Bug 1067051 - DTGov processes one deployment several times in clustered environment
Summary: DTGov processes one deployment several times in clustered environment
Keywords:
Status: MODIFIED
Alias: None
Product: JBoss Fuse Service Works 6
Classification: JBoss
Component: DT Governance
Version: 6.0.0 GA
Hardware: Unspecified
OS: Unspecified
unspecified
urgent
Target Milestone: ---
: FUTURE
Assignee: Eric Wittmann
QA Contact: Matej Melko
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2014-02-19 15:03 UTC by Stefan Bunciak
Modified: 2021-02-09 09:53 UTC (History)
2 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed:
Type: Bug
Embargoed:


Attachments (Terms of Use)
Configuration of servers (38.46 KB, text/xml)
2014-02-19 15:05 UTC, Stefan Bunciak
no flags Details
Log of server1 (1.53 MB, text/x-log)
2014-02-19 15:07 UTC, Stefan Bunciak
no flags Details
Log of server2 (209.67 KB, text/x-log)
2014-02-19 15:08 UTC, Stefan Bunciak
no flags Details
Log of server3 (1.25 MB, text/x-log)
2014-02-19 15:08 UTC, Stefan Bunciak
no flags Details


Links
System ID Private Priority Status Summary Last Updated
Red Hat Issue Tracker DTGOV-123 0 Major Closed Query Monitor needs cluster support 2014-08-18 16:42:40 UTC

Description Stefan Bunciak 2014-02-19 15:03:54 UTC
Description of problem:

After configuring FSW to run in a cluster of 3 nodes,and creating a new deployment in DTGov, it started several (cca 5) workflow instances for a single deployment, and tried to deploy application to the target for multiple times.

Version-Release number of selected component (if applicable):
* FSW 6.0.0.GA

How reproducible:
* 100%

Steps to Reproduce:
1. Set up a cluster of 2 or more standalone FSW servers
2. Deploy dtgov-workflows to one of the nodes (be sure that they are replicated to the remaining servers)
3. Run a quickstart (e.g. dtgov-demos-switchyard)

Actual results:
* See server logs for error messages.

Expected results:
* Same behavior as in a single-server configuration (to execute 1 workflow per deployment)

Additional info:
* Server configuration attached. I used shared Mysql database for all the cluster nodes.

Comment 1 Stefan Bunciak 2014-02-19 15:05:24 UTC
Created attachment 865105 [details]
Configuration of servers

Servers were configured to use shared modeshape repo so that each node can operate on the same data.

Comment 2 Stefan Bunciak 2014-02-19 15:07:46 UTC
Created attachment 865107 [details]
Log of server1

Log contains also error messages produced by BZ #1066618

Comment 3 Stefan Bunciak 2014-02-19 15:08:12 UTC
Created attachment 865108 [details]
Log of server2

Log contains also error messages produced by BZ #1066618

Comment 4 Stefan Bunciak 2014-02-19 15:08:38 UTC
Created attachment 865109 [details]
Log of server3

Log contains also error messages produced by BZ #1066618

Comment 5 Kurt T Stam 2014-05-09 14:32:32 UTC
In a clustered environment we need to make sure one 1 timer run executes at any one given time. At the moment with lets say 2 nodes, you will get a timer run on each nodes, which means there can be race conditions and you may get more then one workflow process for the same artifact. I think the easiest is to add a temporary lock - probably a lock artifact in s-ramp that will expire at the end of the query run, or time out after x amount of time. This means that s-ramp needs to support txn to create this artifact.

Comment 6 JBoss JIRA Server 2014-07-30 16:42:51 UTC
Eric Wittmann <eric.wittmann> updated the status of jira DTGOV-123 to Coding In Progress

Comment 7 JBoss JIRA Server 2014-07-30 18:11:43 UTC
Eric Wittmann <eric.wittmann> updated the status of jira DTGOV-123 to Resolved

Comment 8 JBoss JIRA Server 2014-07-30 18:20:34 UTC
Eric Wittmann <eric.wittmann> updated the status of jira DTGOV-123 to Closed

Comment 9 Eric Wittmann 2014-08-18 16:42:35 UTC
This has been fixed in community and will be available in FSW 6.1. See the associated JIRA for more info.


Note You need to log in before you can comment on or make changes to this bug.