db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Knut Anders Hatlen <knut.hat...@oracle.com>
Subject Re: Largedata test history
Date Fri, 01 Jul 2011 13:49:04 GMT
Kathey Marsden <kmarsdenderby@sbcglobal.net> writes:

> On 7/1/2011 4:44 AM, Knut Anders Hatlen wrote:
>> Kathey Marsden<kmarsdenderby@sbcglobal.net>  writes:
>>
>>> Myrna noticed that the IBM largeData test results were truncated,
>>> probably because they did not finish and were interrupted by the next
>>> day's run.  I don't have test history, so am not sure if these tests
>>> have become slower or just actually never finished.
>>>
>>> I noticed the last posted results for the Sun JVM were for revision
>>> 835496 which was in the 10.6 timeframe.
>>> http://dbtg.foundry.sun.com/derby/test/trunk15/largedata/testing/Limited/testSummary-current.html
>>>
>>> At that time, the duration was 242.62% of revision 433306 which was in
>>> the 10.3 timeframe.
>>>
>>> I was wondering
>>> 1) Does anyone remember if we understood the duration difference with
>>> revision 835496?
>>> 2)  Is there a more comprehensive history available?
>>> 3) Are there more recent statistics?
>> I cannot find any newer results than revision 835496. Looking at
>> whatever is left of logs from these runs I see this:
>>
>> ,----
>> | Built on SunOS 5.10 i386 i86pc[siv02.norway.sun.com] at 2006-08-21 19:47:02
>> |
>> | Suite   Tests   Passed  Failed  Skippd  Duration
>> | largeData       4       4       0       0       04:27:07
>> |
>> | Started:        2006-08-21 23:30:04     End:    2006-08-22 03:57:15     Host: 
 siv39.norway.sun.com
>> `----
>>
>> and
>>
>> ,----
>> | Built on SunOS 5.10 i386 i86pc[siv02.norway.sun.com] at 2009-11-12 19:47:04
>> |
>> | Suite   Tests   Passed  Failed  Skippd  Duration
>> | largeData       1       1       0       0       10:48:06
>> |
>> | Started:        2009-11-13 02:00:04     End:    2009-11-13 12:48:17     Host: 
 nanna18.norway.sun.com
>> `----
>>
>> So it looks like the 433306 run was on a different machine than the
>> 835496 run, which might explain the difference in duration. I have never
>> run this test suite myself, so I don't know what it means that one of
>> them reports four tests passed and the other one just one test.
>>
> Thank you Knut for the information. I think I will just start from
> scratch then and give the test a try on trunk, 10.5 and 10.3 on the
> same machine to see where we are.  I ran it on trunk last night and it
> took just over ten hours.  Embedded passed. There was just one test
> skipped in the skip file for client:
> largeData/largeData.skip:largeData:largeDataNet:largeDataTests
> but the .sum file showed Framework: DerbyNet Cannot run the suite,
> framework driver not found.  Maybe the test number differences were
> related to whether JCC was available to the run.

Maybe. But it's still very strange that running four tests in 2006 was
more than twice as fast as running one test in 2009. Thanks for trying
to solve this mystery!

-- 
Knut Anders

Mime
View raw message