Control-M/EM UNIX Cluster Configuration

The Control-M/EM UNIX installation is not cluster aware. Therefore, you need to configure the UNIX cluster manager to manage Control-M/EM components.

Before you can configure Control-M/EM in a cluster environment on UNIX, you must do the following:

BMC provides demo scripts that start, check, and stop Control-M/EM components. You can build your own scripts and use the provided scripts as an example or update the scripts to match your installation. After your scripts are ready, you can configure the cluster manager to use the scripts.

The following demo scripts are located at <Control-M/EM_HOME>/bin:

  • demo_ha_cluster_start

  • demo_ha_cluster_check

  • demo_ha_cluster_stop

  • demo_ha_cluster (Main script)

To get the usage, run the scripts with –help (or with no parameter).

The following procedures describe how to configure Control-M/EM in a clustered environment on UNIX using BMC provided demo scripts:

Modifying the Demo Ha Cluster Script

This procedure describes how to modify the demo ha cluster script with the Control-M/EM home directory and the UNIX account name. This enables you to provide the cluster manager with the required scripts to start, check, and stop components.

Begin

  1. Navigate to the following directory:

    <Control-M/EM home dir>/ bin

  2. Open the demo_ha_cluster script.

  3. In the User definition section, do the following:

    1. Change the EM install directory to the location where Control-M/EM is installed.

      EM_DIR="/data/emuser/ctm_em"

    2. Change the UNIX install account name to the Control-M/EM user account that you created (see Configuring a User Account on UNIX).

      EM_ACC="emuser"

Configuring the Control-M/EM Configuration Agent as a Cluster Resource

This procedure describes how to configure the Control-M/EM Configuration Agent as a cluster resource using the BMC provided demo scripts. You must perform this procedure to activate your cluster environment.

Begin

  • From your cluster manager, add the following commands:

    • Start: demo_ha_cluster_start -C Config_Agent

    • Check: demo_ha_cluster_check -C Config_Agent

    • Stop: demo_ha_cluster_stop -C Config_Agent

  • The Configuration Agent is now managed by the cluster manager. If there is a failover/move, the cluster manager starts up the Configuration Agent on the other host.

Configuring Control-M/EM Kernel Components as Cluster Resources on UNIX

This procedure describes how to configure the CMS, Naming Service, and PostgreSQL database (if installed) components as cluster resources on UNIX.

This procedure is optional. If you do not do it, the Configuration Agent continues to manage the Control-M/EM components.

Begin

  1. Set the system parameter ClusterInstall to 1, as described in General parameters.

    The CMS, Naming Service, and PostgreSQL database components are no longer managed by the Configuration Agent.

  2. From your cluster manager, add the start, check, and stop commands, as described in Control-M/EM UNIX Cluster Commands.

    The Control-M/EM components are now managed by the cluster manager.

Configuring Additional Control-M/EM Components as Cluster Resources on UNIX

This procedure describes how to configure additional Control-M/EM components as cluster resources on UNIX using the BMC provided demo scripts.

This procedure is optional. If you do not do it, the Configuration Agent continues to manage the Control-M/EM components.

Begin

  1. From the CCM, set the following components to Ignore, as described in Ignoring a component:

    Gateway

    GUI Server

    Web Server

    GCS

    SLA Management

    Forecast Server

    Self Service Server

  2. From your cluster manager, enter the start, check, and stop commands, as described in Control-M/EM UNIX Cluster Commands.

    The Control-M/EM components are now managed by the cluster manager.

Control-M/EM UNIX Cluster Commands

The following table lists the start, check, and stop commands for each Control-M/EM component.

The demo_ha_cluster_check command returns a 0 value if the component is up and a different value if the component is down.

