Skip to content

ringods/pulumi-resource

Repository files navigation

Pulumi Resource Type for Concourse

A Concourse resource type that allows jobs to modify IaaS resources via Pulumi. This resource will work against the Pulumi hosted platform. Cloud based state storage backends are not supported by this resource.

NOTE: This resource is currently under development and might still contain some bugs. Use at least v0.4.0, in combination with a Pulumi 3.x based project. All previous development versions are unsupported.

Community

For usage questions, please post a new message on the Github Discussions board for this project.

If you are not sure whether your problem is a usage problem or a bug, please reach out via the discussions first. Also reach out first via the Discussions if you have a suggestion for an improvement.

Only file a new Github issue when you are really sure there is a bug.

Using the resource type in your pipeline

To use this resource type in your pipeline, you have to register it under the resource_types section in your pipeline:

Example

resource_types:
- name: pulumi
  type: registry-image
  source:
    repository: ghcr.io/ringods/pulumi-resource
    tag: v0.4.0

Concourse will now know about the resource type called pulumi in your list of resources.

Source Configuration

Every stack you want to manage in your pipeline needs to be registered as a resource in your pipeline. Each stack is identified by the following 3 properties, extended with authentication credentials:

  • organization: Required. The name of the organization you use on the Pulumi platform.
  • project: Required. The name of your Pulumi project.
  • stack: Required. Name of the stack to manage, e.g. staging.
  • token: Required. Access token which will be used to login on the Pulumi platform. Use Concourse Credential Management to keep your token safe!

Example

resources:
- name: my-staging-network
  type: pulumi
  source:
    organization: companyname
    project: network
    stack: staging
    token: pul-XXXXXXXXXXXXXXXXX

Behavior

check: Check for new infrastructure deployments

This uses a non-public REST endpoint on the hosted platform, authenticated with the provided access token, to check for new revisions of the configured stack. The check step filters out failed deployments and will only provide new versions for successful deploys.

in: Get the details of a new infrastructure revision

You can use a Pulumi resource in a get step to act as a trigger for downstream builds. This triggering will happen once the check step finds new succesful deploys of the configured stack.

The details of this new infrastructure revision are not yet provided for downstream processing.

Example

- name: after-update-of-my-staging-network
  plan:
  - get: my-staging-network
    trigger: true
  - task: do-something-after-deploying-staging-network-stack
    ...

out: run pulumi to deploy the latest infrastructure code

Use a pulumi resource in a put step if you want to run pulumi in your infrastructure code via Concourse.

Pulumi can use different language runtimes. The amount of possibilities in language runtime, language runtime version as well as additional tools on the side makes it sheer impossible to provide that all within the container image for this resource type. Therefor, this resource type is implemented in a way that you can pass your custom runtime image via config.

  • runtime: should point to a registry-image resource containing your specific language runtime, version and additional tooling.
  • sources: should point to an input retrieved via get or to an output from a previous step.
  • config: this section may contain any valid configuration which you would normally put in your Pulumi.<stack>.yaml file. Secrets are not yet supported.
  • env: this section may contain environment variables which are needed. E.g., for authentication

Let's show this at work with an example of a NodeJS based Pulumi stack:

Example

Pipeline file:

resources:
- name: nodejs14
  type: registry-image
  source:
    repository: node

- name: myinfracode
  type: git
  uri: git@github.com:owner/myinfracode.git
    branch: main

- name: my-staging-network
  type: pulumi
  source:
    organization: companyname
    project: network
    stack: staging
    token: pul-XXXXXXXXXXXXXXXXX

jobs:
- name: update-infra
  plan:
  - get: myinfracode
    trigger: true
  - get: nodejs14
  - task: npm-install
    image: nodejs14
    input_mapping: { code: myinfracode }
    file: code/npm-install.yml
  - put: my-staging-network
    params:
      runtime: nodejs14
      sources: code
      config:
        network:setting1: value1
        network:setting2: value2
      env:
        CLOUD_PROVIDER_CREDENTIALS: |
          {
            "private_key_id": "<private key id string>",
            "private_key": "<private key string here>",
            ...
          }
        CLOUD_PROVIDER_REGION: value1
        CLOUD_PROVIDER_ZONE: value2

The example task file:

---
platform: linux

inputs:
- name: code

outputs:
- name: code

run:
  path: /bin/bash
  args:
    - -c
    - |
      set -euo pipefail
      cd ./code
      npm ci

The resources section contains 3 resources:

  • nodejs14: a container image for NodeJS 14 and support tools like npm or yarn. You can configures this fully to your liking.
  • myinfracode: a git resource pointing to your Pulumi code in a git repository, e.g. tracking new commits on branch main
  • my-staging-network: the pulumi resource pointing to our staging network stack.

We then create a job which does 4 steps:

  • retrieves the new revision of the code in the get: myinfracode step.
  • fetches your NodeJS runtime image in the get: nodejs14 step. Do not forget to get your runtime image in the job. If you forget this step, you will not have it available in your put step.
  • runs npm install first on the retrieved code using the nodejs14 image as the container
  • pass your modified sources as an output. If you don't do this and pass your sources using your get resource, you still have your clean sources.
  • runs Pulumi (via the Automation API) on your code, using the provided runtime image. The stack config set via the Concourse pipeline is mixed with any config already provided in the Pulumi.<stack>.yaml file residing in the source repository

Full Example

To wire all the pieces together, here is a full example combining all the previous snippets together:

resource_types:
- name: pulumi
  type: registry-image
  source:
    repository: ghcr.io/ringods/pulumi-resource
    tag: v0.2.0

resources:
- name: nodejs14
  type: registry-image
  source:
    repository: node

- name: myinfracode
  type: git
  uri: git@github.com:owner/myinfracode.git
    branch: master

- name: my-staging-network
  type: pulumi
  source:
    organization: companyname
    project: network
    stack: staging
    token: pul-XXXXXXXXXXXXXXXXX

jobs:
- name: update-infra
  plan:
  - get: myinfracode
    trigger: true
  - get: nodejs14
  - task:
    image: nodejs14
    input_mapping: { code: myinfracode }
    file: code/npm-install.yml
  - put: my-staging-network
    params:
      runtime: nodejs14
      sources: code
      config:
        network:setting1: value1
        network:setting2: value2
      env:
        CLOUD_PROVIDER_CREDENTIALS: |
          {
            "private_key_id": "<private key id string>",
            "private_key": "<private key string here>",
            ...
          }
        CLOUD_PROVIDER_REGION: value1
        CLOUD_PROVIDER_ZONE: value2

- name: after-update-infra
  plan:
  - get: my-staging-network
    trigger: true
  - task: do-something-after-rolling-out-network-stack
    ...