Installing agents from the command line
Installing an agent from the command line involves running a batch file or shell script and specifying information about how the agent connects to the server or agent relay.
- For production environments, create a user account that has the required privileges for the
installation.
If the user account does ot have the necessary privileges, the installation fails.
For AIX, HP, Solaris, Windows and Mac OS, the user account must have access to all working directories used in the process the agent runs.
For Linux and Unix OS, the user account myust have root privileges in order to issue the su or sudo command when installing.
For z/OS environments, review the security requirements for the IBM z/OS computers.
- Make sure that your networks and firewalls allow communication on the required ports. See Firewall and communication configuration.
-
Ensure that each computer where you are installing an agent has a supported version of the Java™ Runtime Environment (JRE) or Java developer kit. For more information, see Software Product Compatability Reports.
-
If the target computer has an IBM JRE or Java developer kit, update to a supported version of the IBM JRE or Java developer kit.
- Set the JAVA_HOME system variable to the location of the JRE or JDK that you are using.
- Ensure that there is no limit on the maximum memory size and virtual memory size.
For example, on most Linux™ systems, you can run the
ulimit -m
andulimit -v
commands and ensure that both return the valueunlimited
. To find out how to remove the limit on the maximum memory size and virtual memory size, see the documentation for the operating system.
Agents can also be installed directly from the IBM UrbanCode Deploy web application, see Installing agents remotely.
For simple evaluations, the administrative user can run the agent on the computer where the server is located. Because agent workloads can be resource-intensive, for production environments, do not install agents on the computer where the IBM UrbanCode Deploy server is located. If you plan to run deployments on several target servers, install a separate agent on each one. If, for example, your testing environment consists of three servers, install an agent on each one. Follow the same procedure for each environment the application uses.
Each agent requires the appropriate rights to communicate with the IBM UrbanCode Deploy server.
At a minimum, each agent must have these permissions:
- Open a TCP connection. A JMS agent uses a TCP connection to communicate with the server's JMS port, and a Web agent uses one for its WebSocket connection.
- Open a HTTP(S) connection. The agent must be able to connect to the IBM UrbanCode Deploy user interface to download artifacts from the CodeStation repository.
- Access the file system. Many agents need read/write permissions to items on the file system.
If the agent communicates with IBM UrbanCode Deploy by using an agent relay, you can install the agent from the command line.