Lambda Event Sources

Scheduled Events

Chalice has support for scheduled events. This feature allows you to periodically invoke a lambda function based on some regular schedule. You can specify a fixed rate or a cron expression.

To create a scheduled event in chalice, you use the @app.schedule() decorator. Let’s look at an example.

app = chalice.Chalice(app_name='foo')

@app.schedule('rate(1 hour)')
def every_hour(event):
    print(event.to_dict())

In this example, we have a single lambda function that we want automatically invoked every hour. When you run chalice deploy Chalice will create a lambda function as well as the necessary CloudWatch events/rules such that the every_hour function is invoked every hour.

The Chalice.schedule() method accepts either a string or an instance of Rate or Cron. For example:

app = chalice.Chalice(app_name='foo')

@app.schedule(Rate(1, unit=Rate.HOURS))
def every_hour(event):
    print(event.to_dict())

The function you decorate must accept a single argument, which will be of type CloudWatchEvent.

You can use the schedule() decorator multiple times in your chalice app. Each schedule() decorator will result in a new lambda function and associated CloudWatch event rule. For example:

app = chalice.Chalice(app_name='foo')

@app.schedule(Rate(1, unit=Rate.HOURS))
def every_hour(event):
    print(event.to_dict())


@app.schedule(Rate(2, unit=Rate.HOURS))
def every_two_hours(event):
    print(event.to_dict())

In the app above, chalice will create two lambda functions, and configure every_hour to be invoked once an hour, and every_two_hours to be invoked once every two hours.

CloudWatch Events

You can configure a lambda function to subscribe to any CloudWatch Event <https://amzn.to/2SCgWA6>.

To subscribe to a CloudWatch Event in chalice, you use the @app.on_cw_event() decorator. Let’s look at an example.

app = chalice.Chalice(app_name='foo')

@app.on_cw_event({"source": ["aws.codecommit"]})
def on_code_commit_changes(event):
    print(event.to_dict())

In this example, we have a single lambda function that we subscribe to all events from the AWS Code Commit service. The first parameter to the decorator is the event pattern that will be used to filter the events sent to the function.

See the CloudWatch Event pattern docs for additional syntax and examples https://amzn.to/2OlqZso

The function you decorate must accept a single argument, which will be of type CloudWatchEvent.

S3 Events

You can configure a lambda function to be invoked whenever certain events happen in an S3 bucket. This uses the event notifications feature provided by Amazon S3.

To configure this, you just tell Chalice the name of an existing S3 bucket, along with what events should trigger the lambda function. This is done with the Chalice.on_s3_event() decorator.

Here’s an example:

from chalice import Chalice

app = chalice.Chalice(app_name='s3eventdemo')
app.debug = True

@app.on_s3_event(bucket='mybucket-name',
                 events=['s3:ObjectCreated:*'])
def handle_s3_event(event):
    app.log.debug("Received event for bucket: %s, key: %s",
                  event.bucket, event.key)

In this example above, Chalice connects the S3 bucket to the handle_s3_event Lambda function such that whenver an object is uploaded to the mybucket-name bucket, the Lambda function will be invoked. This example also uses the .bucket and .key attribute from the event parameter, which is of type S3Event.

It will automatically create the appropriate S3 notification configuration as needed. Chalice will also leave any existing notification configuration on the mybucket-name untouched. It will only merge in the additional configuration needed for the handle_s3_event Lambda function.

Warning

This feature only works when using chalice deploy. Because you configure the lambda function with the name of an existing S3 bucket, it is not possible to describe this using a CloudFormation/SAM template. The chalice package command will fail. You will eventually be able to request that chalice create a bucket for you, which will support the chalice package command.

The function you decorate must accept a single argument, which will be of type S3Event.

SNS Events

You can configure a lambda function to be automatically invoked whenever something publishes to an SNS topic. Chalice will automatically handle creating the lambda function, subscribing the lambda function to the SNS topic, and modifying the lambda function policy to allow SNS to invoke the function.

To configure this, you just need the name of an existing SNS topic you’d like to subscribe to. The SNS topic must already exist.

Below is an example of how to set this up. The example uses boto3 to create the SNS topic. If you don’t have boto3 installed in your virtual environment, be sure to install it with:

$ pip install boto3

First, we’ll create an SNS topic using boto3.

$ python
>>> import boto3
>>> sns = boto3.client('sns')
>>> sns.create_topic(Name='my-demo-topic')
{'TopicArn': 'arn:aws:sns:us-west-2:12345:my-demo-topic',
 'ResponseMetadata': {}}

Next, we’ll create our chalice app:

$ chalice new-project chalice-demo-sns
$ cd chalice-demo-sns/

We’ll update the app.py file to use the on_sns_message decorator:

from chalice import Chalice

app = chalice.Chalice(app_name='chalice-sns-demo')
app.debug = True

@app.on_sns_message(topic='my-demo-topic')
def handle_sns_message(event):
    app.log.debug("Received message with subject: %s, message: %s",
                  event.subject, event.message)

We can now deploy our chalice app:

$ chalice deploy
Creating deployment package.
Creating IAM role: chalice-demo-sns-dev
Creating lambda function: chalice-demo-sns-dev-handle_sns_message
Subscribing chalice-demo-sns-dev-handle_sns_message to SNS topic my-demo-topic
Resources deployed:
  - Lambda ARN: arn:aws:lambda:us-west-2:123:function:...