Control-M/EM component

Commands

CMS

  • Start: demo_ha_cluster_start -C CMS

  • Check: demo_ha_cluster_check -C CMS

  • Stop: demo_ha_cluster_stop -C CMS

Naming Service

  • Start: demo_ha_cluster_start -C Naming_Server

  • Check: demo_ha_cluster_check -C Naming_Server

  • Stop: demo_ha_cluster_stop -C Naming_Server

PostgreSQL Database

  • Start: demo_ha_cluster_start -C DBPostgres

  • Check: demo_ha_cluster_chec -C DBPostgres

  • Stop: demo_ha_cluster_stop -C DBPostgres

Gateway

  • Start: demo_ha_cluster_start -C Gateway -dc <Data Center Name>

  • Check: demo_ha_cluster_check -C Gateway -dc <Data Center Name>

  • Stop: demo_ha_cluster_stop -C Gateway -dc <Data Center Name>

GUI Server

  • Start: demo_ha_cluster_start -C GUI_Server -name <Logical Name>

  • Check: demo_ha_cluster_check -C GUI_Server -name <Logical Name>

  • Stop: demo_ha_cluster_stop -C GUI_Server -name <Logical Name>

GCS

  • Start: demo_ha_cluster_start -C GCS

  • Check: demo_ha_cluster_check -C GCS

  • Stop: demo_ha_cluster_stop -C GCS

Web Server

  • Start: demo_ha_cluster_start -C Web_Server

  • Check: demo_ha_cluster_check -C Web_Server

SLA Management (BIM)

  • Start: demo_ha_cluster_start -C BIM

  • Check: demo_ha_cluster_check -C BIM

  • Stop: demo_ha_cluster_stop -C BIM

Forecast Server

  • Start: demo_ha_cluster_start -C Forecast_Server -name <Logical Name>

  • Check: demo_ha_cluster_check -C Forecast_Server -name <Logical Name>

  • Stop: demo_ha_cluster_stop --C Forecast_Server -name <Logical Name>

Self Service Server

  • Start: demo_ha_cluster_start -C Self_Service_Server

  • Check: demo_ha_cluster_check -C Self_Service_Server

  • Stop: demo_ha_cluster_stop --C Self_Service_Server

Control-M/EM Configuration Agent

In some cases, after setting a virtual host name online, the physical host name of the cluster node becomes unavailable. In such case, additional configuration is required for the Control-M/EM Configuration Agent. This configuration provides the ability to send life check communications to the Control-M/EM Configuration Agent using the virtual host name.

This configuration should only be used if the Control-M/EM Configuration Agent appears in the Control-M Configuration Manager as Not Responding after starting the Control-M/EM Configuration Agent.

Begin

  1. Stop the Control-M/EM Configuration Agent from root_menu or by running the command stop_config_agent from the shell prompt (this requires the Control-M/EM DBO password).

  2. In the Control-M Configuration Manager, choose Tools > System Parameters.

  3. In the general section, select the HostPort parameter and click New to open the "Add New Parameter" dialog box.

  4. In the Add New Parameter dialog box, click Advanced.

  5. Modify the following field values:

    Field

    Value

    Type

    General (the default value)

    Name

    HostPort (the default value)

    Value

    vhemxxx:0, which is specified in the format:

    virtualHostName:port

    By specifying 0 (zero) as the port number, the Control-M/EM Configuration Agent will listen on a random port that is chosen when the Control-M/EM Configuration Agent starts. If a port other than zero is specified, the Control-M/EM Configuration Agent will use the specified port number.

    Advanced field: Type

    Config Agent

    Advanced field: Name

    * (the default value)

    Advanced field: Host

    * (the default value) Computer hostname

    If the installation of the Control-M/EM components is distributed between additional hosts other than the cluster nodes, a separate condition must be specified with the physical host names of each cluster node specified in the Host Name parameter.

  1. Click Save.

  2. Start the Control-M/EM Configuration Agent.

Global Condition Server

In some cases, after setting a virtual host name online, the physical host name of the cluster node becomes unavailable. In such case, additional configuration is required for the GCS. This configuration should only be used if the GCS failed to start after a failover.

Begin

  1. Start the Control-M Configuration Manager and change the desired state of the GCS to DOWN.

  2. Ensure that the GCS is not running on any of the cluster nodes (ps -ef | grep gcs).

  3. In the Control-M Configuration Manager, choose Tools => System Parameters.

  4. In the general section, select the HostPort parameter and click New to open the "Add New Parameter" dialog box.

  5. In the Add New Parameter dialog box, click Advanced.

  6. Modify the following field values:

    7. Field

    Value

    Type

    General (the default value)

    Name

    HostPort (the default value)

    Value

    vhemxxx:55443, which is specified in the format:

    virtualHostName:port

    Do not specify 0 as the port number.

    Advanced field: Type

    GCS

    Advanced field: Name

    * (the default value)

    Advanced field: Host

    * (the default value)

  1. Click Save.

  2. Start the GCS using the Control-M Configuration Manager.