Skip to content

Latest commit

 

History

History
211 lines (176 loc) · 11.7 KB

README.md

File metadata and controls

211 lines (176 loc) · 11.7 KB
Raw

README

What is this?

sceptre-sam-handler is a TemplateHandler for Sceptre (versions 2.7 and up) that lets you use an AWS SAM template (and its associated project) as a stack's template.

This template handler will run sam build and then sam package from the indicated SAM Template's directory in order to compile and generate a CloudFormation-ready template. Additionally, if you define your SAM template with a .j2 extension, you can utilize Jinja2 templating syntax and logic to render the SAM template prior to the build.

By using the SAM Handler, you are letting SAM build your application, compile a SAM template, and upload artifacts to S3, and then using Sceptre to actually do the deployment of the template to a stack. In other words, by using this handler with Sceptre, you skip ever using sam deploy; It's not needed. You also shouldn't need a sam config file with deployment defaults, since you'll be using Sceptre to deploy rather than SAM.

By using this handler, you can now use SAM templates with all your favorite Sceptre commands, like launch, validate, generate, and diff (along with all the rest)!

How to install sceptre-sam-handler

Simply pip install scepre-sam-handler. Additionally, you need SAM CLI installed and accessible on the PATH for the handler to invoke as a subprocess.

There are three main ways you can install SAM CLI:

  • You can follow AWS's own documentation on how to install SAM for your operating system. (Note: at least on Linux, this requires the ability to use sudo; If you need to install SAM where permissions escalation is not possible, this won't work.)
  • You can use pipx to install aws-sam-cli into an isolated virtual environment where it can have exclusive claim to its dependency versions. This can be done without privilege escalations.
  • If you want to install aws-sam-cli along with this handler using pip, you can use the "extra" like pip install sceptre-sam-handler[sam]. However, using pip to install SAM is generally not recommended, according to SAM's own documentation. This can lead to dependency conflicts, since SAM CLI is particular about dependency versions.

How to use sceptre-sam-handler

The template "type" for this handler is sam. There are two file extensions supported by this handler:

  • .yaml: Use this for a "normal" SAM template according to the SAM specification. This template will be directly sent to the SAM CLI for building and packaging.
  • .j2: Use this if you need to use Jinja2 templating syntax and logic in order to render a SAM template, such as to interpolate values into the template prior to building it. See the section below on Jinja SAM templates for more details.

This handler takes several arguments, two of which are required.

Arguments:

  • path (string, required): The path from the current working directory (NOT the project path) to the SAM Template. The path must end in either ".yaml" or ".j2".
  • artifact_bucket_name (string, required): The bucket name where artifacts should be uploaded to on S3 during the packaging process. If your project has a template_bucket_name, you can set this to {{ template_bucket_name }}.
  • artifact_prefix (string, optional): The prefix to apply to artifacts uploaded to S3. This can be the project's {{ template_key_prefix }}.
  • build_args (dict, optional): Additional key/value pairs to supply to sam build. For flag-type arguments that have no value, set the value to "True".
  • package_args (dict, optional): Additional key/value pairs to apply to sam package. The same is true here as for build_args for flag-type arguments.
  • skip_jinja_cleanup (bool): Can be set to True to retain the jinja file after processing. This can be used to troubleshoot issues.

How does this handler work?

When using only sam CLI (not Sceptre) to deploy using sam deploy, SAM CLI effectively performs 3 steps:

  1. SAM CLI builds the all the various resources special SAM resources, resolving dependencies. These would include Lambda functions and Lambda layers. It copies any locally-referenced files and resolves any dependencies into a directory called .aws-sam. This is the same behavior as running sam build.
  2. SAM CLI then transforms all SAM template URIs that reference local filepaths to S3 keys (among other) transformations it applies, uploads any built artifacts to those s3 keys, and saves the transformed template. This is the same behavior as running sam package.
  3. SAM CLI finally takes that transformed template (along with a local sam config and any other CLI arguments) and performs CloudFormation stack create/update with them.

When you use Sceptre with this handler, the SAM handler performs steps 1-2 above to create a template that Sceptre can use, but it does not use SAM to deploy it!. Instead, Sceptre can use that template produced in step 2 above (via sam package) to perform all it's usual commands with all it's usual magic!

In other words, using this handler lets you use resolvers, put your SAM stack into StackGroups, let you name your stack according to Sceptre's naming conventions, validate, diff, and more! Basically, the SAM stack can be managed using Sceptre just like any other.

Default behavior

SAM commands are invoked using the system shell in a subprocess, with stdout redirected to stderr. Artifacts will be uploaded using the artifact_bucket_name and artifact_prefix arguments, the project_code, and the Sceptre stack name.

