Browse Prior Art Database

Logging Techniques in WebSphere Vertical Clustering Environment

IP.com Disclosure Number: IPCOM000198346D
Publication Date: 2010-Aug-05
Document File: 3 page(s) / 91K

Publishing Venue

The IP.com Prior Art Database

Abstract

This article presents logging techniques to work with Log4J and WebSphere vertical clustering environment.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 46% of the total text.

Page 1 of 3

Logging Techniques in WebSphere Vertical Clustering Environment

Problem Statement - With Vertical clustering (single machine and multiple JVM's) in place, there would be log files locking issues and roll-over contention as all singleton Log4J instances will try to use/write to the same log file. This could sometime end up not logging at all, sometimes logging on to a backup log file etc.

Known Solution, it's drawbacks and why a need for additional solution - Keep the log4

                                                                ar inside application's server main lib rather packaging the jar inside web archive or application archive. This way it will be initialized at earlier and will be invoked as singleton only once. This doesn't seem to be feasible on production environment which might be hosting other applications utilizing the same web sphere server. And the other applications might not want it that way or the infrastructure team might not allow you to keep an open source library in the server main lib. So the additional solution here is required which without affecting other applications on same web sphere server can assure that all the logging is done properly. Below is the only one technical forum discussion which I could find on developerworks related to this topic.
http://www.ibm.com/developerworks/forums/thread.

j

_LOG

Overview on Vertical clustering -

A single application server, which is implemented by a single JVM process, cannot always fully utilize the processor power and RAM of a large multiprocessor machine. Clustering with vertical cluster members provides a straightforward mechanism to create multiple JVM processes, which together can fully use all the processing power and memory available.

Prerequisite -

I assume that the application is already extending the existing log4

                                           framework to write it's own application specific logging components so that we can give a name to our application log file and not only just one log file, we can have one log file for each of the type of components deployed in our application. For example GUI component can have its own log file, backend and web services component can have one, all EJB/MDB log can go in one log file etc.

How the technique works, the implementation details :

(1) Modify log4

j

j.properties, create system property in WebSphere using admin console.

j.properties (we might have one for each of the componen

j.appender.ROOT.File

= MyAppLog.log

   I presume that earlier the entry would be logs/MyAppLog.log. So here our aim is to create one log file for each of the server in the vertical cluster. For example if we have two clusters in the server say App001 and App002, we would have two log files created as our ultimate goal. One

1

j

j

-version.

spa?threadID=220498

The Solution I designed and implemented in my project - I discovered a very simple approach in Virtual Solution Center (VSC) application to resolve the issue - I was able to create the same log file under each server in the cluster thus no locking and roll-over content...