eFix (APAR): pq47372
Status: efix

For Release: WebSphere 3.0.2.3

For Operating System: AIX only

Supercedes eFixes: N/A

CMVC defect: pq47372, pq46826, pq46480, 88543, 89841, 89228, 90715, 89349, 81301 

Byte size of APAR: Part 1 297KB, part 2 4KB 

Date: 4/4/01

Abstract: Enhancement in dispatching algorithm for the Web server plugin

Description/sypmtom of problem:

- Enhancement in dispatching algorithm for the Web server plugin
- CST_AIX: Internal Server errors by IHS plug-in
- Make identifier id configurable for url rewriting
- OSE Remote failover broken when two or more clones are down
- Default values for OSE connection retries are too high
- Error on some request when  Servlet Transport Type = Local Pipe      
- Fixing minor compiler warnings for AIX in sslink.c                     
- ASV302 Session Affinity for re-written URLs - align w/sessmgr          
- Added caching of clone IP addresses (performance improvement)
- CPU spikes when the default server is started 

DIRECTIONS TO APPLY EFIX:

Part 1:
1) Stop the Web Server(s).
2) Stop all WebSphere clones
3) Stop WebSphere Admin Server and Nanny process
4) Verify that all java processes are stopped
5) Run utility "slibclean"
6) Save the old versions of the affected "*.so" files in /usr/WebSphere/Appserver/bin
7) Copy the new *.so files into /usr/WebSphere/AppServer/bin
8) End of Part 1 Install Instructions; Start Part2.


Part 2:
9) Create "efix" directory to store the efix jar file(s):
   AIX: /usr/WebSphere/AppServer/efix
10) Copy pq47372_part2.tar to the directory
11) Add the directory/zip file to the beginning of the admin server's classpath
  in admin.config:
  com.ibm.ejs.sm.adminserver.classpath=C:/WebSphere/AppServer/efix/pq47372_part2.tar...
12) Restart the web server(s) and restart WebSphere AdminServer


ADDITIONAL INFORMATION:


Notes on configuration parameters:
----------------------------------
a. In bootstrap.properties file the parameter ose.refresh.interval specifies
   how often the properties files are checked for any changes. For production
   environments this should be increased from the default of 20 seconds.
   (suggested value greater than 60)

b. In bootstrap.properties file the parameter ose.failure.retry_timer specifies
   how often clones should be re-checked after an error condition. When an
   error is detected on a clone, new requests are not sent to it until this
   timer expires. Any down clones will also be checked after the propreties
   files have been re-loaded, as controlled by a) above. (suggested value 20)

c. The clone selection can be configured to be random by adding a parameter 
   to the queues.properties file as follows (example is for queue ibmoselink): 
   ose.srvgrp.ibmoselink.clone.selection=RANDOM. If this parameter is not 
   specified it will default to round-robin clone selection.


The following files are included in the fix:
--------------------------------------------
    libascommon.so
    libascommonst.so
    libasin.so
    libasoutc.so
    libasoutcst.so
    libasoutsj.so
    libstubj.so
    SQEventImp.class