And now we can test our app by publishing a few SNS messages to our topic. We’ll do this using boto3. In the example below, we’re using list_topics() to find the ARN associated with our topic name before calling the publish() method.

$ python
>>> import boto3
>>> sns = boto3.client('sns')
>>> topic_arn = [t['TopicArn'] for t in sns.list_topics()['Topics']
...              if t['TopicArn'].endswith(':my-demo-topic')][0]
>>> sns.publish(Message='TestMessage1', Subject='TestSubject1',
...             TopicArn=topic_arn)
{'MessageId': '12345', 'ResponseMetadata': {}}
>>> sns.publish(Message='TestMessage2', Subject='TestSubject2',
...             TopicArn=topic_arn)
{'MessageId': '54321', 'ResponseMetadata': {}}

To verify our function was called correctly, we can use the chalice logs command:

$ chalice logs -n handle_sns_message
2018-06-28 17:49:30.513000 547e0f chalice-demo-sns - DEBUG - Received message with subject: TestSubject1, message: TestMessage1
2018-06-28 17:49:40.391000 547e0f chalice-demo-sns - DEBUG - Received message with subject: TestSubject2, message: TestMessage2

In this example we used the SNS topic name to register our handler, but you can also use the topic arn. This can be useful if your topic is in another region or account.

SQS Events

You can configure a lambda function to be invoked whenever messages are available on an SQS queue. To configure this, use the Chalice.on_sqs_message() decorator and provide the name of the SQS queue and an optional batch size.

The message visibility timeout of your SQS queue must be greater than or equal to the lambda timeout. The default message visibility timeout when you create an SQS queue is 30 seconds, and the default timeout for a Lambda function is 60 seconds, so you’ll need to modify one of these values in order to successfully connect an SQS queue to a Lambda function.

You can check the visibility timeout of your queue using the GetQueueAttributes API call. Using the AWS CLI, you can run this command to check the value:

$ aws sqs get-queue-attributes \
    --queue-url https://us-west-2.queue.amazonaws.com/1/testq \
    --attribute-names VisibilityTimeout
{
    "Attributes": {
        "VisibilityTimeout": "30"
    }
}

You can set the visibility timeout of your SQS queue using the SetQueueAttributes API call. Again using the AWS CLI you can run this command:

$ aws sqs set-queue-attributes \
    --queue-url https://us-west-2.queue.amazonaws.com/1/testq \
    --attributes VisibilityTimeout=60

If you would prefer to change the timeout of your lambda function instead, you can specify this timeout value using the lambda_timeout config key if your .chalice/config.json file. See Lambda Specific Configuration for a list of all supported lambda configuration values in chalice. In this example below, we’re setting the timeout of our handle_sqs_message lambda function to 30 seconds:

$ cat .chalice/config.json
{
  "stages": {
    "dev": {
      "lambda_functions": {
        "handle_sqs_message": {
          "lambda_timeout": 30
        }
      }
    }
  },
  "version": "2.0",
  "app_name": "chalice-sqs-demo"
}

In this example below, we’re connecting the handle_sqs_message lambda function to the my-queue SQS queue. Note that we are specifying the queue name, not the queue URL or queue ARN. If you are connecting your lambda function to a FIFO queue, make sure you specify the .fifo suffix, e.g. my-queue.fifo.

from chalice import Chalice

app = chalice.Chalice(app_name='chalice-sqs-demo')
app.debug = True

@app.on_sqs_message(queue='my-queue', batch_size=1)
def handle_sqs_message(event):
    for record in event:
        app.log.debug("Received message with contents: %s", record.body)

Whenever a message is sent to the SQS queue our function will be automatically invoked. The function argument is an SQSEvent object, and each record in the example above is of type SQSRecord. Lambda takes care of automatically scaling your function as needed. See Understanding Scaling Behavior for more information on how Lambda scaling works.

If your lambda functions completes without raising an exception, then Lambda will automatically delete all the messages associated with the SQSEvent. You don’t need to manually call sqs.delete_message() in your lambda function. If your lambda function raises an exception, then Lambda won’t delete any messages, and once the visibility timeout has been reached, the messages will be available again in the SQS queue. Note that if you are using a batch size of more than one, the entire batch succeeds or fails. This means that it is possible for your lambda function to see a message multiple times, even if it’s successfully processed the message previously. There are a few options available to mitigate this:

  • Use a batch size of 1 (the default value).
  • Use a separate data store to check if you’ve already processed an SQS message. You can use services such as Amazon DynamoDB or Amazon ElastiCache.
  • Manually call sqs.delete_message() in your Lambda function once you’ve successfully processed a message.

For more information on Lambda and SQS, see the AWS documentation.