WebSphere Application Server V8.5 Administration and Configuration Guide for the Full Profile

700 WebSphere Application Server V8.5 Administration and Configuration Guide for the Full Profile

8. Repeat steps 5 to 7 on page 205, but enter a custom property with a name of

com.ibm.websphere.zos.mvsservices.enable and a value of true so that the

Administration services Custom properties view contains the properties shown in

Figure 21-85.

Figure 18-85 Administration services custom properties window

9. Repeat steps 2 to 8 for all the node agents tied to your cluster.

10.Restart the affected node agents.

For information about a staged application deployment rollout, go to the following website:

http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP101641

For information about updating a high-availability application manually with a server stop, go

to the following website:

http://pic.dhe.ibm.com/infocenter/wasinfo/v8r5/index.jsp?topic=%2Fcom.ibm.webspher

e.zseries.doc%2Fae%2Fcontainer_update_ha_apps_zos.html

18.4 Enabling multiple servants

Enabling multiple servants on your systems can improve resource availability on your system.

With WLM you can control the number of servants that are running and how they are

performing. WLM manages the response time and throughput of transactions according to

their assigned service class, the associated performance objectives, and the availability of

system resources.

Based on your workload requirements you can control the maximum or minimum number of

servants running on a server. The minimum value is used for starting up a basic number of

servers and the maximum value is used for capping the number of address spaces that are

started by WLM for each server instance.

Manual rollout procedure: For the manual rollout procedure, disable nodeagent

Automatic synchronization and Startup synchronization under System administration 

Node agents  Node agent name  File synchronization service.

Chapter 18. Clustering and high availability 701

To use this service, the multiple server instance needs to be enabled from the administrative

console:

1. Select Servers  Server Types  WebSphere Application Server  server_name.

2. Under Server Infrastructure  Java and Process Management  Server Instance,

(Figure 18-86).

Figure 18-86 Selecting Server Instance

3. Select the Multiple Instance Enabled option and then set the values for the maximum

and minimum number of instances, (Figure 18-87). If you leave these fields blank, WLM

will determine the numbers.

Figure 18-87 Configuration value setting

4. Click Apply to apply the changes and then click Save to save the changes.

You can use the z/OS MODIFY command to dynamically increase the minimum and maximum

values. For example, your server is currently configured with minimum set to two and

maximum set to four, and currently you have only two servants running, but you want all four

to be running. Issue the following MODIFY command to change the minimum value and start

the additional servants.

F Z85SR02B,WLM_MIN_MAX=4,6

The above command changed the minimum value to 4 and maximum value to 6. Similarly if

we want to modify from min=4, max=6 to min=5 and max=6, we can use the following

command:

F Z85SR02B,WLM_MIN_MAX=5,6

702 WebSphere Application Server V8.5 Administration and Configuration Guide for the Full Profile

When using the MODIFY command, if the minimum value is greater than the configured value,

the excess servants are started. If the minimum value is less than the configured value, the

excess servants are stopped.

Similarly, if the current number of servants running is four, and there are requirements for

more servants, WLM starts the additional servants until the maximum configured value is

reached.

For more information about configuring the servant region, refer to the following website:

http://pic.dhe.ibm.com/infocenter/wasinfo/v8r5/index.jsp?topic=%2Fcom.ibm.webspher

e.zseries.doc%2Fae%2Ftrun_control_servants.html

18.4.1 Balancing workload with WLM

When using multiple servants, there can be situations where one servant is utilized more than

the other servants because WLM will not bring another servant region in use until the first one

is fully used or a problem is encountered with the servant region.

To help evenly distribute the workload between servant regions, you can use the environment

variable for enabling WLM stateful session placement.

To set the variable on: wlm_stateful_session_placement_on.

This tells WLM to balance session affinities across different servant regions. Session affinity

is when a session is established in a particular JVM, subsequent requests that belong to the

same session are assigned to the JVM that has the established session affinity.

By using the wlm_stateful_session_placement_on variable, WLM balances the load between

servant regions. So if a new user request comes in, WLM directs the request to another

servant region.

This setting is mainly used for applications having requirements for session affinity.

18.4.2 Balancing workload without WLM

There can be situations where you do not want to use WLM for load balancing but rather

based load balancing on application requirements. In this case, work distribution can be

controlled using WebSphere queuing, following an algorithm based on application

requirements, such as round robin or session affinity. The application server custom property

server_work_distribution_algorithm can be used.