Script
caution
This is the SST v1.x Constructs doc. SST v2 is now released. If you are using v2, see the v2 Constructs doc. If you are looking to upgrade to v2, check out the upgrade steps.
The Script
construct is a higher level CDK construct that makes it easy to run a script in a Lambda function during the deployment process. It provides a simple way to build and bundle the script function; and allows you to pass parameter values based on outputs from other constructs in your SST app. So you don't have to hard code values in your script. You can configure a script to run before or after any of the stacks or resources are deployed in your app.
Since the script is running inside a Lambda function, it can interact with resources like the RDS databases, that are inside a VPC; and make AWS API calls to services that the IAM credentials in your local environment or CI might not have permissions to.
A few things to note:
- It does not run locally. It runs inside a Lambda function.
- It gets run on every deployment.
- It can run for a maximum of 15 minutes.
- Live Lambda Dev is not enabled for these functions.
Examples
Minimal config
import { Script } from "@serverless-stack/resources";
new Script(stack, "Script", {
onCreate: "src/function.create",
onUpdate: "src/function.update",
onDelete: "src/function.delete",
});
Configuring parameters
The params
will be passed in as the event
object to the function.
import { Table, Script } from "@serverless-stack/resources";
const table = new Table(this, "Table", {
fields: {
userId: "string",
},
primaryIndex: { partitionKey: "userId" },
});
new Script(stack, "Script", {
onCreate: "src/script.create",
params: {
hello: "world",
tableName: table.tableName,
},
});
So in the above example, the event.params.tableName
will be available in the onCreate function in src/script.create
.
Note that, the value for tableName
will be resolved at deploy time. For example, in this case, the Table
construct will get created first, and the Script
construct will be run afterwards. And if you were to print out the value of event.params.tableName
inside the onCreate function, you will see the name of the table.
Configuring functions
Specifying function props for all the functions
You can extend the minimal config, to set some function props and have them apply to all the functions.
new Script(stack, "Script", {
defaults: {
function: {
timeout: 20,
environment: { tableName: table.tableName },
permissions: [table],
},
},
onCreate: "src/script.create",
});
Configuring an individual function
Configure each Lambda function separately.
new Script(stack, "Script", {
onCreate: {
srcPath: "src/",
handler: "script.create",
environment: { tableName: table.tableName },
permissions: [table],
},
});
Note that, you can set the defaults.function
while configuring a Lambda function. The function's props will just override the defaults.function
. Except for the environment
, the layers
, and the permissions
properties, that will be merged.
new Script(stack, "Script", {
defaults: {
function: {
timeout: 20,
environment: { tableName: table.tableName },
permissions: [table],
},
},
onCreate: {
handler: "src/script.create",
timeout: 10,
environment: { bucketName: bucket.bucketName },
permissions: [bucket],
},
onUpdate: "src/script.update",
});
So in the above example, the onCreate
function doesn't use the timeout
that is set in the defaults.function
. It'll instead use the one that is defined in the function definition (10 seconds
). And the function will have both the tableName
and the bucketName
environment variables set; as well as permissions to both the table
and the bucket
.
Attaching permissions
You can grant additional permissions to the script.
const script = new Script(stack, "Script", {
onCreate: "src/script.create",
onUpdate: "src/script.update",
onDelete: "src/script.delete",
});
script.attachPermissions(["s3"]);
Running before deploy
You can configure the Script
to run at the beginning of the deployment, before any resources are deployed.
Create a stack for the construct. Let's call it BeforeDeployStack
and add it to your stacks/index.js
.
import { dependsOn, StackContext, Script } from "@serverless-stack/resources"
function BeforeDeployStack({stack}: StackContext) {
new Script(stack, "BeforeDeploy", {
onCreate: "src/script.create",
});
}
function ApiStack(ctx: StackContext) {
dependsOn(BeforeDeployStack)
}
function DBStack(ctx: StackContext) {
dependsOn(BeforeDeployStack)
}
By making both ApiStack
and DBStack
depend on BeforeDeployStack
, they will get deployed after BeforeDeployStack
is done deploying.
Here we are making use of the idea of Stack dependencies in CDK.
Now when you deploy this app, the BeforeDeployStack
will get deployed first, which runs the Script
.
Note that, if the script fails to run, the deploy fails. And the ApiStack
and the DBStack
will not get deployed. In this case, you can fix the script, and deploy again.
Running after deploy
Similarly, you can configure a Script
to run at the end of the deployment, after all resources are deployed.
Create a AfterDeployStack
in stacks/index.js
.
import { dependsOn, StackContext, Script } from "@serverless-stack/resources"
function AfterDeployStack({stack}: StackContext) {
dependsOn(ApiStack)
dependsOn(DBStack)
new Script(stack, "AfterDeploy", {
onCreate: "src/script.create",
});
}
function ApiStack(ctx: StackContext) {
}
function DBStack(ctx: StackContext) {
}
By making the AfterDeployStack
depend on both ApiStack
and DBStack
, it will get deployed after the two stacks are done deploying.
Here we are making use of the idea of Stack dependencies in CDK.
Now when you deploy this app, the AfterDeployStack
will get deployed at the end and run the Script
.
Note that, if the script fails to run, the entire deploy is marked as failed. And the updates made to the ApiStack
and the DBStack
will get rolled back. In this case, you can fix the script, and deploy again.
Running multiple Scripts
Multiple scripts within the same Stack can run concurrently. You can manage the order in which they get run by specifying a dependency relationship.
const scriptA = new Script(stack, "Script", {
onCreate: "src/scriptA.create",
});
const scriptB = new Script(stack, "Script", {
onCreate: "src/scriptB.create",
});
scriptB.node.addDependency(scriptA);
In this case, scriptB
will run after scriptA
is completed.
Here we are making use of the idea of Construct dependencies in CDK.
Constructor
new Script(scope, id, props)
Parameters
- scope Construct
- id string
- props ScriptProps
ScriptProps
defaults.function?
Type : FunctionProps
The default function props to be applied to all the Lambda functions in the API. The environment
, permissions
and layers
properties will be merged with per route definitions if they are defined.
new Script(stack, "Api", {
defaults: {
function: {
timeout: 20,
}
}
});
onCreate?
Type : string | Function | FunctionProps
Creates the function that runs when the Script is created.
new Script(stack, "Api", {
onCreate: "src/function.handler",
})
onDelete?
Type : string | Function | FunctionProps
Create the function that runs when the Script is deleted from the stack.
new Script(stack, "Api", {
onDelete: "src/function.handler",
})
onUpdate?
Type : string | Function | FunctionProps
Creates the function that runs on every deploy after the Script is created
new Script(stack, "Api", {
onUpdate: "src/function.handler",
})
params?
Type : Record<string, any>
An object of input parameters to be passed to the script. Made available in the event
object of the function.
import { Script } from "@serverless-stack/resources";
new Script(stack, "Script", {
onCreate: "src/script.create",
params: {
hello: "world",
},
});
Properties
An instance of Script
has the following properties.
createFunction?
Type : Function
The internally created onCreate Function
instance.
deleteFunction?
Type : Function
The internally created onDelete Function
instance.
updateFunction?
Type : Function
The internally created onUpdate Function
instance.
Methods
An instance of Script
has the following methods.
attachPermissions
attachPermissions(permissions)
Parameters
- permissions Permissions
Grants additional permissions to the script
script.attachPermissions(["s3"]);
bind
bind(constructs)
Parameters
- constructs Array<SSTConstruct>
Binds additional resources to the script
script.bind([STRIPE_KEY, bucket]);