A cluster is a set of application servers that you manage
together as a way to balance workload.
Before you begin
Before you create a cluster:
- Review the content of the topic "Clusters and workload management,"
especially the information about setting cluster weights.
- Decide if you want enterprise bean requests routed to the node
on which the client resides.
- Decide if you want to use HTTP memory-to-memory replication.
- Determine the appropriate configuration settings
for the first cluster member. A copy of the first cluster member that
you create is stored as part of the cluster data and becomes the template
for all additional cluster members that you create.
- Decide on which node you want the first cluster member to reside.
About this task
You might want to create a cluster if you need to:
- Balance your client requests across multiple application servers.
- Provide a highly available environment for your applications.
A cluster enables you to manage a group of application
servers as a single unit, and distribute client requests among the
application servers that are members of the cluster.
On the z/OS platform, if you plan to create a cluster
of servers that spans multiple systems in a sysplex and has stateful
session beans with an activation policy of Transaction deployed in
them, the passivation directory should reside on an HFS (hierarchical
file system) that is shared across the multiple systems in the sysplex
on which the clustered servers are running.
To create a cluster:
Procedure
- In the administrative console, click Servers > Clusters
> New. The Create a new cluster wizard starts.
- Specify a name for the cluster.
- Optional: Specify a short name
for the cluster. This field only appears if you are running
on a z/OS system.
For clustered servers, the WLM application environment
is the default value for the cluster short name. If you specify a
short name for a cluster, the name:
- Must be one to eight characters in length
- Must contain only uppercase alphanumeric characters
- Cannot start with a number.
- Must be unique in the cell
- Cannot be the same as the value specified on the ClusterTransitionName custom
property of any non-clustered server. Do not specify a cluster transition
name for a server that is part of a cluster.
Important: If you specify a short
name, make sure that you set up a RACF SERVER class profile that includes
this short name.
- Select Prefer local if you want to enable node-scoped
routing optimization. This option is enabled by default.
When this option is enabled, if possible, EJB requests are routed
to the client node. This option improves performance because client
requests are sent to local enterprise beans.
Supported configurations: If you enable the preferLocal
optimization, the deployment manager must be running to affect the
configuration. If the deployment manager is shut down, preferLocal
optimization is not performed and requests might be dispersed across
all the members of the cluster.
sptcfg
- Select Configure HTTP session memory-to-memory replication if
you want a memory-to-memory replication domain created for this cluster.
The replication domain is given the same name as the cluster
and is configured with the default settings for a replication domain.
When the default settings are in effect, a single replica is created
for each piece of data and encryption is disabled. Also, the Web container
for each cluster member is configured for memory-to-memory replication.
To
change these settings for the replication domain, click Environment
> Replication domains > replication_domain_name. To modify
the Web container settings, click Servers > Clusters > cluster_name >
Cluster members > cluster_member_name > Web container
settings > Session management > Distributed environment settings in
the administrative console. If you change these settings for one cluster
member, you might also need to change them for the other members of
this cluster.
- Click Next.
- Choose whether to create an empty cluster or to create
the first member of the cluster.
If you decide to create
an empty cluster, to add members to this cluster, in the administrative
console, click Servers > Clusters > cluster_name >
Cluster members > New.
To create an empty cluster:
- Select None. Create an empty cluster.
- Click Next to display a summary of the defined
cluster.
- Click Finish to create the cluster, or click Cancel if
you decide not to create this cluster.
When you create
the first cluster member, remember that a copy of the first cluster
member that you create is stored as part of the cluster data and becomes
the template for all additional cluster members that you create.
- Specify the name of the first cluster member.
- Select the node on which you want this cluster member
to reside.
- Specify a short name for this cluster
member. The short name is the default z/OS job name and
identifies the cluster member to the native facilities of the operating
system, such as Workload Manager (WLM), Automatic Restart Manager,
SAF (for example, RACF), started task control, and others.
- Specify the weight value for the cluster member.
The weight value controls the amount of work that is directed
to the application server. If the weight value for this server is
greater than the weight values that are assigned to other servers
in the cluster, then this server receives a larger share of the workload.
The weight value represents a relative proportion of the workload
that is assigned to a particular application server. The value can
range from 0 to 20.
On the z/OS platform, weight is used to balance
some of the workload types, but others are balanced by the z/OS system.
- For HTTP requests, weights are used to distribute HTTP traffic between
the Web server plug-in and the controller handling the clustered application
server. Assign a higher weight value to the application server that should
receive the HTTP traffic.
