The composition of the configuration system is shown below:
Ø 172.18.41.112
Web Server, use IHS6 (remote installation, HTTP port 80)
WAS6, application node, join 98, Server1
Ø 172.18.41.98
WAS6, management node
Ø 172.18.41.69
WAS6, application node, join 98, Server2 to establish a cluster:
Cluster: Server1 (9080), Server2 (9081)
Virtul Host: Default Host, and increase the alias *: 9081
Several questions that need attention:
1. If you do not select a copy domain when establishing a cluster, you need to design a session replication policy for each web capacity.
2, applying to the configuration of the host
3, the generation of plug-in configuration. Addition in httpd.conf:
LoadModule WAS_AP20_Module "C: /IBM/websphere/plugins/bin/websphere/plugins/bin/mod_was_ap20_http.dll" WebSpherePluginConfig "C: /IBM/websphere/plugins/config/webserver112/plugin-cfg.xml"
The path is determined by the specific situation of your installation plugin.
For the plugin-cfg.xml file, pay attention to the defined part of the cluster and the forwarding policy portion.
The IHS6 is configured to forward requests to cluster AngelCluster1.
Tested, confirmed:
1, Cluster session can be synchronized (to configure multiple synchronization policies)
2. After a machine DOWN machine in the cluster can be implemented, the service is automatically forwarded to another server.
3. Enable load balancing (policy can be combined) for testing applications containing the following simple features:
http: // ip: Port / App / Servlet / TestCluster? Cluster =
Among them: If the cluster value is empty, the system will return the value of the Cluster in Session, and printed the session ID. If the cluster value is worth, use this value to update the value of Cluster in the session, and print the session ID
Test 1: Do not forward, direct requests, such as:
http://172.18.41.69:9081/app/servet/testcluster?cluster=12345
http://172.18.41.1129080/app/servlet/testcluster? cluster =
The system can work normally and printed the value in the desired value (the session of the two machines).
Test 2: Forwarded through Web Server:
http://172.18.41.112/app/Servlet/testCluster?cluster=12345
http://172.18.41.112/app/servlet/testcluster?cluster=
You can get a response normally and printed out the desired information.
Test 3: Abnormal situation
During the request, kill the process / deployment manager process / node agent process for one of the application servers, see the availability of the application.
RESULTS: Application is still available, can be seen in the log, plug-in finds that the application forwards to another when the application server is not available. At this point, the client will feel slow in response time.
Test, confirm: 1. Cluster session can be synchronized (configurable in multiple synchronization policies)
2. After a machine DOWN machine in the cluster can be implemented, the service is automatically forwarded to another server.
3, can achieve load balancing (policy can be combined)