4.6 Starting and Stopping the Oracle Directory Integration and Provisioning Server in a High Availability Scenario

The Oracle directory integration and provisioning server can, with certain restrictions, execute in various high availability scenarios. This section discusses the Oracle directory integration and provisioning server as it operates in a Real Application Clusters environment and in an Oracle Application Server Cold Failover Cluster (Infrastructure). It contains these topics

The Oracle Directory Integration and Provisioning Server in a Real Application Clusters Environment
The Oracle Directory Integration and Provisioning Server in an Oracle Application Server Cold Failover Cluster (Infrastructure)

In either type of high availability environment, there are two common scenarios for configuring Oracle Directory Integration and Provisioning. They are:

Collocated–The Oracle directory integration and provisioning server is located within the cluster on the same node as Oracle Internet Directory.
Outside the cluster–The Oracle directory integration and provisioning server is installed on a separate node, outside the cluster.

4.6.1 The Oracle Directory Integration and Provisioning Server in a Real Application Clusters Environment

The Oracle Internet Directory infrastructure is configured to work in a Real Application Clusters mode. In Real Application Clusters, the Oracle directory integration and provisioning server can execute against any directory node.

A particular configuration set can be executed by only one instance of the Oracle directory integration and provisioning server. For this reason, during the default installation only one server instance—namely, instance 1—is started on the Real Application Clusters master node. This server instance executes configuration set 0. Although it is started only on the master node, the server is nevertheless registered on all the nodes.

If the master node fails, then the Oracle directory integration and provisioning server instance is started by the OID Monitor on a secondary node. If there are multiple secondary nodes, then the server is started by the first OID Monitor to recognize the master node failure.

When it starts the server, the OID Monitor uses the same instance number and configuration set that was used on the master node. This is a transparent to the end user, and, once it is done, the Oracle directory integration and provisioning server on the secondary node behaves as if it is the primary server. The server continues executing on the secondary node as long as that node is available.

Two separate instances of the Oracle directory integration and provisioning server running on two nodes cannot simultaneously execute the same configuration set. Although the OID Monitor does not check for this, the Oracle directory integration and provisioning server itself fails to start.

You can stop the Oracle directory integration and provisioning server at any time by using the OID Control utility. However, if you do this, then the server does not start automatically on any other node. To start it on another node, do so manually by using the OID Control utility.

If you execute the command opmnctl stopall, and subsequently execute opmnctl startall, then the Oracle directory integration and provisioning server starts.

In summary, unless an OID Control command stops the Oracle directory integration and provisioning server, the OID Monitor always ensures that the server is running.

4.6.1.1 Collocated Configurations

In a collocated configuration, you can start Oracle Directory Integration and Provisioning from any node in the cluster. Once the Oracle directory integration and provisioning server is started on the first node, you do not need to start it on any other node. On failure of the Oracle Directory Integration and Provisioning node, another node in OracleAS Cluster (Identity Management) will detect the failure and start the Oracle directory integration and provisioning server. No additional OID Control command is required to register the Oracle directory integration and provisioning server.

In most cases, the Oracle Directory Integration and Provisioning server communicates with only the single, default instance of the Oracle directory server. It is possible, however, to have manually configured the Oracle directory integration and provisioning server to communicate with a second instance of the Oracle directory server. If the second instance of the Oracle directory server is not configured on the other nodes, then on failover, the surviving node will start both Oracle Directory Integration and Provisioning and a second instance of the Oracle directory server.

In a collocated configuration, node failure is handled as follows: the OID Monitor on a surviving node keeps polling all other nodes every 10 seconds. When a node detects that one node is not responding, the OID Monitor on the surviving node starts the Oracle directory integration and provisioning server and possibly the LDAP server (if it is not on the default node).

4.6.1.2 Outside-the-Cluster Configurations

In an outside-the-cluster configuration, the Oracle directory integration and provisioning server node does not have failover capability. In this configuration, you can configure Oracle Directory Integration and Provisioning to connect to the Oracle Internet DirectoryLDAP server using a load balancer or virtual server in front of the multiple Oracle Internet Directory nodes.

4.6.2 The Oracle Directory Integration and Provisioning Server in an Oracle Application Server Cold Failover Cluster (Infrastructure)

In this configuration, you should start the Oracle directory integration and provisioning server with a virtual hostname. This is the default configuration on installation.

If the active node fails, then the OID Monitor on a standby node starts the Oracle directory integration and provisioning server instance on the standby node. When it does this, it uses the same instance number and configuration set as previously used on the active node. This is a transparent to the end user. The server continues executing on the active node as long as the node is available. In an Oracle Application Server Cold Failover Cluster (Infrastructure), the server is registered once for both the active and standby nodes because the virtual host names are the same for both.

You can stop the Oracle directory integration and provisioning server at any time by using the OID Control utility. However, if you do this, then the server does not start again on this node. Moreover, if this node fails over, then the OID Monitor on the standby node does not start the Oracle directory integration and provisioning server. To start the server, you must use the OID Control utility.

If you execute the command opmnctl stopall, and subsequently execute opmnctl startall, then the Oracle directory integration and provisioning server starts.

In summary, unless an OID Control command stops the Oracle directory integration and provisioning server, OID Monitor always ensures that the server is running.

4.6.2.1 Collocated Configurations

In a collocated configuration, start the Oracle Directory Integration and Provisioning server using this command line:

oidctl connect=connStr host=virtualHost server=odisrv instance=1 \
      flags="host=virtualHost port=OIDPORT" start

4.6.2.2 Outside-the-Cluster Configurations

In an outside-the-cluster configuration, use this command to start the Oracle Directory Integration and Provisioning server:

oidctl connect=connStr server=odisrv instance=1 \
      flags="host=OIDvirtualHost port=OIDPORT" start

Note:

There are two host parameters in the command-line examples for the collocated and outside-the-cluster configurations:

The host parameter outside the flags specifies the node where the OID Control utility runs and originates requests to the OID Monitor
The host parameter inside the flags specifies the LDAP server that the Oracle Directory Integration and Provisioning and replication servers should connect to. This parameter is valid only for those servers.