Run ready to use actions from StackStorm Exchange as AWS Lambda with serverless framework. Serverless and Stackstormless.

• Serverless framework
• NodeJS >= 8.4.0
• Docker - used to build and local-run Lambda on any OS
• Build tools (build-essentials package on Ubuntu)
• SSL dev files (libssl-dev package on Ubuntu)

Install serverless dependency globally

npm install -g serverless

Init with package.json:

npm init

Install the plugin:

npm i --save-dev serverless-plugin-stackstorm

Browse StackStorm Exchange to find the integration pack and an action you'd like to use. In the example below we use github.get_issue from GitHub integration pack.

Configure your service to use the plugin by creating serverless.yml file.

service: my-service

provider:
  name: aws
  runtime: python2.7 # StackStorm runners are based on Python 2

functions:
  get_issue:
    stackstorm: # `stackstorm` object replaces `handler`. The rest is the same.
      action: github.get_issue
      config:
        token: ${env:GITHUB_TOKEN}
      input:
        user: "{{ input.pathParameters.user }}"
        repo: "{{ input.pathParameters.repo }}"
        issue_id: "{{ input.pathParameters.issue_id }}"
      output:
        statusCode: 200
        body: "{{ output }}"
    events:
      - http:
          method: GET
          path: issues/{user}/{repo}/{issue_id}

plugins:
  - serverless-plugin-stackstorm

There are few new options inside the function definition (see serverless.example.yml for more options):

• stackstorm.action allows you to pick up a function you want to turn into a lambda
• stackstorm.config sets config parameters for the action. Config parameters are pack-wide in stackstorm and are commonly used for authentication tokens and such.
• stackstorm.input defines how input event parameters should be transformed to match the parameters list stackstorm action expects
• stackstorm.output defines the transformation that should be applied to the action output to form a result of lambda execution

If you are in doubt on the list of parameters given StackStorm action expects, check action info:

$ sls stackstorm info --action github.get_issue
github.get_issue .............. Retrieve information about a particular Github issue.
Parameters
  issue_id [string] (required)  Issue id
  repo [string] (required) .... Repository name.
  user [string] (required) .... User / organization name.
Config
  base_url [string] (required)  The GitHub URL, for GitHub Enterprise please set enterprise_url.
  deployment_environment [string] (required)  The environment for this StackStorm server.
  enterprise_url [string]  .... GitHub API url (including /api/v3) of your GitHub Enterprise hostname.
  github_type [string] (required)  Default to either github or enterprise.
  password [string]  .......... GitHub Password
  repository_sensor [object]  . Sensor specific settings.
  token [string] (required) ... GitHub oAuth Token
  user [string]  .............. GitHub Username

Then deploy your function to the cloud and invoke it:

sls deploy

sls invoke --function get_issue --log \
--data '{"pathParameters": {"user": "StackStorm", "repo": "st2", "issue_id": "3785"}}'

You can also invoke a function locally for testing. It runs in docker container to ensure compatibility with AWS lambda environment.

sls stackstorm docker run -f get_issue --verbose --passthrough -d '{"pathParameters": {"user": "StackStorm", "repo": "st2", "issue_id": "3785"}}'

Note the options:

• --passthrough: skips actual invocation - comes handy to ensure the input maps to action parameters right, without invoking the body of the lambda.
• --verbose: shows the transformation routine that happened for a particular input and output.

Here is an example of a verbose output:

Incoming event ->
{
  "issue_id": "222"
}
-> Parameter transformer ->
{
  "repo": "st2",
  "issue_id": "222",
  "user": "StackStorm"
}
-> Action call ->
{
  "result": {
    "url": "https://github.com/StackStorm/st2/pull/222",
    "created_at": "2014-07-14T19:25:46.000000+00:00",
    ...
  },
  "exit_code": 0,
  "stderr": "",
  "stdout": ""
}
-> Output transformer ->
{
  "result": "2014-07-14T19:25:46.000000+00:00"
}

The plugin also provides a few optional commands. You don't have to use them as they are all included into sls package, but they still might be handy in some situations.

• sls stackstorm - Build λ with StackStorm
• sls stackstorm clean - Clean StackStorm code
• sls stackstorm docker pull - Pull λ docker image
• sls stackstorm docker start - Start λ docker container
• sls stackstorm docker stop - Stop λ docker container
• sls stackstorm docker exec - Execute a command in λ docker container
• sls stackstorm docker run - Execute a function in λ docker container
• sls stackstorm install adapter - Install StackStorm adapter
• sls stackstorm install deps - Install StackStorm dependencies
• sls stackstorm install packs - Install a pack
• sls stackstorm install packDeps - Install dependencies for packs
• sls stackstorm info - Print information on the action

The available packs can be discovered in StackStorm Exchange (https://exchange.stackstorm.com/). At the moment, the collection consist of 6500+ actions spread across 130 packs. We've yet to try them all, though, but the one we did are marked with serverless tag.

The StackStorm packs this plugin allows you to run on serverless infrastructure are all part of StackStorm Exchange. We encourage community members to contribute to this packs to enrich the entire ecosystem. The most simple way to help us is to try different packs, mark the one that works with serverless keyword and report ones that don't work for some reason. For now, the plugin only supports stackstorm's python runner, but they represent more than 90% of exchange actions.