[JBoss-dev] partitioning jboss for SPECjAppSever2002 ?

[Christofer Dutz]

Hi, 

sorry for contacting the dev-mailinglist, but my problem couldn't be
solved in the user-list. I hoped mabe the developers could help me. 

As I am still working on SPECjAppServer2002 as a unversity project. The
benchmark is runing nicely but I am trying to get even more performance
out of jboss.

The application consists of 5 parts : Corporate, Manufacturiung,
Supplychainmanagement, Customer and Utility. 

I could monitor quite a lot of inter-cluster-traffic when runing the
benchmark and wanted to use the Jboss partitioning to minimize this
load. I configured the EJBs in the parts to use different partitions 
CorpPartition, MfgPartition, SupplyParttion, CustPartitiion and
UtilPartition. 

My netx step was to create 4 jboss configurations. Each configuration
should contain UtilPartition and one of the 4 remaining partitions. 
I did have some Problems configuring this, but I think I managed to do
this, by ading a new partition-element in cluster-service.xml and to
change the other Services to depend on UtilPartition and the other one.
I also added a  element (now having 2 of
these elements - I don't know if this is correct). 

If I deploy the benchmark now I only get some Errors from the EJBs of
Partitions that don't run on this server (I woud realy like to get rid
of these errors).

After deploying the application on all servers and starting the
benchmark driver I can see some activity on one host and after a few
seconds I get NamingExceptions: could not dereference Object -
javax.naming.NameNotBoundException

It seems that the HAJNDI seems to fail with this config. 

Could you please help me with configuring this the right way? It seems
that almost no other applicationserver supports this kond of
partitioning and I think this could be a major advantige.

Christofer Dutz



---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
___
JBoss-Development mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-development

[JBoss-dev] SPECjAppServer2002

[Christofer Dutz]

Hi,

as part of my student research project I am working on deployment of the
SPECjAppServer2002 benchmark on a JBoss server. At the moment I am stuck
at one problem I just can't solve. I contacted Dain Sundstrom for help
and I was told to contact this mailinglist.
In the application there are several entity beans using CMP.
Unfortunately when using CMR the column used for storing the CMR-Data is
also a field of the corresponding entity. This results in SQL-Statements
containing one column twice. After discussing this problem with my ...
tutor ... we seem to have two options:
1. As the field containing the CMR data is nor directly used in the
application I could comment-out both the CMP-Field definition on the
definition of the finder (for example findByOrder). Then I would have to
change the all calls to this method to the getOrdLines from the CMR.
2. I could get rid of the CMR and modify all calls to getOrdLines to
findByOrders
Since testing of CMR should be part of the benchmark I would prefer
solution number one. Are there any other suggestions?
This is the last part I am hanging at. Plese help me.

Christofer Dutz
Student at
Technical Univeristy Darmstadt (Germany)




---
This SF.net email is sponsored by:Crypto Challenge is now open! 
Get cracking and register here for some mind boggling fun and 
the chance of winning an Apple iPod:
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0031en
___
Jboss-development mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-development