Domain Manager Fail over

Document ID : KB000026336
Last Modified Date : 14/02/2018
Show Technical Document Details

Question:

What is the recommended Domain Manager Failover approach?

 

Environment:                                          

CA Client Automation - All Versions

 

Answer:

The failover support feature is available for Windows operating environments only:

Here are the steps to create a failover setup on a VMWare server:

  1. Create a shared disk (This disk has to be connected to SCSI bus and not to SATA / IDE).
  2. Create a quorum disk.
  3. Enable reservation protocol on Shared Disk (SCSI -2 protocol).
  4. Now create a node and while creating the node add the shared and quorum disks to the node. Also create 2 NICs (one custom used for external world and another used as heart-beat NIC between the two nodes).
  5. Enable the SCSI-2 reservation protocol on node1.
  6. After node 1 is up, run sysprep (This is located on the Windows installation CD).
  7. Create a clone copy of node1 (you can use VMWare?vdiskmanager tool for this).
  8. After the clone is ready, add the two disks (shared and Quorum disks) to the node and enable reservation protocol on this node too.
  9. Note that there should be an ADS running separately (on separate box).
  10. Install ADS on node1 and make Node1 as additional Domain Controller to existing ADS. You should not install ADS on any of the nodes in the cluster (means which handle HA) as CCS cannot be installed on Domain Controllers.
  11. Similarly, install ADS on Node2 and make Node2 as additional Domain Controller for the existing ADS.
  12. Create an account on the ADS box and use this account for cluster service.
  13. Now install MS-cluster on both the nodes.
  14. Using Cluster Administrator tool, create a cluster on node1 and check whether all the resources are identified properly or not.
  15. Repeat steps 13 and 14 on Node2 as well.
  16. Restart both the boxes.
  17. Notice that when Node1 is accessing the Shared and Quorum disks node2 should not get access and vice-versa.
  18. Now Install SQL on Node1. Before installing SQL on Node1 check whether IIS is installed on both the nodes of the cluster, verify whether name resolution is OK and network share access should be working fine from both the nodes of the cluster. Also create an SQL In user and add it to the admin group.
  19. Start installation of SQL.
  20. In the "Components to Install Screen" select check SQL Server Database Services , Create a SQL Server failover cluster and Workstation components, Books Online and Development tools options.
  21. In the Feature Selection Screen Click on Database Services option and then select Data Files option, select Browse and select the Shared Disk (we need to store data files on the Shared Disk).
  22. In the Virtual Server Name screen mention virtual server name and select appropriate Group in the Cluster Group Selection screen and proceed with the installation.
  23. SQL will be installed on both the nodes and data files will be installed on shared disk.
  24. For CA Client Automation installation refer the implementation guide.

Following are some important facts taken from the implementation guide:

Failover Support

The failover support feature as described above is available for Windows operating environments only.
Failover supports installation of the manager components on two different computers running Microsoft Cluster.
One system needs to be the active node running the manager functionality and the other (or maybe others) are standby passive systems, having the manager software installed but not activated.
Within the installer's Configure Manager dialog there is a button called 'Recovery' which leads into the failover dialog.
In this dialog the 'Enable Recovery Support' and 'Active' options must be selected to install an active CA Client Automation Manager. For a cluster installation the installation path for the configuration data must point to the shared disk of the cluster. The Shared Components should be placed on a local disk.
CA Client Automation has no functionality in place to detect system failures and to automatically switch from the active to the passive node in case of a system failure. This needs to be done manually or to be configured manually within the cluster software being in use.
If you have installed active and passive manager systems and really run into a hardware problem where you must switch over from the active to the passive manager system, it is important to notify the passive manager about it. You notify the manager by running the ActivateManagerNode.bat file which is located in the bin directory at the installation location. If you run a cluster management system you can also configure the cluster manager to execute the content of the .bat file.
If a domain manager is installed on a cluster system, the CCS installation is performed non-silent.
Before installing a domain manager on a cluster system with a non-default named SQL instance check that the TCP/IP port numbers of the instance are the same on all nodes of the cluster. If this is not the case, modify the port numbers accordingly.
If you want to install a CA Client Automation Domain Manager with a local Microsoft SQL Server in a Microsoft Cluster environment, CA Client Automation must be installed into the Microsoft SQL Server Cluster Group.

Note: Currently, boot server is not supported on a cluster.