Bug 724606 (BRMS-518)

Summary: Performance issue on RHEL6
Product: [JBoss] JBoss Enterprise BRMS Platform 5 Reporter: Tomas Schlosser <tschloss>
Component: unspecifiedAssignee: Tihomir Surdilovic <tsurdilo>
Status: CLOSED WORKSFORME QA Contact:
Severity: unspecified Docs Contact:
Priority: urgent    
Version: 5.1.0 GACC: jlocker, lpetrovi, psiroky, tsurdilo
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
URL: http://jira.jboss.org/jira/browse/BRMS-518
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
RHEL6 (both 32bit and 64bit architecture) Sun JDK 1.6.0_20 Oracle 11g R2 EAP 5.1 JackRabbit Firefox 3.6.9
Last Closed: 2011-02-07 10:36:10 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:
Attachments:
Description Flags
brms-leak-os.png
none
brms-leak-os_2.png
none
brms-leak-test-rhel6_x86-oracle11gR2-operations.png
none
brms-leak-test-rhel6_x86_64-derby-operations.png none

Description Tomas Schlosser 2010-12-22 07:44:38 UTC
securitylevel_name: Public

Comapred to other operation systems the test time increases significantly over time. See included chart for details. Also memory usage chart is included.

Comment 1 Tomas Schlosser 2010-12-22 07:45:13 UTC
Attachment: Added: brms-leak-os.png


Comment 2 Tomas Schlosser 2010-12-22 07:45:36 UTC
Attachment: Added: brms-leak-os_2.png


Comment 3 Tihomir Surdilovic 2011-01-06 14:21:30 UTC
What is the test performed?

Comment 4 Lukáš Petrovický 2011-01-06 14:30:19 UTC
It is a BRMS Leak tests. It is a test that does hundreds of loops of the following:

                guvnor.login("admin");
                Browser b = guvnor.browseTechnicalRules("mortgages");
                DrlRule r = b.openDrlRule("Dummy rule");
                r.define("when\nthen System.out.println(" + i + ");");
                r.validate();
                r.saveChanges("Some commit");
                Package p = guvnor.openPackage("mortgages");
                p.viewSource();
                p.saveAndValidate();
                build.startMeasuring();
                p.createSnapshotExisting("TEST", "My TEST Snapshot");
                guvnor.signOut();

(Please understand the above code as a pseudocode representing operations in the BRMS GUI. If any of them are unclear, I can explain more.)

Comment 5 Tihomir Surdilovic 2011-01-06 14:36:58 UTC
Can you identify which of the above steps cause the memory jumps on RHEL? What are ways I can reproduce the issue so it can be debugged?

Comment 6 Tihomir Surdilovic 2011-01-06 14:44:23 UTC
Do you get same/similar results running against Drools trunk ( https://github.com/droolsjbpm/droolsjbpm ) ?

Comment 7 Petr Široký 2011-01-11 11:10:17 UTC
I added chart with duration of mentioned operations. It seems that the duration of each operation is slowly increasing over time.

I am not sure if we'll be able to run the test against Drools trunk (we need to compile it, right?), but we'll try it.



Comment 8 Petr Široký 2011-01-11 11:10:17 UTC
Attachment: Added: brms-leak-test-rhel5_x86-oracle11gR2-operations.png


Comment 9 Petr Široký 2011-01-11 11:12:21 UTC
Attachment: Removed: brms-leak-test-rhel5_x86-oracle11gR2-operations.png 


Comment 10 Petr Široký 2011-01-11 11:12:36 UTC
Attachment: Added: brms-leak-test-rhel6_x86-oracle11gR2-operations.png


Comment 11 Petr Široký 2011-01-11 11:19:07 UTC
Added one more chart where the increasing time of each operation is more obvious.

The test was ran with derby, but the results are same as with oracle 11g.

Comment 12 Petr Široký 2011-01-11 11:19:07 UTC
Attachment: Added: brms-leak-test-rhel6_x86_64-derby-operations.png


Comment 13 Petr Široký 2011-01-12 11:56:13 UTC
I wasn't able to set up the the Drools trunk (guvnor war). The compilation, packaging and deploy is ok, but when accessing the guvnor web page, the message "Loading data, please wait" is shown and even after hour does not go away. So I am not able to run the test. 

Comment 14 Petr Široký 2011-01-20 20:28:59 UTC
I ran some more tests and the result is quite surprising. The performance issue only occurs when the test is running on RHEL 6 with Firefox 3.6. The results from other combinations like RHEL 5 + Firefox 3.6 (or 3.0) and RHEL 6 + Firefox 3.0 are quite good and the issue is not there. 

I also noticed that the test on RHEL 6 + Firefox 3.6 is taking twice as long than with other configurations. I am not sure why it is happening, but I'll try to look at it more closely.

Comment 15 Petr Široký 2011-02-07 10:36:10 UTC
Closing this issue, because the test result on RHEL 6 + FF 3.6 is now pretty much same as with other configurations. The update of RHEL6 probably solved the problem.