Bug 779558 (SOA-1934)

Summary:	jBPM : Process instances stuck in a fork
Product:	[JBoss] JBoss Enterprise SOA Platform 4	Reporter:	Darran Lofthouse <darran.lofthouse>
Component:	JBPM - within SOA	Assignee:	Alejandro Guizar <alex.guizar>
Status:	CLOSED NEXTRELEASE	QA Contact:
Severity:	urgent	Docs Contact:
Priority:	urgent
Version:	4.3 CP02	CC:	dlesage
Target Milestone:	---
Target Release:	4.3 CP04 ER1
Hardware:	Unspecified
OS:	Unspecified
URL:	http://jira.jboss.org/jira/browse/SOA-1934
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2010-03-30 11:36:18 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Deadline:	2010-03-08

Description Darran Lofthouse 2010-02-08 11:37:23 UTC

Date of First Response: 2010-02-25 15:19:39
Help Desk Ticket Reference: https://enterprise.redhat.com/issue-tracker/404803
project_key: SOA

There are problems with a jBPM process which contains a fork/join when running in a cluster.

If run with just one node, the processes end properly, but with two nodes the child tokens don't reactivate the parent token

Any ideas? Such as:
* specifying <transaction-isolation>TRANSACTION_READ_COMMITTED</transaction-isolation> on the JbpmDS
* trying lock="UPGRADE_NOWAIT" on the join

Comment 1 Len DiMaggio 2010-02-09 15:32:04 UTC

Link: Added: This issue related SOA-1939

Comment 2 Alejandro Guizar 2010-02-15 16:57:34 UTC

Link: Added: This issue incorporates JBPM-2787

Comment 3 Alejandro Guizar 2010-02-16 06:28:36 UTC

Link: Added: This issue is related to SOA-1942

Comment 4 Anne-Louise Tangring 2010-02-25 20:19:39 UTC

Approved, but not required, for SOA 4.3 CP03. If it is not resolved by Due Date, it's out.

Comment 5 Alejandro Guizar 2010-03-11 10:09:38 UTC

JBPM-2787 is done and the fix got positive feedback from the customer. SOA-P team, please close this issue as you see fit.

Comment 6 David Le Sage 2010-03-15 01:16:13 UTC

Draft text for Resolved Issues section of Release Notes:

https://jira.jboss.org/jira/browse/JBPM-2787

    If a user attempted to run a jBPM process on two nodes in a cluster and that process contains a
    join, the child tokens would not reactivate the parent token.

    To prevent concurrent tokens from being stuck in a join node, each of them ends when it enters
    that node. Normally, ending a token would mean it is unable to reactivate its parent but this
    fix restores that ability to it. The join execute method checks that all child tokens have
    ended before it marks the parent token's reactivation flag. The join node has also been made
    asynchronous. As a result of these changes, it is now less likely that a process will become stuck
    in a join.

Comment 7 Marek Baluch 2010-03-30 11:36:18 UTC

Verified on 4.3 CP03.