Serverless Framework: Deploy on Scaleway Functions

The Scaleway functions plugin for Serverless Framework allows users to deploy their functions and containers to Scaleway Functions with a simple serverless deploy.

Serverless Framework handles everything from creating namespaces to function/code deployment by calling APIs endpoint under the hood.

• Serverless Framework: Deploy on Scaleway Functions
  • Requirements
  • Create a Project
  • Configure your functions
  • Functions Handler
    • Node
    • Python
    • Golang
    • Events
    • Managing containers
  • Logs
  • Documentation and useful Links
  • Contributing
  • License

Requirements

• Install node.js
• Install Serverless CLI (npm install serverless -g)

Let's work into ~/my-srvless-projects

# mkdir ~/my-srvless-projects
# cd ~/my-srvless-projects

Create a Project

The easiest way to create a project is to use one of our templates. The list of templates is here

Let's use python3

serverless create --template-url https://github.com/scaleway/serverless-scaleway-functions/tree/master/examples/python3 --path myService

Once it's done, we can install mandatory node packages used by serverless

cd mypython3functions
npm i

Note: these packages are only used by serverless, they are not shipped with your functions.

Configure your functions

Your functions are defined in the serverless.yml file created:

service: scaleway-python3
configValidationMode: off

provider:
  name: scaleway
  runtime: python310
  # Global Environment variables - used in every functions
  env:
    test: test
  scwToken: <scw-token>
  scwProject: <scw-project-id>
  # region in which the deployment will happen, (default fr-par)
  scwRegion: <scw-region>

plugins:
  - serverless-scaleway-functions
  
package:
  patterns:
    - '!node_modules/**'
    - '!.gitignore'
    - '!.git/**'

functions:
  first:
    handler: handler.py
    # Local environment variables - used only in given function
    env:
      local: local

Note: provider.name and plugins MUST NOT be changed, they enable us to use the scaleway provider

This file contains the configuration of one namespace containing one or more functions (in this example, only one) of the same runtime (here python3)

The different parameters are:

• service: your namespace name
• provider.runtime: the runtime of your functions (check the supported runtimes above)
• provider.env: environment variables attached to your namespace are injected to all your namespace functions
• provider.secret: secret environment variables attached to your namespace are injected to all your namespace functions, see this example project
• scwToken: Scaleway token you got in prerequisites
• scwProject: Scaleway org id you got in prerequisites
• scwRegion: Scaleway region in which the deployment will take place (default: fr-par)
• package.patterns: usually, you don't need to configure it. Enable to include/exclude directories to/from the deployment
• functions: Configure of your fonctions. It's a yml dictionary, with the key being the function name
  • handler (Required): file or function which will be executed. See the next section for runtime specific handlers
  • env (Optional): environment variables specific for the current function
  • secret (Optional): secret environment variables specific for the current function, see this example project
  • minScale (Optional): how many function instances we keep running (default: 0)
  • maxScale (Optional): maximum number of instances this function can scale to (default: 20)
  • memoryLimit: ram allocated to the function instances. See the introduction for the list of supported values
  • timeout: is the maximum duration in seconds that the request will wait to be served before it times out (default: 300 seconds)
  • runtime: (Optional) runtime of the function, if you need to deploy multiple functions with different runtimes in your Serverless Project. If absent, provider.runtime will be used to deploy the function, see this example project.
  • events (Optional): List of events to trigger your functions (e.g, trigger a function based on a schedule with CRONJobs). See events section below

Functions Handler

Based on the chosen runtime, the handler variable on function might vary.