For example, given an artifact_bucket_name of "bucket", artifact_prefix of "prefix", a project_code of "project" and a stack config located at "config/indigo/sam-application.yaml", SAM artifacts will be uploaded to:

s3://bucket/prefix/project/indigo/sam-application/sam_artifacts/

By default, these will be the sam commands that are run from the template's directory:

sam build --cached --template-file [path as absolute path]
sam package \
  --s3-bucket [artifact_bucket_name argument] \
  --region [the stack region] \
  --s3-prefix [the prefix described above] \
  --template-file [path as absolute path]

If any additional arguments are desired for to be passed to SAM, you can specify those with dicts for the build_args and package_args template handler arguments. These key/value pairs will override the defaults. For any flag-type arguments, set the value to True. If you want to remove a default argument (such as the --cached flag for sam build), set the value to None.

IAM and authentication

This handler uses the stack's connection information to generate AWS environment variables and sets those on the sam process, ensuring that the AWS authentication configuration on the stack config and project is carried over to SAM without any need for additional arguments.

If you desire to use a different profile or region when invoking sam package than what is set on the stack, you should specify "profile" and/or "region" values for "package_args".

Important: SAM creates CloudFormation-ready templates via sam package, which uploads built artifacts to S3 in the process. This means that Sceptre commands that do not normally require S3 actions (such as generate, validate, diff, and others) will require them when using this handler. You will need to ensure that any user or role executing these commands has proper permissions for these operations. For more information on required permissions, see the documentation for SAM permissions.

Jinja SAM Templates

The SAM Handler supports using SAM templates that have Jinja logic in them. These Jinja templates will have access to the sceptre_user_data just like Jinja templates via the normal file handler do. This can be useful for implementing additional template logic (such as loops and other actions) in the template.

If you need to pass variables into the Jinja template for reference via Jinja syntax, you should pass those variables via sceptre_user_data. Remember, resolvers can be used with sceptre_user_data, so this can be a powerful tool to pre-render your templates or reference values in areas that a SAM Template cannot use parameters (such as in Transforms).

Resolvers in the SAM Handler parameters

It's likely that you'll want to use your template_bucket_name as your artifact_bucket_name, so you don't need to have a separate bucket for your sam artifacts. However, since template_bucket_name is technically a resolvable property that could be set via !stack_output on your StackGroup config, you cannot directly reference it in your Stack Config with {{ template_bucket_name }} if you defined it on the StackGroup Config using a resolver. For more information on why you can't do this, you should read about the Resolution order of values for Sceptre Stack Configs on Sceptre's documentation.

Nonetheless, the bottom line is this: {{ template_bucket_name }} may or may not actually work for you, but !stack_attr template_bucket_name will always work for you, no matter how you've set up your project.

Example Stack Config (simple .yaml)

# By using the SAM handler, you let SAM build and package the template and upload artifacts to S3
# and Sceptre will use the packaged template to create the CloudFormation stack, using the stack
# config.
template:
    type: sam
    path: path/from/my/cwd/template.yaml
    # If your template_bucket_name is inherited from your StackGroup as a resolver, you cannot use
    # {{ template_bucket_name }} to reference it, so you need to use the !stack_attr resolver. But
    # using !stack_attr will always work for you anyway, so you might as well configure the SAM
    # handler that way.
    artifact_bucket_name: !stack_attr template_bucket_name
    # It makes the most sense to use the same prefix as your template_key_prefix so that your SAM
    # artifacts are foldered similar to your other templates... but it's not strictly necessary.
    artifact_prefix: !stack_attr template_key_prefix
    build_args:
        use-container: True

# You can use resolvers to pass parameters, just like any other Sceptre stack!
parameters:
    long_parameter: !file my/file/path
    my_template_parameter: !stack_output some/other/stack.yaml::SomeOutput

# The SAM Handler will work with all the other stack parameters you might want to use too!
profile: my_profile
iam_role: arn:aws:iam::1111111111:role/My-Deployment-Role
region: us-east-1
stack_tags:
    SomeTag: SomeValue

Example Stack Config (Using Jinja2 .j2 template)

template:
    type: sam
    path: path/from/my/cwd/template.j2
    artifact_bucket_name: !stack_attr template_bucket_name
    artifact_prefix: !stack_attr template_key_prefix

# Remember, Jinja logic cannot access parameter values; Those are accessed via CloudFormation functions
# like !Ref and !Sub when the stack is being deployed. If you need values to use with your Jinja logic,
# use sceptre_user_data instead.
parameters:
    my_template_parameter: !stack_output some/other/stack.yaml::SomeOutput

# sceptre_user_data is resolved PRIOR TO building and deploying the template and it is passed to
# Jinja. So you can use sceptre_user_data to control Jinja logic or render values into the template.
# And because sceptre_user_data is resolvable, you can use resolvers to pass values and even whole
# template segments to render into the final SAM template before SAM build is ever invoked.
sceptre_user_data:
     template_segmant: !file my/template/segment