Use Swift signal resources with Cloud Orchestration user data scripts

  • Last updated on: 2018-05-01
  • Authored by: Mike Asthalter

Introduction

Cloud Orchestration users can implement Swift signal resources that help determine the status of user data scripts. For example, implementing the following Swift signal resources adds functionality that indicates whether a user data script completed, and if so, whether it has succeeded or failed.

Use the OS::Heat::SwiftSignal resource to coordinate resource creation with signals coming from sources that are external or internal to the stack. This resource is often used in conjunction with the OS::Heat::SwiftSignalHandle resource.

Use the SwiftSignalHandle resource to create a temporary URL that applications and scripts use to send signals. The SwiftSignal resource waits for this URL to receive a specified number of SUCCESS signals within a specified amount of time. The temporary URL is created by using Rackspace Cloud Files.

The following tutorial walks you through the process of setting up a single-node Linux server that signals the success or failure of user_data script execution at a given URL.

Add the top-level template sections

At the top of the template, add the following information:

heat_template_version: 2014-10-16

description: |
  A single-node linux server with swift signaling.

resources:

outputs:

Add a SwiftSignalHandle resource

In the resources section, add a SwiftSignalHandle resource. The following code example shows how to add this resource:

signal_handle:
  type: "OS::Heat::SwiftSignalHandle"

Add a SwiftSignal resource

In the resources section, add a SwiftSignal resource.

Provide the URL in the handle property and the number of signals in the count property.

The following example shows how to add a SwiftSignal resource that waits for 600 seconds to receive one signal on the handle.

wait_on_server:
  type: OS::Heat::SwiftSignal
  properties:
    handle: {get_resource: signal_handle}
    count: 1
    timeout: 600

The stack is marked as a failure if the specified number of signals is not received within the amount of time specified in the timeout property, or if a signal other than SUCCESS is received. A data string and a reason string might be attached to the success or failure notification. The data string is an attribute that can be displayed as template output.

Add a server resource

Add a Linux server. In the user_data property, include a bash script. At the end of the script execution, send a SUCCESS or FAILURE message to the temporary URL that is created by the SwiftSignalHandle resource that you added earlier.

linux_server:
  type: OS::Nova::Server
  properties:
    image: 4b14a92e-84c8-4770-9245-91ecb8501cc2
    flavor: 1 GB Performance
    user_data:
      str_replace:
        template: |
          #!/bin/bash -x
          # assume you are doing a long-running operation here
          sleep 300

          # Assuming the long-running operation completed successfully,
          # notify success signal
          wc_notify --data-binary '{"status": "SUCCESS", "data": "Script execution succeeded"}'

          # Alternatively, if operation fails, a FAILURE with reason and data may be sent,
          # notify failure signal example below
          # wc_notify --data-binary '{"status": "FAILURE", "reason":"Operation failed due to xyz error", "data":"Script execution failed"}'

        params:
          # Replace all occurrences of "wc_notify" in the script with an
          # appropriate curl PUT request using the "curl_cli" attribute
          # of the SwiftSignalHandle resource
          wc_notify: { get_attr: ['signal_handle', 'curl_cli']

Add the Swift signal URL to the outputs section

The following example shows how to add the Swift signal URL to the outputs section:

# Get the signal URL which contains all information passed to the signal handle
signal_url:
  value: { get_attr: ['wait_handle', 'curl_cli'] }
  description: Swift signal URL

# Obtain data describing script results. If nothing is passed, this value will be NULL
signal_data:
  value: { get_attr: ['wait_on_server', 'data'] }
  description: Data describing script results

# Obtain IPv4 address of server
server_public_ip:
  value:{ get_attr: [ linux_server, accessIPv4 ] }
  description: Linux server public IP

View the full template

The following code shows what the complete template looks like:

heat_template_version: 2014-10-16

description: |
  Single node linux server with swift signaling.

resources:

  signal_handle:
    type: "OS::Heat::SwiftSignalHandle"

  wait_on_server:
    type: OS::Heat::SwiftSignal
    properties:
      handle: {get_resource: signal_handle}
      count: 1
      timeout: 600

  linux_server:
    type: OS::Nova::Server
    properties:
      image: 4b14a92e-84c8-4770-9245-91ecb8501cc2
      flavor: 1 GB Performance
      user_data:
        str_replace:
          template: |
            #!/bin/bash -x
            # assume you are doing a long-running operation here
            sleep 300

            # Assuming long-running operation completed successfully, notify success signal
            wc_notify --data-binary '{"status": "SUCCESS", "data": "Script execution succeeded"}'

            # Alternatively, if operation fails, a FAILURE with reason and data may be sent,
            # notify failure signal example below
            # wc_notify --data-binary '{"status": "FAILURE", "reason":"Operation failed due to xyz error", "data":"Script execution failed"}'

          params:
            wc_notify: { get_attr: ['signal_handle', 'curl_cli'] }

outputs:
  # Get the signal URL which contains all information passed to the signal handle
  signal_url:
    value: { get_attr: ['signal_handle', 'curl_cli'] }
    description: Swift signal URL

  # Obtain data describing script results. If nothing is passed, this value will be NULL
  signal_data:
    value: { get_attr: ['wait_on_server', 'data'] }
    description: Data describing script results

  # Obtain IPv4 address of server
  server_public_ip:
    value: { get_attr: [ linux_server, accessIPv4 ] }
    description: Linux server public IP

Continue the conversation in the Rackspace Community.