title | linkTitle | description |
---|---|---|
Serverless Framework |
Serverless Framework |
Use the Serverless Framework with LocalStack
|
This guide explains how to integrate LocalStack with the Serverless Framework. Although it probably requires a few code changes, integrating LocalStack with the Serverless Framework is fairly straightforward.
In particular, the setup consists of the following two steps.
- Installing and configuring the Serverless-LocalStack plugin.
- Adjusting AWS endpoints in Lambda functions.
{{< callout "note" >}}
Currently, the Serverless Framework integration with LocalStack via the serverless-localstack
plugin only provides support up to Serverless Framework version 3.39.
Serverless Framework V4 requires a subscription and a license to use.
The serverless-localstack
plugin has not been updated or tested against version 4 of the Serverless Framework, and breaking changes are expected if attempted.
Currently there is no timeline on when an update will be released to enable compatibility with Serverless Framework V4.
{{< /callout >}}
This guide assumes that you have the following tools installed.
It also assumes that you already have a Serverless app set up consisting of a couple of Lambda functions and a serverless.yml
file similar to the following.
An example Serverless app integrated with LocalStack can be found here: Simple REST API using the Serverless Framework and LocalStack
service: my-service
frameworkVersion: ">=1.1.0 <=2.50.0"
provider:
name: aws
runtime: python3.8
environment:
DYNAMODB_TABLE: ${self:service}-${opt:stage, self:provider.stage}
iamRoleStatements:
- Effect: Allow
Action:
- dynamodb:Query
- ...
Resource: "arn:aws:dynamodb:${opt:region, self:provider.region}:*:table/${self:provider.environment.DYNAMODB_TABLE}"
functions:
create:
handler: todos/create.create
events:
- http:
path: todos
method: post
cors: true
...
resources:
Resources:
TodosDynamoDbTable:
Type: 'AWS::DynamoDB::Table'
DeletionPolicy: Retain
Properties:
...
TableName: ${self:provider.environment.DYNAMODB_TABLE}
To install the plugin, execute the following command in the root of your project. {{< command >}} $ npm install -D serverless-localstack {{< / command >}}
Next, set up the plugin by adding the following properties to serverless.yml
.
...
plugins:
- serverless-localstack
custom:
localstack:
stages:
- local
This sets up Serverless to use the LocalStack plugin but only for the stage "local". Next, you need make minor adjustments to your function code in order to make your application work no matter if it is deployed on AWS or LocalStack.
You are likely using an AWS SDK (such as Boto3 for Python) in your Lambda functions to interact with other AWS services such as DynamoDB.
For example, in Python, your code to set up a connection to DynamoDB may look like this:
...
dynamodb = boto3.resource('dynamodb')
...
By default, this call attempts to create a connection via the usual AWS endpoints. However, when running services in LocalStack, we need to make sure, our applications creates a connection via the LocalStack endpoint instead.
Usually, all of LocalStack's services are available via a specific port on localhost (e.g. localhost:4566
).
However, this endpoint only works when accessing LocalStack from outside its Docker runtime.
Since the Lambda functions execute within the LocalStack Docker container, Lambda functions cannot access other services via the usual localhost endpoint.
Instead, LocalStack provides a special environment variable AWS_ENDPOINT_URL
which contains the internal endpoint of the LocalStack services from within its runtime environment.
Hence, you need to configure the Lambda functions to use the AWS_ENDPOINT_URL
endpoint when accessing other AWS services in LocalStack.
In Python, this may look something like.
The code detects if it is running in LocalStack by checking if the AWS_ENDPOINT_URL
variable exists and then configures the endpoint URL accordingly.
...
if 'AWS_ENDPOINT_URL' in os.environ:
dynamodb = boto3.resource('dynamodb', endpoint_url=os.environ['AWS_ENDPOINT_URL'])
else:
dynamodb = boto3.resource('dynamodb')
...
In LocalStack Pro, no code changes are required using our [Transparent Endpoint Injection]({{< ref "user-guide/tools/transparent-endpoint-injection" >}}).
You can now deploy your Serverless service to LocalStack.
First, start LocalStack by running {{< command >}} $ localstack start {{< / command >}}
Then deploy the endpoint by running {{< command >}} $ serverless deploy --stage local {{< / command >}}
The expected result should be similar to:
Serverless: Packaging service...
Serverless: Excluding development dependencies...
Serverless: Creating Stack...
Serverless: Checking Stack create progress...
........
Serverless: Stack create finished...
Serverless: Uploading CloudFormation file to S3...
Serverless: Uploading artifacts...
Serverless: Uploading service my-service.zip file to S3 (38.3 KB)...
Serverless: Validating template...
Serverless: Skipping template validation: Unsupported in Localstack
Serverless: Updating Stack...
Serverless: Checking Stack update progress...
.....................................
Serverless: Stack update finished...
Service Information
service: my-service
stage: local
region: us-east-1
stack: my-service-local
resources: 35
api keys:
None
endpoints:
http://localhost:4566/restapis/XXXXXXXXXX/local/_user_request_
functions:
...
layers:
None
Use the displayed endpoint http://localhost:4566/restapis/XXXXXXXXXX/local/_user_request_/my/custom/endpoint
to make requests to the deployed service.
serverless-localstack supports a feature for lambda functions that allows local code mounting:
# serverless.yml
custom:
localstack:
# ...
lambda:
mountCode: True
When this flag is set, the lambda code will be mounted into the container running the function directly from your local directory instead of packaging and uploading it.
By default localstack generates random deployment ids for the API gateway which in turn get populated into the endpoint URLs. As these IDs can literally change every time you deploy (depending on your setup) This makes continual local development practically impossible in situations where you rely on consistant urls (eg the url for your api for your frontend code)
localstack provides a method to use a fixed custom id instead by passing a tag custom_id to the API during creation. It has to be done during creation and it has to be on either the v2 API or the v1 RestApi.
Serverless can do this out of the box with resource extensions. See below for the v1/v2 implementation.
This will produce endpoints like http://localhost:4566/restapis/mytag/development/_user_request_ (v1)
or for the newer localstack http://localhost:4566/_aws/execute-api/mytag/development
rather than a continually changing ID http://localhost:4566/restapis/jh345798dx/development/_user_request_
resources:
extensions:
ApiGatewayRestApi: #for v1
Properties:
Tags:
- Key: _custom_id_
Value: mytag
MyHTTPApi: #for v2
Properties:
Tags:
- Key: _custom_id_
Value: mytag
To get hot-reload up and running quickly, put the following into a script before you run sls deploy.
You need the region set to the same region in your config otherwise the bucket and ssm param can't be found by the sls deployment
- note: this requires the localstack awslocal cli to be installed
awslocal s3api create-bucket --bucket hot-reload --create-bucket-configuration LocationConstraint=ap-southeast-2
AWS_DEFAULT_REGION=ap-southeast-2 awslocal ssm put-parameter \
--name "/serverless-framework/deployment/s3-bucket" \
--type "String" \
--value "{\"bucketName\":\"hot-reload\",\"bucketRegion\":\"ap-southeast-2\"}"
Then in your serverless.yaml use the custom variables to set the names. This uses the default option to set the name to the same as serverless does when deploying
custom:
development:
bucket: 'hot-reload'
provider:
deploymentBucket: ${self:custom.${self:provider.stage}.bucket, serverless-deployment-bucket-${self:service}-${self:provider.stage}}
If you run into any issues or problems while integrating LocalStack with your Serverless app, please submit an issue.