Skip to content

Custom Code

This microservice is capable of running custom user functions (NodeJS or Python).

Installation prerequisites

Here is a few thing that should be considered:

  • The service must have access to a npm registry. You can use the NPM_CONFIG_REGISTRY environment variable to specify your own.
  • The service needs a volume, we recommend using volumes with a high I/O speed as it might be highly solicited when installing dependencies for example.

Environment variables

Name Description Default value
PYTHON_FUNCTIONS_RUN_TIMEOUT Python Functions execution timeout (ms) 20000
PYTHON_API_URL Python API url http://localhost:8000
PORT HTTP port 4000
FUNCTIONS_STORAGE_FILESYSTEM_DIRPATH Functions directory path data/functions/
FUNCTIONS_RUN_TIMEOUT Functions execution timeout (ms) 20000
FUNCTIONS_WORKERS_MAX_LRU Maximum number of function workers kept in memory 500
NODE_BUILTIN_MODULES Allowed builtin modules http, https, url, util, zlib, dns, stream, buffer, crypto
NODE_WORKER_MAX_OLD_GENERATION_SIZE_MB NodeJS function worker maxOldGenerationSizeMb 100MB
INACTIVE_NODE_WORKER_DELETION_TIMEOUT Inactive period in seconds after which node workers are automatically terminated 3600
UPDATE_SCRIPTS_ON_STARTUP If set to yes or true, makes sure function scripts are in-sync with corresponding workspace.yaml files on start-up. Useful to migrate functions after an upgrade with breaking changes. Please care that in case of a multi-instances deployment, every instance will run this migration. no
REQUEST_MAX_SIZE Maximum request body size (format from bodyParser.json) 1mb

Microservice testing

  1. Create a function "workspace" (named test and containing a hello function):
curl --location 'http://localhost:4000/v2/functions/test' \
--header 'Content-Type: application/json' \
--data '{
    "functions": {
        "hello": {
            "code": "return \"Hello \" + name;",
            "parameters": {
                "name": {
                    "type": "string"

If successful, the same body should be returned in response.

  1. Run the function:
curl --location 'http://localhost:4000/v2/functions/test/run/hello' \
--header 'Content-Type: application/json' \
--data '{
    "parameters": {
        "name": "world"

If successful, the answer should be similar to this:

    "result": "Hello world",
    "logs": [],
    "duration": 38

Congratulations, you service is up and running!


Async functions

Even if they do not use any async call, every functions are always annotated as "async".
As such, calling a function from the same workspace must be done with await keyword. See Shared execution context for a full example.

Shared execution context

Functions from the same workspace are executed within a generated JS file gathering these functions together. This allows functions to call each others :

funcA :

  return "world";

funcB :

  return "hello " + (await funcA());

Shared memory

Functions from the same workspace are executed within the same worker NodeJS and VM2 instance. Although their inner variables are automatically destroyed after each execution, a special cache variable is provided to share memory between functions and accross executions :

funcA :

let counter = cache['counter']

counter = (counter || 0) + 1

cache['counter'] = counter;

return counter;

funcB :

  let counter = cache['counter'];
  return "Counter : " + counter;