Installation Script

The same installation script is used in all of the installation methods. This is either a bash script on Linux or a PowerShell script on Windows. The script is agent-specific, and a separate script is rendered from a template for each agent during the install workflow. The script downloads the agent package from the manager (over port 53333), extracts it on the agent host, creates a daemon and starts it.

Installation Methods

There are several methods by which the installation script is distributed to the agent host and executed.

remote (default)

In this method, the installation script is pushed to the agent host, and executed remotely using SSH on Linux hosts and WinRM on Windows hosts. This is the default and simplest way for systems that allow SSH/ WinRM access. The pre-requisites for remote installation are:

The output (both standard output and standard error streams) generated during the agent installation process is accumulated in memory, and echoed to Cloudify’s logger once the agent installation is completed.


For systems that don’t have SSH/ WinRM access, userdata (e.g. cloud-init) may be used. In this method, an install script is rendered, and a temporary link pointing to it is created. This link is then embedded in a separate download script, which is injected into the agent host’s userdata. When the host is booted for the first time, the download script is executed. It downloads the installation script from the temporary link and executes it. To use the init_script method, the IaaS provider and Cloudify plugin need to support userdata. Currently, the Openstack, AWS and AWS-SDK plugins support this installation method.

The output generated during the agent installation process is not echoed to Cloudify’s logger; instead, you can find it in:


For systems that have neither SSH/ WinRM access, nor the ability to inject a script into userdata, the plugin installation method allows implementing a custom way to download and execute the installation script. In this method, a temporary download link for the installation script is created and made available to the plugin using the ctx. It is up to the plugin’s developer then to implement a custom method for downloading and executing the installation script on the agent’s host.


For systems that require the agent to be embedded in the image, users can use the provided method. In this mode, it is up to the user to make sure the agent is already installed on the image.
During the install workflow, a configuration script will be rendered and a temporary link to it will created and made available via Cloudify’s logs (reading the logs is the only way to retrieve the temporary link). This script is similar to the installation script, except that it doesn’t download or install the agent package, but only configures and starts the agent daemon. This script needs to be downloaded and executed manually for every agent.

Provided Agent Step-by-Step-Guide

For the purposes of this example we will be using a manager IP of and deploying an agent for the node with node instance ID container_nwxqiu.

  1. Prepare a Virtual Machine or Image with a custom agent. Either Prepare a custom agent or just upload an agent archive from:
  2. Install the blueprint:
        type: cloudify.nodes.Compute
            install_method: provided
            user: centos
  3. Find out the node instance name using cfy node-instances list in the CLI, or through the Deployments -> Deployment Info -> Deployment Nodes widget in the UI:
    Node instance
  4. Get the configuration script link from the runtime properties of the node instance, or form the deployment logs:
    Script in logs
  5. Login to the agent’s VM.
  6. Prepare the agent directory
    sudo mkdir -p /opt/cloudify-agent-6.3.0/
    sudo chown $(whoami) /opt/cloudify-agent-6.3.0/
    mkdir -p /opt/cloudify-agent-6.3.0/container_nwxqiu/cloudify/ssl
  7. Untar the archive to a folder with the same as the node instance. (“container_nwxqiu” in the above example):
    tar xzfv centos-core-agent.tar.gz --strip=1 -C /opt/cloudify-agent-6.3.0/container_nwxqiu
  8. Download the configuration script using the link from step 4:
    wget --no-check-certificate --user admin --password admin
  9. The script doesn’t copy the SSL certificate, so just do it manually — copy cloudify_internal_ca_cert.pem from the manager into /opt/cloudify-agent-6.3.0/container_nwxqiu/cloudify/ssl/cloudify_internal_cert.pem in the VM, e.g. using scp.
  10. Run the downloaded script:
    chmod 700
    sudo ./

The Agent has now joined to the Cloudify Manager. Note: It is recommended to automate the above steps if used often.


In some cases, the user cannot or prefers not to install an Agent on Cloudify-managed VMs. This might be due to a security restriction, or because a VM is a pre-configured closed appliance that the user cannot access or modify.

This has the following implication:

Specifying the Installation Method

To specify the installation method that will be used, set the install_method key in the agent_config structure. This can be done in various ways, as described in the agent configuration documentation.

For example, you can set the install_method as a property on the compute node’s template as follows:

    type: cloudify.nodes.Compute
        install_method: remote