Building your manifest file

Overview

A manifest file allows you to be more explicit about your application composition and control settings that are not usually available through the user interface or Cloud 66 toolbelt. See Getting started with manifest files for an introduction.

For Rails/Rack applications the path for manifest.yml is:

<application-root>/.cloud66/manifest.yml

The .cloud66 folder must in the root of your source code & repo.

Problems upgrading?

The structure of a Manifest file

Manifest files are made up of blocks of settings that define the infrastructural elements of your application. This includes both the “hardware” (virtual or real servers) and the “software” (such as the version of Docker your application uses).

A typical block of settings will contain some combination the following:

• Environment (optional) - the environment to which the settings apply
• Component settings (required) - the settings for the component being configured (e.g. MySQL or an AWS Load Balancer)
• Server definition (optional) - the specifications for the server used by the component defined in the settings above
• Custom log files (optional) - each component can be configured to use custom LiveLog files

You can also control and customize some other elements in your Manifest including:

• Environment variables - you can set environment variables for your app
• Background processes - you can customize the Linux signals used to control your background processes

Important

Environment settings

The (optional) environment setting allows you to use the same manifest.yml for multiple applications with different environments. Some examples are:

• production
• staging
• development

You can also use your own custom environment names in your Manifest file.

Applying settings across all environments

If you would like your manifest setting to apply to all environments, you can simple drop the root level environment node from your YAML. This will ensure a component is always provisioned, regardless of environment.

Component settings

The (mandatory) component settings define how a component is configured. The Manifest file currently supports the following components (click on any component name for a detailed doc):

• ElasticSearch
• Gateway
• GlusterFS
• Load balancers
• Memcached
• MongoDB
• MySQL
• Nginx
• Node version
• PostGIS
• PostgreSQL
• Redis
• Sinatra
• Rails
• Post-deployment global availability checks

Note

Server definitions

Every component defined in the manifest file must be bound to a server. However, if you’d like configurations to apply to all your servers, you don’t need to specify a server type.

Servers can be deployed specifically to host a single component, be shared between multiple components (e.g. Rails and MySQL on the same server) or be an external server (e.g. using an external database).

Here is an example of a simple server definition:

rails:
  servers:
  - server:
      unique_name: app

These are the parameters that the server section can take:

Important

A more complex server example

rails:
  servers:
  - server:
      unique_name: app
      vendor: aws
      key_name: Default
      region: us-east-1
      size: m3.medium
      root_disk_size: 100
      root_disk_type: ssd
      subnet_id: subnet-40000000
      availability_zone: us-east-1c

Deploy to your own server

You can deploy to one of your Registered servers via manifest settings. The server must be registered before deployment.

Example of registered server

redis:
  servers:
  - server:
      unique_name: redis-main
      address: 123.123.123.123

Shared Servers

You can share a server between two components. This allows you to, for example, use a server for both your front-end app and the database server backing it.

Example of shared server

rails:
  servers:
  - server:
      unique_name: my_rails_app
      vendor: aws
      key_name: Default
      region: us-east-1
      size: m3.medium
mongodb:
  servers:
  - server:
      same_as: my_rails_app

External Servers

If you would like to use an external server for a component (like using your own MySQL or AWS RDS instance, for example), you can define that server as external.

External server definitions specify that the component is hosted on a server external to Cloud 66. This is not a valid target for your main application (e.g. rails) but may be appropriate for another component (e.g. MongoDB):

mysql:
  servers:
  - server: external

Custom LiveLog files

You can add custom live log files for each component. For example:

rails:
  configuration:
    custom_log_files: ["/tmp/mylog/*/*.log"]

…or

docker:
  configuration:
    custom_log_files: ["/tmp/dockerlog/*/*.log"]
postgresql:
  configuration:
    custom_log_files: ["/tmp/your-other-log.log"]

For more information about LiveLogs and additional examples, please see the LiveLogs help page.

Environment variables

You can add environment variables to your manifest files, either globally or per environment. For example:

environment_variables:
  SOME_VARIABLE: value
  ANOTHER_ONE: another_value
  THIRD_ONE: AUTO_GENERATE
  LONG_ONE: AUTO_GENERATE_15

If you need to auto generate a value, you can use the AUTO_GENERATE keyword. It generates a 10 character long random string unless you specify the length after it: AUTO_GENERATE_15 which generates a 15 character random string.

Environment variables set in your manifest file will only apply during the initial build of your application. Please refer to our documentation on environment variables if you’d like to set them beyond this point.

Any environment variable that is generated by the result of the code analysis (like database addresses) will override any value specified in the manifest file. In other words, you cannot specify a value like MYSQL_ADDRESS in your manifest file as it will be ignored.

Detailed examples of Manifest files

In this example we’re defining a MySQL server with all the bells and whistles:

mysql:
  groups:
    default:
      configuration:
        version: 8.0
        root_disk_size: 1000
        root_disk_type: ssd
        engine: percona
        iam_instance_profile_name: mysql-perms
        custom_log_files: [ "/tmp/mysqllog/*/*.log" ]
      servers:
      - server:
          unique_name: mysql-main
          vendor: aws
          key_name: Default
          region: us-east-1
          size: m3.medium
          subnet_id: subnet-40000000
          availability_zone: us-east-1c

In this example we’re defining a Rails app and an accompanying MySQL instance, and applying these settings only in the production environment:

production:
  rails:
    configuration:
      ruby_version: 2.7.2
      asset_pipeline_precompile: true
      do_initial_db_schema_load: false
      reserved_server_memory: 0 #default value
      passenger_process_memory: 200 #default value
      memory_allocator: jemalloc # malloc is default
    servers:
    - server:
      unique_name: rails-main
      vendor: aws
      key_name: Default
      region: us-east-1
      size: m3.medium
      subnet_id: subnet-40000000
      availability_zone: us-east-1c
  mysql:
    groups:
      default: 
        configuration:
          version: 8.0

Notice that we haven’t specified a server for the MySQL instance in this YAML. In cases like this the Cloud 66 Dashboard will prompt you to specify a server during the deployment process, and that server will be installed with MySQL V8.0.

Processes

Background processes can be deployed and managed by Cloud 66. Any process in a Procfile will be picked up, deployed and monitored by the system.

If you would like more flexibility over the signals used to control the processes, you can use the procfile_metadata section. Here is an example:

procfile_metadata:
  worker:
    stop_sequence: ttin, 120, term, 5, kill
  web:
    restart_signal: usr1
    stop_sequence: usr1, 30, kill
  nsq:
    restart_on_deploy: false

In this example, a process called worker is stopped using a TTIN signal first. After waiting for 120 seconds, if the process is still running, a TERM signal will be sent. If it is still running after 5 seconds, it will be killed.

As for web or custom_web processes, you can specify a restart_signal which will be sent to the process serving web. This is useful for web servers that can do “phased” or zero-downtime restarts.

All processes restart during each redeployment of the application. If you want to avoid this, you can set restart_on_deploy to false.

The default values for process signals depend on which web server and/or process manager you are using.

For the default signals for web servers, click the links below:

• Puma
• Unicorn
• Thin

For non-web processes:

• systemd (our default process manager)
• Bluepill (legacy pre-June 2020)

Need help with deployments?

Contact Support

More Questions?

Join the Slack Channel