[#JDO-438] More Derby "ERROR 40XL1" issues with multithreaded tests

JDO

More Derby "ERROR 40XL1" issues with multithreaded tests

Created: 20/Oct/06 05:37 PM Updated: 08/Jan/07 07:03 PM

Return to search

Component/s:

tck2

Affects Version/s:

JDO 2 final

Fix Version/s:

JDO 2 maintenance release 1

Time Tracking:

Not Specified

File Attachments:

File Name	Date Attached	Attached By	Size
derby-dsid-pm-junit.txt	2006-10-20 05:41 PM	Michael Bouschen	20 kB
RefreshAllSideEffects.java	2006-10-20 11:42 PM	Michelle Caisse	7 kB

Resolution Date:

08/Jan/07 07:03 PM

Description

« Hide

I see the same DERBY error: "ERROR 40XL1: A lock could not be obtained within the time requested" as reported in ~~JDO-324~~ for two different test cases:
- RefreshAllWithArraySideEffects
- RefreshAllWithCollectionSideEffects
The reason might be that I switched to a dual core machine.

~~JDO-324~~ fixed test case RefreshAllNoParameterSideEffects. Maybe we need to apply the same fix to the other test cases RefreshAllWithArraySideEffects and RefreshAllWithCollectionSideEffects.

All

Comments

Work Log

Change History

Subversion Commits

Sort Order:

[ Permlink | « Hide ]

Michael Bouschen added a comment - 20/Oct/06 05:41 PM

The log file showing the derby exception.

[ Permlink | « Hide ]

Michelle Caisse added a comment - 20/Oct/06 11:42 PM

I've attached a new file RefreshAllSideEffects.java that replaces both RefreshAllWithArraySideEffects.java and RefreshAllWithCollectionSideEffects.java. The new test invokes refreshAll() with both a Collection and Object[] parameter sequentially depending on the value of a flag:

            if (useCollection)
                pm1.refreshAll(points1);
            else pm1.refreshAll(points1.toArray());

Michael, it would be great if you could test this on a multicore machine. To test, you will need to modify conf/pm.conf or create a simple test conf file referencing the new file name.

Please review. It's a bit ugly.

[ Permlink | « Hide ]

Craig Russell added a comment - 20/Oct/06 11:56 PM

Looks good to me. It's not a multithreaded test any more so I don't know about the value of running on a multicore machine...

[ Permlink | « Hide ]

Michael Bouschen added a comment - 23/Oct/06 08:31 PM

Hi Michelle,

I ran the new test case several times on a multicore machine w/o any failures. So I think the trest case is ready to be checked in.

Regards Michael

[ Permlink | « Hide ]

Michelle Caisse added a comment - 23/Oct/06 08:41 PM

Thanks, Michael. I checked in the new class and an updated pm.conf that invokes the new test, but not the two that it replaces (revision 467118). The tests all pass.

I did not yet remove the two replaced tests from the repository. Is there any reason not to "svn delete" them?

[ Permlink | « Hide ]

Michelle Caisse added a comment - 08/Jan/07 07:03 PM

Replaced test files were deleted with revision: 494153