serverless-offline-lambda-docker-plugin
v1.0.2
Published
Use Docker images in AWS, but function handlers for Local development with Serverless Offline.
Downloads
52
Maintainers
Readme
serverless-offline-lambda-docker-plugin
Use Docker images in AWS, but function handlers for Local development with Serverless Offline.
The plugin is designed to use Docker images when deploying to AWS environment; and to use the classic file-handler approach when working in the local environment.
Instructions:
- Install the plugin:
npm i --save serverless-offline-lambda-docker-plugin
- Add the plugin to the list of plugins in
serverless.yml
, before theserverless-offline
plugin:
plugins:
- (some plugin)
- serverless-offline-lambda-docker-plugin
- (another plugin)
- serverless-offline
- Configure the plugin in
custom.serverless-offline-lambda-docker-plugin
as follows:
custom:
serverless-offline-lambda-docker-plugin:
enabled: true
dockerImage: your-docker-image:version
localStages:
- offline
- dev
- any_other_stage
functionsToSkip:
- some_function
- another_function
enableTransitionSuffix: false
- enabled: allows to disable the plugin altogether.
- dockerImage: define the image name. You can use
${env:SOME_VALUE}
in case you define the image to use through environment variables. - functionsToSkip: list of functions for which the plugin won't kick in. This let's you disable the plugin for specific functions if needed.
- enableTransitionSuffix: important setting for plugin state changes (either enabling it or disabling it, see documentation below for further information)
- localStages: list of stages the plugin will ignore.
- Create a docker-handler file, that will route traffic to the actual files. Find a Python example below, and feel free to submit a NodeJs example if you get it done:
docker.py
import importlib
import os
def find_module():
# For this to work you need to pass the actual
# function handler as an environment variable
# See https://www.npmjs.com/package/serverless-offline-lambda-docker-plugin for further information
handler_path = os.environ.get("REAL_EVT_HANDLER")
(module_name, function_name) = handler_path.split(".")
return (module_name.replace("/", "."), function_name)
def handler(lambda_event, lambda_context):
(module_name, function_name) = find_module()
loaded_module = importlib.import_module(module_name)
event_handler = getattr(loaded_module, function_name)
return event_handler(lambda_event, lambda_context)
Define your serverless functions as you normally do, having a
handler
key pointing to the event handler.Update your Docker image following the AWS documentation, and making the docker-image to have
docker.handler
as theCMD
:
CMD [ "path/to/file/docker.handler" ]
- Try it with
serverless offline start --stage development
or the stage name you normally use.
** Important **
Serverless-offline requires the command to be serverless offline start
and not serverless offline
alone, otherwise the offline:start:init
and offline:start:end
lifecycle hooks won't kick in.
How does it work?
The plugin updates the function definitions by:
- Adding an
image
key pointing to the image you want to use - Copying the value of the
handler
key as aREAL_EVT_HANDLER
environment variable of the function, which is later on used by the handler. - Removing the
handler
key from the function definition
First deployment after enabling or disabling the plugin
The Lambda documentation specifies that an existing function can't change from a normal handler towards a Docker Image and vice versa. Updating the Lambda resource requires a deletion and a creation.
This means that, in order to start using the plugin (and in order to stop using it) you will need to change the names of all functions, in such a way that the current function is deleted and the new function is created, with a different name.
This plugin offers an option to ease the transition towards using the plugin by enabling and disabling the enableTransitionSuffix
option.
When the enableTransitionSuffix
option is set to true
, the plugin will add the Dkr
suffix to the function key in Serverless. As a result, your previous function won't exist in the stack and, therefore, it will be removed from CloudFormation.
Enabling the plugin
When using the plugin for the first time, set enableTransitionSuffix: true
in the plugin configuration and attempt a deployment. Once the deployment is successful you can set enableTransitionSuffix: false
and deploy again.
Disabling the plugin
When attempting to disable the plugin, set enableTransitionSuffix: true
, keeping the plugin enabled, and attempt to deploy. Upon success, disable the plugin and deploy again. This will revert the function to use normal handlers and will also rename the functions back as they were originally.
To do:
- Add tests
- Include a NodeJs sample file
- Add linter
- Test with Serverless 3
Issues, bug reports and collaboration:
- Issues can be created on the Issues section
- Pull requests are accepted
Disclaimer:
- This plugin won't build any docker image. Building the images and pushing them to an image repository should be done manually
- This plugin wasn't tested with Serverless v3 as I'm using v2.