>
> 
> This has been broken since at least October 9th according to my e-mail.
> I took a look at the tests in question, they're (perhaps necessarily)
> somewhat fragile tests. I suspect a combination of fast hardware + JDK 6
> causes the test failure.
> 
> I also somewhat suspect the bug is in the test not in the actual code
> (though the actual code itself does look a bit scary -- not really how
> I'd write this in 2009 :-D ).
> 
> I tried jiggling timers and such a bit and making some small changes to
> the test but I only seem to make it worse (can get deadlocks, even).
> 
> Anyone around that knows this code? (The SVN history tells us that it
> didn't see meaningful changes since the CVS -> SVN conversion in 2003...)
> 
> Any ideas?
> 
> ...if I don't hear anything I'm going to comment out some tests or move
> them to a different maven profile so they don't run out-of-the-box.
> 
+1 :)

I think those tests are failing for a long time (usually they pass if
invoked several times) :) I looked into them a very long time ago, but
came to the same conclusion as you (bug in the test)

Carsten
-- 
Carsten Ziegeler
cziegeler@apache.org

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@excalibur.apache.org
For additional commands, e-mail: dev-help@excalibur.apache.org


Mime
View raw message