You must install CloudBees Flow components on all the nodes in your CloudBees Flow cluster. Where you install the individual components depends on the type of cluster configuration you need to create. For more information on how to install CloudBees Flow, see Installing CloudBees Flow.
Use the reliability approach if you want to minimize single points of failure in your CloudBees Flow installation; use the performance approach if (in addition to minimizing single points of failure), you want to maximize throughput of your CloudBees Flow server at the cost of using more hardware.
In the reliability approach, other CloudBees Flow components such as agents, repositories, and web servers are placed on the same machine as a node of the CloudBees Flow server cluster; in the performance approach, they are placed on other servers to leave as many resources as possible available for the CloudBees Flow server node.
Choose one of the following four installation approaches for your environment:
For any of these approaches, when you install agent, repository, and web server services, you can save time by configuring the software to point to a remote server location. You must also register all of these service agents as resources on the CloudBees Flow server. For more information, see Duplicating Repository Contents .
-
For a command-line installation, set the option
--remoteServer <load_balancer_FQDN>. If you are doing an advanced installation on Linux, when prompted for the remote CloudBees Flow server, enter the <load_balancer_FQDN> . -
For a graphical user interface installation, set the Server Host Name field in the "Remote CloudBees Flow server" installer page to < load_balancer_FQDN >:8000. The load_balancer_FQDN is the fully qualified domain name of your CloudBees Flow server’s load balancer machine.
For details about the overall steps for installing DevOps Insight on a group of servers to create a DevOps Insight server cluster, see Installing the DevOps Insight Server in Cluster Mode .
New CloudBees Flow Installation for Reliability
The reliability approach allows multiple CloudBees Flow services to run on a machine, but multiple instances of the service should exist to prevent single points of failure.
-
Install the CloudBees Flow server and agent software on one node in the CloudBees Flow cluster.
If you do not already have a CloudBees Flow web server that you can temporarily point at this CloudBees Flow server node, you might want to also install a CloudBees Flow web server that can be used for the following two steps in this section. Before you install the CloudBees Flow server and agent software on the remaining nodes in the CloudBees Flow cluster, turn off the web server. You turn off the web server on Linux by using the command /etc/init.d/commanderApache stop, or on Windows by stopping the service and setting the Startup Type to Manual. -
Configure CloudBees Flow to use an external database. At this time, the CloudBees Flow node is in a single-server configuration.
For more information, see Switching to an Alternate Database .
-
Move the plugins directory on the CloudBees Flow server software node to a location on the shared file system.
For more information, see Universal Access to the Plugins Directory and Moving the Plugins Directory to a Pre-Configured Network Location.
-
Install the CloudBees Flow server and agent software on the remaining nodes in the CloudBees Flow cluster.
-
Install the CloudBees Flow repository service on one or more machines.
-
Register agents on these machines as resources on the CloudBees Flow server.
-
Install the CloudBees Flow web server service on one or more machines.
New CloudBees Flow Installation for Performance
The performance approach requires separate machines for each CloudBees Flow service.
-
Install just the CloudBees Flow server software on all the nodes in the CloudBees Flow cluster.
For more information, see Silent Unattended Installation Method .
If you do not already have a CloudBees Flow web server that you can temporarily point at this CloudBees Flow server node, you may also want to install a CloudBees Flow web server that can be used for the following two steps in this section. Before you install the CloudBees Flow server and agent software on the remaining nodes in the CloudBees Flow cluster, turn off the web server . You turn off the web server on Linux by using the command /etc/init.d/commanderApache stop, or on Windows by stopping the service and setting the Startup Type to Manual. -
Configure one instance of the CloudBees Flow server software to use an external database.
At this time, the CloudBees Flow node is in a single-server configuration.
For more information, see Switching to an Alternate Database .
-
Move the plugins directory on the CloudBees Flow server software node to a location on the shared file system.
For more information, see Universal Access to the Plugins Directory and Moving the Plugins Directory to a Pre-Configured Network Location.
-
Install the following software services on one or more individual machines.
Each service should not be installed with any other CloudBees Flow software components. CloudBees Flow agent CloudBees Flow repository server ** CloudBees Flow web server
-
Remove any agents that were automatically installed with the CloudBees Flow server.
For more information, see Separating Agents from CloudBees Flow Servers and Verifying CloudBees Flow Services .
Converting an Existing CloudBees Flow Installation for Reliability
Because this is a conversion of an existing CloudBees Flow system, one or more machines with the CloudBees Flow server, agent, web server, and repository software already exist. The reliability approach allows multiple CloudBees Flow services to run on a machine, but multiple instances of the service should exist to prevent single points of failure.
-
Upgrade the existing CloudBees Flow software according to the instructions in Roadmap to Upgrade CloudBees Flow.
Horizontal scalability is supported starting with CloudBees Flow 5.0.
-
Verify that CloudBees Flow is pointing to an external database.
To verify which database is in use: sign in to CloudBees Flow and select Administration > Database Configuration to see the current database.
-
The database connection is successfully configured if you can log into CloudBees Flow.
-
See Switching to an Alternate Database if additional configuration is required.
-
-
Verify that CloudBees Flow is configured to use a plugins directory located on the shared file system.
For more information, see Universal Access to the Plugins Directory.
-
Install the CloudBees Flow server and agent software on the remaining nodes for the CloudBees Flow cluster.
-
Install the CloudBees Flow repository server on one or more machines.
-
Install the CloudBees Flow web server on one or more machines.
-
Register the machine agents as resources on the CloudBees Flow server.
Converting an Existing CloudBees Flow Installation for Performance
Because this is a conversion of an existing CloudBees Flow system, one or more machines with the CloudBees Flow server, agent, web server, and repository software already exist. The performance approach requires separate machines for each CloudBees Flow service.
-
Upgrade the existing CloudBees Flow software according to the instructions in Roadmap for the Upgrade Process .
Horizontal scalability is supported starting with CloudBees Flow 5.0.
-
Verify that CloudBees Flow is pointing to an external database.
To verify which database is in use sign in to CloudBees Flow and select Administration > Database Configuration to see the current database.
The database connection is successfully configured if you can log into CloudBees Flow.
See Switching to an Alternate Database if additional configuration is required.
-
Verify that CloudBees Flow is configured to use a plugins directory located on the shared file system.
For more information, see Universal Access to the Plugins Directory xref:configure:plugins-directory/plugins-dir.adoc#moving-the-plugins-dir.
-
Remove any web server or agent software that is installed with the original CloudBees Flow machine.
This software will be reinstalled on a separate system.
-
Install the CloudBees Flow server software on the nodes for the CloudBees Flow cluster.
-
Install the following software services on one or more individual machines.
Each machine should not be installed with any other CloudBees Flow software services.
-
CloudBees Flow agent
-
CloudBees Flow repository server
-
CloudBees Flow web server
-
-
Remove any agents that were automatically installed with the CloudBees Flow server, web server, and repository services.
The original CloudBees Flow machine as well as the new installations should be checked to verify that the agent software is removed.
For more information, see Separating Agents from CloudBees Flow Servers and Verifying CloudBees Flow Services .
-
Remove any repository server software that is installed with the original CloudBees Flow machine after you duplicate the repository server contents.
For more information, see Duplicating Repository Contents.
If necessary, install the repository software on additional machines.