Chapter 1. Introduction to cloud-init

cloud-init is a software package that automates the initialization of cloud instances during system boot. You can configure cloud-init to perform a variety of tasks. Some sample tasks that cloud-init can perform include:

  • Configuring a host name
  • Installing packages on an instance
  • Running scripts
  • Suppressing default virtual machine (VM) behavior

Where you obtain your image for configuring cloud-init depends on how you intend to use it.

  • The cloud-init package is installed on KVM Guest Images that you download from the Red Hat Customer Portal. When you launch an instance, cloud-init is enabled. KVM Guest Images that you download from the Red Hat Customer Portal are intended for use with Red Hat Virtualization (RHV), the Red Hat OpenStack Platform (RHOSP), and Red Hat OpenShift Virtualization.
  • You can also download a RHEL ISO image from the Red Hat Customer Portal to create your own custom guest image. In this case, you need to install the cloud-init package on your guest image yourself.
  • If you plan to use an image with a cloud provider (for example, AWS or Azure), use Red Hat Image Builder to create the image. Image Builder images are customized for use for specific cloud providers. The image types AMI, VHD, and qcow2 include cloud-init already installed. Refer to Composing a Customized RHEL System Image for information on Image Builder.

Most cloud platforms support cloud-init, though configuration procedures and supported options vary. Alternatively, you can configure cloud-init for a NoCloud environment.

You can configure cloud-init on one VM and then use that VM as a template for additional VMs or clusters of VMs.

Specific Red Hat products (for example, Red Hat Virtualization) have documented procedures for configuring cloud-init for use with those products.

This document refers to the cloud-init documentation in a number of places. Refer to the referenced cloud-init documentation for complete information on cloud-init.


1.1. cloud-init configuration

cloud-init uses YAML-formatted file instructions to perform tasks. You decide the initial configuration you want cloud-init to perform by providing instructions within the YAML files. When an instance boots, the cloud-init service starts and searches for and executes the instructions. Tasks complete during the first boot or on subsequent boots of your VM, based on your cloud-init configuration.

You define the tasks by configuring the /etc/cloud/cloud.cfg file and adding directives under the /etc/cloud/cloud.cfg.d/ directory.

  • The cloud.cfg file includes directives, such as those for user access and authentication and system information.

    The file also includes default and optional modules for cloud-init. The modules are executed in order within three phases that include the cloud-init initialization phase, the configuration phase, and the final phase. Within the cloud.cfg file, modules for the three phases are listed under cloud_init_modules, cloud_config_modules, and cloud_final_modules, respectively.

  • The cloud.cfg.d directory is where you can add additional directives for cloud-init. When you add directives to the cloud.cfg.d directory, you typically add them to a file named *.cfg, and you always include #cloud-config at the top of the file.

1.2. cloud-init operates in stages

cloud-init operates in five stages during a system boot. Those stages determine whether cloud-init runs and where it finds its datasources, among other tasks. The stages are as follows:

  1. The cloud-init generator stage, through the systemd service, determines whether to run cloud-init upon the boot.
  2. During the local stage, cloud-init finds local datasources and applies network configuration.
  3. During the network stage, cloud-init processes user data and runs the modules listed under cloud_init_modules in your cloud.cfg file. You can enable, disable, or add modules to the cloud_init_modules section.
  4. During the config stage, cloud-init runs the modules listed under cloud_config_modules in your cloud.cfg file. You can enable, disable, or add modules to the cloud_config_modules section.
  5. During the final stage, cloud-init can run what you have included under cloud_final_modules in your cloud.cfg file. You can include package installations that you would typically run after a system boots and can also include configuration management plug-ins and user scripts. You can enable, disable, or add modules to the cloud_final_modules section.

The five boot stages are described in the cloud-init Documentation section Boot Stages.

1.3. cloud-init modules execute in phases

When cloud-init runs, it executes the modules within cloud.cfg in order within three phases:

  1. The network phase (cloud_init_modules)
  2. The configuration phase (cloud_config_modules)
  3. The final phase (cloud_final_modules)

When cloud-init runs for the first time on a VM, all the modules you have configured run in their respective phases. On a subsequent running of cloud-init, whether a module runs within a phase depends on the module frequency of the individual module. Some modules run every time cloud-init runs; some modules only run the first time cloud-init runs, even if the instance ID changes.


An instance ID uniquely identifies an instance. When an instance ID changes, cloud-init treats the instance as a new instance.

The possible module frequency values are as follows:

  • Per instance means that the module runs on first boot of an instance. For example, if you clone an instance or create a new instance from a saved image, the modules designated as per instance run again.
  • Per once means that the module runs only once. For example, if you clone an instance or create a new instance from a saved image, the modules designated per once do not run again on those instances.
  • Per always means the module runs on every boot.

You can override a module’s frequency when you configure the module or by using the command line.

1.4. cloud-init acts upon user data, metadata, and vendor data

cloud-init consumes and acts upon user data, metadata, and vendor data.

  • User data includes directives you specify in the cloud.cfg file and in the cloud.cfg.d directory, for example, user data can include files to run, packages to install, and shell scripts. Refer to the cloud-init Documentation section User-Data Formats for information on the types of user data that cloud-init allows.
  • Metadata includes data associated with a specific datasource, for example, metadata can include a server name and instance ID. If you are using a specific cloud platform, the platform determines where your instances find user data and metadata. Your platform may require that you add metadata and user data to an HTTP service; in this case, when cloud-init runs it consumes metadata and user data from the HTTP service.
  • Vendor data is optionally provided by the organization (for example, a cloud provider) and includes information that can customize the image to better fit the environment where the image runs. cloud-init acts upon optional vendor data and user data after it reads any metadata and initializes the system. By default, vendor data runs on the first boot. You can disable vendor data execution.

    Refer to the cloud-init Documentation section Instance Metadata for a description of metadata; Datasources for a list of datasources; and Vendor Data for more information on vendor data.

1.5. cloud-init identifies the cloud platform

cloud-init attempts to identify the cloud platform using the script ds-identify. The script runs on the first boot of an instance.

Adding a datasource directive can save time when cloud-init runs. You would add the directive in the /etc/cloud/cloud.cfg file or in the /etc/cloud/cloud.cfg.d directory. For example:


Beyond adding the directive for your cloud platform, you can further configure cloud-init by adding additional configuration details, such as metadata URLs.

datasource_list: [Ec2]
    metadata_urls: ['']

After cloud-init runs, you can view a log file (run/cloud-init/ds-identify.log) that provides detailed information about the platform.

1.6. Additional resources