Created by Kenneth Falck <kennu@nordcloud.com>. Copyright Nordcloud 2017-2021. Released under the MIT license.
This plugin provides support for managing multiple AWS CloudFormation stacks in your Serverless 1.x service (serverless.yml).
The plugin requires Node.js 12.x or later due to use of Object.fromEntries().
Normally, Serverless creates a single CloudFormation stack which contains all your CloudFormation resources. If you ever delete your service, these resources are also deleted, including data stored in S3 or DynamoDB.
The Additional Stacks Plugin allows you to move your "permanent" resources into a separate CloudFormation stack, which is not tightly coupled with the service. If you delete your Serverless service and deploy it again from scratch, the permanent resources and their stored data remain untouched.
The cost of separation is that Ref:
no longer works for directly referencing
resources between stacks. If you need such references, you'll need to define
exported stack output values and then use Fn::ImportValue
to refer to the
value in the other stack.
However, you can also use shared Serverless variables in stack definitions. So
you can define resource names in serverless.yml
and then refer to them
using variables like ${self:custom.tableName}
.
You can use the plugin to manage as many additional stacks as you like
in your serverless.yml, placed under the custom.additionalStacks
section.
To install with npm, run this in your service directory:
npm install --save serverless-plugin-additional-stacks
To install with yarn, run this in your service directory:
yarn add serverless-plugin-additional-stacks
Then add this to your serverless.yml
:
plugins:
- serverless-plugin-additional-stacks
To define additional stacks, add an additionalStacks
section like this
in your serverless.yml
:
custom:
additionalStacks:
permanent:
Resources:
S3BucketData:
Type: AWS::S3::Bucket
Properties:
BucketName: ${self:service}-data
The syntax of the Resources section is identical to the normal resources
in serverless.yml
, so you can just cut-paste resource definitions around.
The full name of the deployed CloudFormation stack will include the service
name. If your service is called my-service
and you deploy it to the dev
stage, the additional stack would be called my-service-dev-permanent
.
The plugin supports some directives you can use to control how your additional stacks are deployed. They are described below.
custom:
additionalStacks:
stackName:
Deploy: After
DeployParameters: []
StackName: CustomName
Resources: ...
By default, additional stacks are deployed before the main Serverless CloudFormation resources. This ensures they are immediately available when your Lambda functions start running.
If you need to deploy an additional stack after other CloudFormation
resources, you can add Deploy: After
to its definition.
By default, stacks are deployed with empty parameters ([]). You can use
DeployParameters: [...]
to specify parameters as an array with the same syntax
that the AWS SDK createStack() and updateStack() functions use. It looks like
this in YAML:
DeployParameters:
- ParameterKey: param1
ParameterValue: value1
- ParameterKey: param2
ParameterValue: value2
By default, additional CloudFormation stacks are named by appending the
specified stackName to the Serverless stack name. If you want to customize
the CloudFormation name of the stack, you can specify StackName: CustomName
.
This is useful for controlling existing stacks.
Your additional stacks will be deployed automatically when you run:
sls deploy
To deploy all additional stacks without deploying the Serverless service, you can use:
sls deploy additionalstacks
To deploy an additional stack individually, you can use:
sls deploy additionalstacks --stack [stackname]
To only deploy the Serverless service without deploying additional stacks, you can use:
sls deploy --skip-additionalstacks
If you want to remove an additional stack, you need to run:
sls remove additionalstacks --stack [stackname]
Or you can remove all additional stacks in the service with:
sls remove additionalstacks
Alternatively, you can remove stacks manually in AWS CloudFormation Console.