Node

Path to your handler file (from serverless.yml), omit ./, ../, and add the exported function to use as a handler:

- src
  - handlers
    - firstHandler.js  => module.exports.myFirstHandler = ...
    - secondHandler.js => module.exports.mySecondHandler = ...
- serverless.yml

In serverless.yml:

provider:
  # ...
  runtime: node16 # or node10, node14, node17
functions:
  first:
    handler: src/handlers/firstHandler.myFirstHandler
  second:
    handler: src/handlers/secondHandler.mySecondHandler

Python

Similar to node, path to handler file src/testing/handler.py:

- src
  - handlers
    - firstHandler.py  => def my_first_handler
    - secondHandler.py => def my_second_handler
- serverless.yml

In serverless.yml:

provider:
  # ...
  runtime: python310 # or python37, python38, python39
functions:
  first:
    handler: src/handlers/firstHandler.my_first_handler
  second:
    handler: src/handlers/secondHandler.my_second_handler

Golang

Path to your handler's package, for example if I have the following structure:

- src
  - testing
    - handler.go -> package main in src/testing subdirectory
  - second
    - handler.go -> package main in src/second subdirectory
- serverless.yml
- handler.go -> package main at the root of project

Your serverless.yml functions should look something like this:

provider:
  # ...
  runtime: go118
functions:
  main:
    handler: "."
  testing:
    handler: src/testing
  second:
    handler: src/second

Events

With events, you may link your functions with specific triggers, which might include CRON Schedule (Time based), MQTT Queues (Publish on a topic will trigger the function), S3 Object update (Upload an object will trigger the function).

Note that we do not include HTTP triggers in our event types, as a HTTP endpoint is created for every function. Triggers are just a new way to trigger your Function, but you will always be able to execute your code via HTTP.

Here is a list of supported triggers on Scaleway Serverless, and the configuration parameters required to deploy them:

• schedule: Trigger your function based on CRON schedules
  • rate: CRON Schedule (UNIX Format) on which your function will be executed
  • input: key-value mapping to define arguments that will be passed into your function's event object during execution.

To link a Trigger to your function, you may define a key events in your function:

functions:
  handler: myHandler.handle
  events:
    # "events" is a list of triggers, the first key being the type of trigger.
    - schedule:
        # CRON Job Schedule (UNIX Format)
        rate: '1 * * * *'
        # Input variable are passed in your function's event during execution
        input:
          key: value
          key2: value2

You may link Events to your Containers too (See section Managing containers below for more informations on how to deploy containers):

custom:
  containers:
    mycontainer:
      directory: my-directory
      # Events key
      events:
        - schedule:
            rate: '1 * * * *'
            input:
              key: value
              key2: value2

You may refer to the follow examples:

• NodeJS-10 with schedule trigger
• Container with Schedule Trigger

Managing containers

Requirements: You need to have Docker installed to be able to build and push your image to your Scaleway registry.

You must define your containers inside the custom.containers field in your serverless.yml manifest. Each container must specify the relative path of its application directory (containing the Dockerfile, and all files related to the application to deploy):

custom:
  containers:
    mycontainer:
      directory: my-container-directory
      # port: 8080
      # Environment only available in this container 
      env:
        MY_VARIABLE: "my-value"

Here is an example of the files you should have, the directory containing your Dockerfile and scripts is my-container-directory.

.
├── my-container-directory
│   ├── Dockerfile
│   ├── requirements.txt
│   ├── server.py
│   └── (...)
├── node_modules
│   ├── serverless-scaleway-functions
│   └── (...)
├── package-lock.json
├── package.json
└── serverless.yml

Scaleway's platform will automatically inject a PORT environment variable on which your server should be listening for incoming traffic. By default, this PORT is 8080. You may change the port in your serverless.yml.

You may use the container example to getting started.

Logs

The serverless logs command lets you watch the logs of a specific function or container.

Pass the function or container name you want to fetch the logs for with --function:

serverless logs --function <function_or_container_name>

Documentation and useful Links

• Official Scaleway Serverless Functions Documentation
• Official Scaleway Serverless Containers Documentation
• Scaleway Functions Golang Runtime (you MUST use this library if you plan to develop with Golang).
• Serverless Framework documentation
• Scaleway Cloud Provider
• Examples/Templates for different runtimes on Scaleway Functions

Contributing

This plugin is mainly developed and maintained by Scaleway Serverless Team but you are free to open issues or discuss with us on our Community Slack Channels #serverless-containers and #serverless-functions.

License

This project is MIT licensed.