Is there a way to stop/start request directed to cluster member of a websphere application server?

The use-case is
- Stop the member
- Stop request directed to a cluster member (Anyway the cluster member will be notified, hence no request will be directed), But ensure the requests are not redicted unless enabled
- Start cluster member
- Wait for application server (member) to stablize
- The start request direction

Any help, would be much appreciated.

You can write script for this requirement. You can define individual JVM of cluster to start/stop, wait for couple of mins, start/stop second member of cluster.

Cluster member can be started stopped using console. No Issues. But wanted to know if there is a way to stop request direction to the member that is going to be stopped.

Thanks for the reply.

try changing the runtime weight of the cluster member..

details : pic.dhe.ibm.com/infocenter/clmhelp/v4r0/...

Does that require server restart? or Just changing the parameter is good enough?

I got to know changing the profile would require server restart.

Please throw some light.

runtime weight - no restart
configured weight - require restart

Seems restart not required for configured weight as well.

pic.dhe.ibm.com/infocenter/clmhelp/v4r0/...

Can you confirm?

changing weight change means .. you are making change to master configuration file. This needs restart and the change in value statys even after restart.

changing runtime is temporary change to runtime, nothing will be saved to master configuration files. 

Ok.. I got it.

Here is my problem, On server restart usetime DBCP Connection goes high and exausting total connections from the pool. And usetime settle downs to acceptable limit once the application stablizes. I wanted to avoid this situation by

- Stop request direction to the member
- Stop the server
- Do some configuration changes
- Start the server
- and start request direction

Please post if any insights.

so.. ur actual issue is with managing load.

why are so many requests queued up and waiting ? you might need to control the no of requests by tuning webContainer thread pool.

Try this
determine how many requests are coming to web server (check using server-status module)..
check your web container thread pool settings
check your database connection pool settings

for example: if your web server is handling 100 requests per minute, then your web container should accept around 60-75 and then your database pool should be around half of web container.  [these things are debatable but you should always make requests queue at web server first and then web container but not for database connection]

If above method looks like a complicated approach .. then if you have enough system resources, try adding another JVM. If you do this be sure to check what is the max connections/sessions setting on database.

Here is my problem, On server restart usetime DBCP Connection goes high and exausting total connections from the pool. And usetime settle downs to acceptable limit once the application stablizes. I wanted to avoid this situation by

- Stop request direction to the member
- Stop the server
- Do some configuration changes
- Start the server
- and start request direction

Please post if any insights.

[quote author=204708499 post=536662889]Here is my problem, On server restart usetime DBCP Connection goes high and exausting total connections from the pool. And usetime settle downs to acceptable limit once the application stablizes. I wanted to avoid this situation by

- Stop request direction to the member
- Stop the server
- Do some configuration changes
- Start the server
- and start request direction

Please post if any insights.[/quote]

is this a copy paster mistake?
you've posted same thing before my last post above

Hi Nageswara,

  If you have requests balanced by a WebSphere Plugin, you can put the server as backup rol.
 
  From 8.5 infocenter but still valid from v6.1.
  pic.dhe.ibm.com/infocenter/wasinfo/v6r1/...

     "The plug-in does not load balance across the backup application servers. A backup server is only used if a primary server is not available."" Your other cluster members need be working fine.
    
  You only need to go through admin console:
 
  Servers > Server Types > WebSphere application servers > server_name, and then, in the Additional Properties section, click Web server plug-in properties.
 
  Change Server Role from Primary to backup.
 
  Regenerate and propagate the plugin.  Live connections will finish and new ones will redirected to other cluster members. When you see that you haven't activity you can "work" with server configuration and stop/start. When the server is up, you can change the role and teh server will receive again requests.
 
  For maintenance purpose I have used and in a lot of clients.
 
  As Joseph's suggest you need to review your environment configuration to avoid exaustion of the connection pool.
 
  Hope this helps. Tell us if you need more support.
 
Regards

Yes, this seems to be feasible solution. Thankyou You guys Rocks!!!