- For Web services calls, information is transferred from a servant in
one application server to a controller in another application server. The
application server that receives the call has the highest weight value.
- Weight has no affect on Internet Inter-ORB Protocol (IIOP) requests.
IIOP requests are distributed to the correct application server using the
sysplex distributor.
- Select Generate unique HTTP ports if you want
to generate unique port numbers for every HTTP transport that is defined
in the source server. When this option is selected, which
is the default setting, this cluster member does not have HTTP transports
or HTTP transport channels that conflict with any of the other servers
that are defined on the same node. If you unselect this option, all
of the cluster members will share the same HTTP ports.
- Select the core group to which you want this cluster
member to belong. You are prompted for the core group only
if you have more than one core group defined for this cluster.
- Select one of the following options as the basis for
the first cluster member.
- Create the member using an application server template.
If you select the defaultZOS template, which is the
only one that is listed unless you used the createServerTemplate command
for the AdminTask object to create additional templates, the first
cluster member uses the default port assignments for the z/OS platform.
If some of these ports are already defined for use elsewhere in your
z/OS system, your newly created cluster member might not start, might
function incorrectly, or might generate unexpected error messages.
Therefore, you must resolve any port conflicts before you start this
server.
- Create the member using an existing application server as a template.
- Create the member by converting an existing application server.
Avoid trouble: You can only add an existing
application server to the cluster if you select that server as the
first cluster member. You cannot add other existing application servers
to that cluster after you create the first cluster member. If you
add an existing server to a cluster, the only way to remove that server
from the cluster is to delete the server. Therefore, you might want
to use the existing server as a template for the first cluster member
instead of as the cluster member. If you keep the original application
server out of the cluster, you can reuse that server as the template
if you need to rebuild the configuration.
gotcha
- Click Next.
- Create additional cluster members. Before you
create additional cluster members, check the configuration settings
of the first cluster member. These settings are displayed at the bottom
of the Create additional cluster members panel of the Create a new
cluster wizard. For each additional member that you want to create:
- Specify a unique name for the member. The
name must be unique within the node.
- Select the node to which you want to assign the cluster
member.
- Specify the weight you want given to this member.
The weight value controls the amount of work that is directed
to the application server. If the weight value for the server is greater
than the weight values that are assigned to other servers in the cluster,
then the server receives a larger share of the workload. The value
can range from 0 to 20.
- Specify a short name for this cluster
member. The short name is the default z/OS job name and
identifies the cluster member to the native facilities of the operating
system, such as Workload Manager (WLM), Automatic Restart Manager,
SAF (for example, RACF), started task control, and others.
- Select Generate unique HTTP ports if you want
to generate unique port numbers for every HTTP transport that is defined
in the source server.
- Click Add member. You can edit the
configuration settings of any of the newly created cluster members
other than the first cluster member, or you can create additional
cluster members. Click Previous to edit the properties of the
first cluster member.
- When you finish creating cluster members, click Next.
- View the summary of the cluster and then click Finish to
create the cluster, click Previous to return to the previous
wizard panel and change the cluster, or click Cancel to exit
the wizard without creating the cluster.
- To further configure a cluster, click Servers > Clusters,
and then click the name of the cluster. Only the Configuration and Local
Topology tabs appear until you save your changes.
- Click Review to review your cluster configuration
settings. Repeat the previous step if you need to make
additional configuration changes.
- If you do not want to make any additional configuration
changes, select Synchronize changes with Nodes and then click Save.
Your changes are saved and synchronized across all of your nodes.
Avoid trouble: If you click
Save, but do not
select Synchronize changes with Nodes, when you restart the cluster,
WebSphere Application Server does not start the cluster servers because
it cannot find them on the node. If you want to always synchronize
your configuration changes across your nodes, you can select Synchronize
changes with Nodes as one of your console preferences.
gotcha
- Restart the cluster.
Results
You have a configured cluster to which you can assign work
requests. The Runtime and Local Topology tabs appear
the next time you access this page.
What to do next
- Create additional cluster members.
- Use the administrative console to view or change the configuration
settings for a cluster. For example, if you are running in a high
availability environment, you can click Servers > Clusters > cluster_name,
and then select the Enable failover of transaction log recovery option
for this cluster. This option allows the recovery of transactions
to failover from one cluster member to another.
- Create additional cluster members.
- Start the cluster.
- Use scripting to automate the task of creating clusters.