Building layers
You can use Amazon SAM to build custom layers. For information about layers, see Amazon Lambda layers in the Amazon Lambda Developer Guide.
To build a custom layer, declare it in your Amazon Serverless Application Model (Amazon SAM) template file and include a
Metadata
resource attribute section with a BuildMethod
entry. Valid
values for BuildMethod
are identifiers for an Amazon Lambda runtime, or makefile
.
Include a BuildArchitecture
entry to specify the instruction set architectures that
your layer supports. Valid values for BuildArchitecture
are Lambda instruction set
architectures.
If you specify makefile
, provide the custom makefile, where you declare a build
target of the form build-
that contains
the build commands for your layer. Your makefile is responsible for compiling the layer if
necessary, and copying the build artifacts into the proper location required for subsequent
steps in your workflow. The location of the makefile is specified by the layer-logical-id
ContentUri
property of the layer resource, and must be named Makefile
.
Note
When you create a custom layer, Amazon Lambda depends on environment variables to find your
layer code. Lambda runtimes include paths in the /opt
directory where your layer
code is copied into. Your project's build artifact folder structure must match the runtime's
expected folder structure so your custom layer code can be found.
For example, for Python you can place your code in the python/
subdirectory.
For NodeJS, you can place your code in the nodejs/node_modules/
subdirectory.
For more information, see Including library dependencies in a layer in the Amazon Lambda Developer Guide.
The following is an example Metadata
resource attribute section.
Metadata: BuildMethod: python3.8 BuildArchitecture: arm64
Note
If you don't include the Metadata
resource attribute section, Amazon SAM doesn't
build the layer. Instead, it copies the build artifacts from the location specified in the
CodeUri
property of the layer resource. For more information, see the ContentUri property of the
AWS::Serverless::LayerVersion
resource type.
When you include the Metadata
resource attribute section, you can use the
sam build
command to build the layer,
both as an independent object, or as a dependency of an Amazon Lambda function.
-
As an independent object. You might want to build just the layer object, for example when you're locally testing a code change to the layer and don't need to build your entire application. To build the layer independently, specify the layer resource with the
sam build
command.layer-logical-id
-
As a dependency of a Lambda function. When you include a layer's logical ID in the
Layers
property of a Lambda function in the same Amazon SAM template file, the layer is a dependency of that Lambda function. When that layer also includes aMetadata
resource attribute section with aBuildMethod
entry, you build the layer either by building the entire application with thesam build
command or by specifying the function resource with thesam build
command.function-logical-id
Examples
Template example 1: Build a layer against the Python 3.9 runtime environment
The following example Amazon SAM template builds a layer against the Python 3.9 runtime environment.
Resources: MyLayer: Type: AWS::Serverless::LayerVersion Properties: ContentUri: my_layer CompatibleRuntimes: - python3.9 Metadata: BuildMethod: python3.9 # Required to have Amazon SAM build this layer
Template example 2: Build a layer using a custom makefile
The following example Amazon SAM template uses a custom makefile
to build the
layer.
Resources: MyLayer: Type: AWS::Serverless::LayerVersion Properties: ContentUri: my_layer CompatibleRuntimes: - python3.8 Metadata: BuildMethod: makefile
The following makefile
contains the build target and commands that will be
executed. Note that the ContentUri
property is set to my_layer
, so
the makefile must be located in the root of the my_layer
subdirectory, and the
filename must be Makefile
. Note also that the build artifacts are copied into
the python/
subdirectory so that Amazon Lambda will be able to find the layer
code.
build-MyLayer: mkdir -p "$(ARTIFACTS_DIR)/python" cp *.py "$(ARTIFACTS_DIR)/python" python -m pip install -r requirements.txt -t "$(ARTIFACTS_DIR)/python"
Example sam build commands
The following sam build
commands build layers that include the
Metadata
resource attribute sections.
# Build the 'layer-logical-id' resource independently
$
sam build
# Build the 'function-logical-id' resource and layers that this function depends onlayer-logical-id
$
sam build
# Build the entire application, including the layers that any function depends onfunction-logical-id
$
sam build