When you are planning your failover system, consider the following:
-
Ensure that you have superuser (User ID 0) or administrative privileges to perform the procedures to set up a failover system.
-
Failover Manager is supported for Linux platform versions listed in the Support Matrix for Domain Managers.
-
Determine where to install the components of the Failover System.
Products installed in the Active location must have corresponding products installed in the Standby location, except for IP Configuration Manager. Since IP Configuration Manager is not supported for failover, only one instance is allowed.
In this document, all instructions assume that Location A or Side A is the Active site and Location B or Side B is the Standby site. For example, instructions about domain configuration changes on the Active side, assume that the changes are made on the host defined as Site A.
Each Active and Standby component that is a part of the Failover System can be installed on a separate host or multiple components can be installed on one host. At the very least, an Active component and its Standby component should operate from distinct installation areas. Ensure that your hosts have sufficient resources.
-
It is recommended that installation locations are the same in the Active and Standby sides. If that is not possible, set the StrictSitemod parameter to False in the BASEDIR/conf/failover/failover.conf file.
-
The mixing of hardware is not allowed. Failover support is available from one physical box to another physical box, or from one virtual machine to another virtual machine, only if they are running the same operating system.
-
Both, the Active and Standby components must be running the same version of software.
Supported products for failover provides a list of supported products. Support Matrix for Domain Managers provides resource (hardware and memory) information.
-
-
Determine which ports to use for Active and Standby components.
Service Assurance Manager, Adapter Platform, BIM, and all Domain Managers must be started on predefined ports. The ports are also specified in the ServerSection entries in the failover.conf file.
For Linux, the sm_service install --port value for the Broker and the sm_service install --port value and the --sport value for the Trap Exploder must be changed to values greater than 1024. The Failover Manager uses a non-root user account when restarting these components, so privileged ports (below 1024) cannot be used.
-
The Failover Manager software is included in any product installation. No additional installation task for Failover Manager is required. Later, for one of the deployment tasks, you issue a command to manually install the service for the Failover Manager and then start the service.
When deciding on where to run the Failover Manager, consider the following scenario:
-
If the Failover Manager is running with Standby SAM from the host on Location B and the Active SAM on Location A fails, the failover occurs and Standby SAM on Location B is promoted to Active. Then, if the newly-promoted Active SAM host on Location B fails, the Failover Manager will fail with it. You lose the capability to failover.
To avoid losing failover capability, you need to initiate a failback as soon as the failed SAM on Location A is operational.
-
If the Failover Manager is running on a separate dedicated host, you do not lose the capability to failover. The Failover Manager will change the failed SAM host on Location B to Active.
The Failover Manager may reside in a separate location or on the Standby side. Ideally, the Failover Manager should run on a highly available host that is separate from the components it monitors.
-
If you plan to run the Failover Manager on a separate host, you need to install an product on that host.
-
If you do not plan to run the Failover Manager from a separate host, start the Failover Manager service from the host where the Standby SAM is running.
-
-
Determine which ports to use for Active and Standby components.
-
The Failover Manager monitors any number of Brokers you have defined. The Brokers can be running on the same host as the Failover Manager or on different hosts. VMware recommends that the Broker operate on the same host as the Failover Manager. The Broker requires minimal resources and recovers very quickly after a failure.
-
The Active Broker to which the Failover Manager is connected must be specified as the default Broker for every component in the runcmd_env.sh file. The Failover Manager checks for this default Broker when it monitors the components.
-
Run the Linux hostname command for each host. The output is used later when you configure the failover.conf file.
The host name specified in the ServerSection of the failover.conf file must exactly match the name displayed when you run the Linux hostname command. If the host names do not match, the hosts will not be registered with the Failover Manager.