Stepfunctions-local provides a local AWS Step Functions server. This package only aims at replacing AWS Step Functions in a local context. Its API is totally compliant with AWS service, thus you can use it for your tests.
In February 2019, AWS released an offical local version of Step Functions, available as a docker image. We will still accept pull requests, but we encourage you to use the official local version.
More information: Setting Up Step Functions Local (Downloadable Version)
- Ease development and tests. You don't have to upload all your resources on AWS to run a state machine.
- 100% compliant with AWS API. You can query it using the AWS cli by changing the endpoint. Errors and responses follow the same format.
- Works well with localstack.
You only need to configure your activity worker to use this stepfunctions
instance. In javascript:
AWS.config.stepfunctions = {
region: 'local',
endpoint: 'http://localhost:4584',
}
Then, start stepfunctions-local
server and you will be able to execute requests to StepFunctions API (GetActivityTask
, SendTaskSuccess
, ...).
Simply configure your lambda endpoint and region when starting the server:
$> stepfunctions-local start --lambda-endpoint http://hostname.com:1337 --lambda-region my-region
stepfunctions-local
will directly query lambda using this configuration.
stepfunctions-local
does not aim to emulate Lambda. To do this you need a local Lambda server that is compliant to AWS API. We recommand to use localstack for that. See how to here.
Simply configure your ECS endpoint and region when starting the server:
$> stepfunctions-local start --ecs-endpoint http://hostname.com:1337 --ecs-region my-region
stepfunctions-local
will directly query ECS using this configuration.
stepfunctions-local
does not aim to emulate ECS. To do this you need a local ECS server that is compliant to AWS API. You may have to create a mock server to do this yourself.
# Use it using command lines
$> npm install -g stepfunctions-local
# Use it in your code
$> cd /your/project/using/stepfunctions
$> npm install --save stepfunctions-local
$> docker build -t stepfunctions-local .
You will find some help on the Wiki page.
$> stepfunctions-local start
Options:
-V, --version output the version number
--port <port> the port the server should run on
--region <region> the region the server should run on
--lambda-region <lambda-region> the region for lambda
--lambda-endpoint <lambda-endpoint> the endpoint for lambda
--ecs-region <ecs-region> the region for ECS
--ecs-endpoint <ecs-endpoint> the endpoint for ECS
-h, --help output usage information
# Options are same as above
$> docker run -it --rm -p 4584:4584 stepfunctions-local start <options>
const stepfunctionsLocal = require('stepfunctions-local');
stepfunctionsLocal.start({
port: 4584,
region: 'local',
lambdaRegion: 'local',
lambdaEndpoint: 'http://localhost:4574',
ecsRegion: 'local',
ecsEndpoint: 'http://localhost:4600',
});
- port: 4584
- region: local
- lambda-region: local
- lambda-endpoint: http://localhost:4574
- ecs-region: local
- ecs-endpoint: http://localhost:4600
The service does not log anything by default. It uses the debug package which is based on the DEBUG
environment variable. You can log process info by setting it.
Example:
$> DEBUG=stepfunctions-local:* stepfunctions-local start
# List state machines
$> aws stepfunctions --endpoint http://localhost:4584 list-state-machines
# Create a new state machine
$> aws stepfunctions --endpoint http://localhost:4584 create-state-machine --name my-state-machine --definition '{"Comment":"A Hello World example of the Amazon States Language using a Pass state","StartAt":"HelloWorld","States":{"HelloWorld":{"Type":"Pass","End":true}}}' --role-arn arn:aws:iam::0123456789:role/service-role/MyRole
# Describe state machine
$> aws stepfunctions --endpoint http://localhost:4584 describe-state-machine --state-machine-arn arn:aws:states:local:0123456789:stateMachine:my-state-machine
# Start state machine execution
$> aws stepfunctions --endpoint http://localhost:4584 start-execution --state-machine-arn arn:aws:states:local:0123456789:stateMachine:my-state-machine --name my-execution --input '{"comment":"I am a great input !"}'
# List state machine executions
$> aws stepfunctions --endpoint http://localhost:4584 list-executions --state-machine-arn arn:aws:states:local:0123456789:stateMachine:my-state-machine
# Describe execution
$> aws stepfunctions --endpoint http://localhost:4584 describe-execution --execution-arn arn:aws:states:local:0123456789:execution:my-state-machine:my-execution
# Describe state machine related to execution
$> aws stepfunctions --endpoint http://localhost:4584 describe-state-machine-for-execution --execution-arn arn:aws:states:local:0123456789:execution:my-state-machine:my-execution
# Get execution history
$> aws stepfunctions --endpoint http://localhost:4584 get-execution-history --execution-arn arn:aws:states:local:0123456789:execution:my-state-machine:my-execution
Start a local Lambda server using localstack
(you need to clone the repository first):
$> docker-compose up
Note: you may have to run TMPDIR=/private$TMPDIR docker-compose up
if you are on Mac OS.
If you need to access AWS services from within your Lambda, the variable LOCALSTACK_HOSTNAME
will contain the name of the host where Localstack services are available.
For instance, in a NodeJS Lambda function, you can use the following to access S3 functions:
const s3 = new AWS.S3({
endpoint: 'http://' + process.env.LOCALSTACK_HOSTNAME + ':4572',
});
s3.listBuckets({}, function(err, data) {
// your callback
});
Configure your Lambda endpoint and region when starting the server:
$> stepfunctions-local start --lambda-endpoint http://localhost:4574 --lambda-region local
stepfunctions-local
will directly query lambda using this configuration.
Actions | Support |
---|---|
CreateActivity | Following errors are not thrown: ActivityLimitExceeded |
CreateStateMachine | Following errors are not thrown: StateMachineDeleting, StateMachineLimitExceeded |
DeleteActivity | * |
DeleteStateMachine | * |
DescribeActivity | * |
DescribeStateMachine | * |
DescribeStateMachineForExecution | * |
GetActivityTask | Following errors are not thrown: ActivityWorkerLimitExceeded |
GetExecutionHistory | * |
ListActivities | * |
ListExecutions | * |
ListStateMachines | * |
SendTaskFailure | * |
SendTaskHeartbeat | * |
SendTaskSuccess | * |
StartExecution | Following errors are not thrown: ExecutionLimitExceeded |
StopExecution | * |
UpdateStateMachine | Following errors are not thrown: StateMachineDeleting |
AWS added support for executing a variety of AWS services from Step Functions. For now, only Lambda and ECS are supported. Adding new integrations should be quite straightforward (see #44), feel free to submit pull requests.
Service | Support |
---|---|
AWS Lambda | * |
AWS Batch | Not yet |
Amazon DynamoDB | Not yet |
Amazon ECS/Fargate | * |
Amazon SNS | Not yet |
Amazon SQS | Not yet |
AWS Glue | Not yet |
Amazon SageMaker | Not yet |
States | Support |
---|---|
Pass | * |
Task | * |
Choice | * |
Wait | * |
Succeed | * |
Fail | * |
Parallel | * |
Wow, that's great ! Feedback, bug reports and pull requests are more than welcome !
To run the tests, you must first authenticate to AWS,
including setting a default region. You can do this via the aws configure
command or by setting environment variables:
$> export AWS_ACCESS_KEY_ID=(your access key)
$> export AWS_SECRET_ACCESS_KEY=(your secret key)
$> export AWS_DEFAULT_REGION=us-east-1
You can then run the tests as follows:
$> npm run lint
$> npm run test
- Add execution abortion related history events
- Continue services integration
See LICENSE.