I've been running an inclusive set of tests on branch_5x to do what I
can for the release effort.  It kept failing with OOME, so I kept
increasing the heap size. After trying 2GB and 3GB, I finally bumped it
all the way to 8GB and dropped the JVM count to 1, but that resulted in
different problems.  Here's the commandline that I used, followed by the
list of failures:

ant -Dtests.jvms=1 -Dtests.heapsize=8g -Dtests.nightly=true
-Dtests.weekly=true -Dtests.monster=true clean test | tee ~/b5x-testlog.txt

   [junit4] Tests with failures:
   [junit4]   - org.apache.lucene.index.Test2BTerms (suite)
   [junit4]   - org.apache.lucene.index.Test2BNumericDocValues.testNumerics
   [junit4]   - org.apache.lucene.index.Test2BNumericDocValues (suite)
   [junit4]   -
org.apache.lucene.index.Test2BSortedDocValues.testFixedSorted
   [junit4]   - org.apache.lucene.index.Test2BSortedDocValues.test2BOrds
   [junit4]   - org.apache.lucene.index.Test2BSortedDocValues (suite)
   [junit4]
   [junit4]
   [junit4] JVM J0:     0.90 .. 76575.00 = 76574.10s
   [junit4] Execution time total: 21 hours 16 minutes 15 seconds

All of them except for Test2BTerms failed because of this problem:

   [junit4]    > Throwable #1: java.lang.IllegalStateException: number
of documents in the index cannot exceed 2147483519

Test2BTerms failed for an entirely different reason:

   [junit4]   2> NOTE: reproduce with: ant test  -Dtestcase=Test2BTerms
-Dtests.seed=9F2773FB226B1E02 -Dtests.nightly=true -Dtests.weekly=true
-Dtests.slow=true -Dtests.locale=es_PE
-Dtests.timezone=America/Los_Angeles -Dtests.file.encoding=UTF-8
   [junit4] ERROR   0.00s | Test2BTerms (suite) <<<
   [junit4]    > Throwable #1: java.lang.AssertionError: The test or
suite printed 3012118 bytes to stdout and stderr, even though the limit
was set to 8192 bytes. Increase the limit with @Limit, ignore it
completely with @SuppressSysoutChecks or run with -Dtests.verbose=true
   [junit4]    >        at
__randomizedtesting.SeedInfo.seed([9F2773FB226B1E02]:0)
   [junit4]    >        at java.lang.Thread.run(Thread.java:745)

I'm clueless about how to fix the number of documents going too high.  I
could probably fix the other one, if someone can tell me what the
preferred fix is.

I haven't tried this on the 4_10 branch, because it takes so long to
run.  I've started a similar commandline in branch_5x/solr to see what
happens.

Thanks,
Shawn


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to