[jboss-jira] [JBoss JIRA] Updated: (JBPM-1071) Possible problem in concurrent signalling from multiple threads

Tom Baeyens (JIRA) jira-events at lists.jboss.org
Thu Oct 18 06:10:03 EDT 2007


     [ http://jira.jboss.com/jira/browse/JBPM-1071?page=all ]

Tom Baeyens updated JBPM-1071:
------------------------------

    Attachment: LockingTest.java

I attached a test LockingTest.  This test does basically the same as your test, but it fits in our overall test suite.

This test runs fine.  So I can't reproduce the problem.

The way that I check the optimistic locking is a bit different.  First, I spawn of x number of competing threads/transactions.  In each transaction, a comment is added that includes the name of the thread.

Later on, we check if exactly 1 thread managed to commit, while the other were rolled back.

What I think that you do different is that you don't use transactional resource to measure which transactions committed and which transactions rolled back.  A limitation of the current implementation is that you can't see if an optimistic failure rollback occured or not.

> Possible problem in concurrent signalling from multiple threads
> ---------------------------------------------------------------
>
>                 Key: JBPM-1071
>                 URL: http://jira.jboss.com/jira/browse/JBPM-1071
>             Project: JBoss jBPM
>          Issue Type: Bug
>          Components: Core Engine
>    Affects Versions: jBPM jPDL 3.2.2
>         Environment: Linux  2.6.21-1.3228.fc7 #1 SMP Tue Jun 12 14:56:37 EDT 2007 x86_64 x86_64 x86_64 GNU/Linux
> MySQL 5.0.22
>            Reporter: Jiri Pechanec
>         Assigned To: Tom Baeyens
>            Priority: Critical
>         Attachments: expl.tar.gz, LockingTest.java
>
>
> Attached is a simple test case that
> 1) Deploys process definition with two nodes
> 2) Starts the process instance that will go to the wait state on first node
> 3) Starts 20 threads that tries concurrently signal the same process instance
> 4) The second node writes a record to the database
> The test case needs to be executed multiple times to see the incorrect behaviour.
> This is an example of run output
> Isol 8
> Action 1
> Action 2
> Action 2
> Action 2
> Action 2
> Action 2 1
> Action 2 1
> Action 2 1
> Action 2 1
> Signalist 5
> Signalist 6
> Signalist 8
> Signalist 12
> Signalist 7
> Signalist 13
> Signalist 14
> Signalist 15
> Signalist 9
> Signalist 16
> Signalist 17
> Signalist 18
> Signalist 4
> Success 7
> Failure 13
> Explanation of the outcome
> 4 threads successfully executed the node action including database operation. All database opeartion were comitted (4 new records were created)
> 3 threads successfully executed the signal operation but no real action was performed
> 13 threads attempted to execute the signal operation but ended with an exception

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        



More information about the jboss-jira mailing list