Cluster restart

Edit online

To synchronize the various WebSEAL servers, first select a master server. For example, default-webseald-master.ibm.com houses the generic configuration. Any configuration changes that you want made available to all servers must first be manually changed for the master server that includes the configuration files and junction definitions.

About this task

You must only modify the configuration information on the master. If you modify the configuration on a replica, you risk losing information during the next restart of the server when the replica synchronizes its configuration information with the master.

When you configure a WebSEAL cluster in your environment, you can issue a cluster restart server task command from the master server to apply any configuration changes and restart the updated servers. You can use the -ripple option to specify whether to restart the clustered WebSEAL servers in parallel or in sequence. There is also a -status option that you can use to monitor the progress of a cluster restart. See server task cluster restart.

Figure 1. Cluster Support
Cluster restart

Figure 1 shows the cluster restart process that involves the following high-level steps:

Procedure

  1. Run the cluster restart server task command on the master server to initiate the restart.
  2. The master issues a server restart command on each of the replicas in the cluster.
    Note: This is either done in parallel or in sequence depending on whether the -ripple option was used in the cluster restart command.
  3. The replica retrieves the timestamp that indicates when the master configuration information was last modified.
  4. If the configuration data stored locally on the replica is not up to date, then the replica synchronizes with the master configuration data.
  5. If the replica configuration data was updated in step 4, then the replica server restarts. If the local configuration database on the replica did not need updating, then the replica does not restart.
  6. The replica notifies the master that its operation is complete.
  7. Once all replicas are updated and restarted where required, the master server restarts.