The high availability (HA) deployment manager function can be configured
to eliminate single point of failure for administrative functions in a Network
Deployment cell.
Before you begin
This feature is supported only on distributed platforms
and is not available on z/OS.Privileges for the high availability
(HA) deployment manager differ, depending on the various roles. Roles include
monitor, operator, configurator, and administrator. If you are a user with
either a monitor or a configurator role, you can only view the HA deployment
manager information. If you have the role of operator or administrator, you
have all the privileges for the HA deployment manager.
Before you follow
these steps ensure that the system environment has a suitable shared file
system available, such as IBM SAN FS or NFS Version 4. To verify that your
file system is suitable for use with the high availability deployment manager,
use the
IBM File System Locking Protocol Test for WebSphere Application
Server. The shared file system requirements are the same for both the
failover of transaction logs and for the high availability deployment manager.
Each
deployment manager must share the same master configuration repository and
workspace area, which must be on the shared file system. Each deployment manager
must also have read and write permissions to the shared master configuration
repository and workspace area. The master configuration repository and workspace
area default locations are install_root/profiles/deployment_manager/config and install_root/profiles/deployment_manager/wstemp/.
File sharing is typically accomplished by installing
the deployment manager on the shared file system. Sharing the entire installation
is not mandatory, and an alternative is to locate only the deployment manager
profiles on the shared file system. The scenarios described in this topic
require that WebSphere Extended Deployment is installed on the shared file
system.
About this task
Set up a high availability deployment manager environment to eliminate
the deployment manager as a single point of failure. When one deployment manager
fails, a configured deployment manager can resume the work. See
The high availability deployment manager
for
more information.
Procedure
- Choose a topology. The HA deployment manager function
can be configured in a variety of topologies. In a production environment,
it is recommended that each deployment manager runs on its own computer to
increase availability. If a single on demand router (ODR) is in the topology,
it must run on its own computer. To eliminate the ODR as a single point of
failure, you can use multiple ODRs in conjunction with an IP sprayer.
A typical HA deployment manager environment requires at least three
computers:
Computer name |
Processes |
A |
Deployment manager |
B |
Standby deployment manager |
C |
Node agent and on demand router |
- Create a new cell with an HA deployment manager configuration:
- Install WebSphere Application Server on the shared file system.
- Install WebSphere Extended Deployment on the shared file system.
- Create a deployment manager profile using the IP address for
computer A.
- Start the deployment manager on computer A.
- Create a custom profile using the IP address for computer C.
Federate computer C with the deployment manager.
- Create an ODR on the federated custom profile.
- Create a deployment manager profile using the IP address for
computer B.
- Run the xd_hadmgrAdd command within the deployment manager profile
for computer B to convert it into a standby deployment manager. For
more information about the xd_hadmgrAdd command, see xd_hadmgrAdd command
.
- Restart the deployment managers and the ODRs. The first deployment
manager that starts is the active deployment manager, and the other is the
standby deployment manager.
- Convert an existing cell to an HA deployment manager configuration:
- Install WebSphere Application Server on the shared file system.
- Install WebSphere Extended Deployment on the shared file system.
- Create a deployment manager profile using the IP address for
computer A and with the same cell and node name as the existing deployment
manager.
- Use the backupConfig and restoreConfig command line utilities
to relocate the existing cell configuration to the shared file system.
Copy the tmsStorage folder
and its contents from the existing deployment manager profile to the new deployment
manager profile on the shared file system.
- Ensure that the existing cell configuration contains an ODR.
If an ODR does not exist, create one.
- Create a deployment manager profile using computer B's IP address.
- Run the xd_hadmgrAdd command within the deployment
manager profile for computer B to convert it into a standby deployment manager.
- Restart the deployment managers and ODRs. The first deployment
manager that starts is the active deployment manager, and the other is the
standby deployment manager.
Results
Multiple deployment managers are configured. The deployment manager
is eliminated as a single point of failure because a primary deployment manager
hosts the administrative function for the cell, and has backup deployment
managers that are in standby node. If the primary deployment manager fails,
the standby deployment managers can resume without losing work.
What to do next
If you are using WebSphere Extended Deployment
Version 6.0.2 on WebSphere Application Server Version 6.1, you can manage
the deployment manager in the administrative console.
Click System
administration > Deployment manager > All deployment managers to view
and manage the configured deployment managers in your environment.
If
you need to remove a high availability deployment manager from the cell, use
the command line utility to remove the function. See xd_hadmgrRemove command
for
more information.