This is a utility project designed to cater neccessities for warming up Lambda functions to prevent cold starts.

Table Of Contents

• Table Of Contents
• Installing Warmer
• Using Warmer
• Setting up Event Bridge Notifications
• Working on enhancements

Installing Warmer

To install module, run the below command:

python3 -m pip install py_lambda_warmer

# or

python3 -m pip install py_lambda_warmer==<release-version>

This does not specify boto3 library as hard core requirements while installing as it expects lambda environment to already have it in order to reduce the layer size created. This would not be the case with development. You will have to explicitely install dependencies using dev_requirements.txt file. ( Changes made since release 0.1.5 )`

Using Warmer

This is very easy to incorporate in your existing Python Lambda Handlers. Follow the below code.

from warmer import warmer
@warmer(flag="custom_event_key", _concurrency=1)
def handler(event, context):
    pass

Parameters: flag (type: str)- name of the event flag to look for _concurrency (type: int)- (optional) Number of concurrent handlers to warm up, default: 1

If your handler is a Flask/FastApi application, you may follow below steps:

from warmer import warmer
from flask import Flask
app = Flask()
@warmer(flag="custom_event_key",_concurrency=1)
def application(event, context):
    return app(event, context)

# or

application = warmer(flag="custom_event_key",_concurrency=1)(app)

# you may now use application as your handler

warmer will help you cater the custom events that are coming for warming Lambda function. Though _concurrency is optional and by default it only warms up current execution. In case you want to warm up multiple instances of lambda handler, you may need to adjust _concurrency to number of handlers running.

Warmer uses threading mechnism to ensure that the warming calls are actually happening concurrently and not serially.

Setting up Event Bridge Notifications

You can also setup you custom event bridge schedule for Lambda function using the Terraform Resource code mentioned in the repository.

Simply download the Terraform code attached in the release and unzip it.

Note: Make sure your function execution role has required permission to invoke your Lambda function else concurrent call executions might fail.

wget https://github.com/satyamsoni2211/LambdaWarmerPy/releases/download/${release}/terraform_code.zip
unzip terraform_code.zip -d terraform_code/
cd terraform_code/
# creating variable file required by terraform
cat << EOF > .auto.tfvars
arn = <arn of your lambda function>
profile = <profile alias for aws>
region = <region for aws lambda>
EOF
# initiating and applying
terraform init
terraform plan -out tfplan
terraform apply tfplan

You may also modify resource names as per your requirements in the code.

You may also use AWS SAM template to create Event Bridge Notifications to warm up Lambda function.

TransactionCompsAPI:
  Type: "AWS::Serverless::Function"
  Properties:
    FunctionName: fake-function
    Policies:
        - LambdaInvokePolicy:
            FunctionName: "fake-function"
    Events:
      WarmerSchedule: # add this event to the same template
        Type: Schedule
        Properties:
          Schedule: cron(*/5 * ? * 2-6 *)
          Name: fake-function-warmer-event
          Description: Warmer Event for Lambda Function
          Enabled: true
          Input: '{"warmer": true}' # this refers to the warmer flag

In case you want to include concurrent executions, you may add below to include concurrent invocations.

Note: Using concurrency would also require you to add sufficient permissions to the role to call lambda:invokeFunction action on the Lambda function. Below code snippet includes policy to grant action to the role.

TransactionCompsAPI:
  Type: "AWS::Serverless::Function"
  Properties:
    FunctionName: fake-function
    Policies:
        - LambdaInvokePolicy:
            FunctionName: "fake-function"
    Events:
      WarmerSchedule: # add this event to the same template
        Type: Schedule
        Properties:
          Schedule: cron(*/5 * ? * 2-6 *)
          Name: fake-function-warmer-event
          Description: Warmer Event for Lambda Function
          Enabled: true
          Input: '{"warmer": true, "concurrency": 5}' # this refers to the warmer flag and concurrency

Working on enhancements

If you want to work on enhancements or development, you may clone the project and run the below commands to setup environment:

python -m pip install pipenv
pipenv shell

# or

python -m pip install virtualenv
virtualenv venv
source venv/bin/activate
python -m pip install -r dev_requirements.txt

You may also raise a PR to get merged into existing project.

Happy Warming.