Kappa is a command line tool that (hopefully) makes it easier to deploy, update, and test functions for AWS Lambda.
There are quite a few steps involved in developing a Lambda function. You have to:
- Write the function itself
- Create the IAM role required by the Lambda function itself (the executing role) to allow it access to any resources it needs to do its job
- Add additional permissions to the Lambda function if it is going to be used in a Push model (e.g. S3, SNS) rather than a Pull model.
- Zip the function and any dependencies and upload it to AWS Lambda
- Test the function with mock data
- Retrieve the output of the function from CloudWatch Logs
- Add an event source to the function
- View the output of the live function
Kappa tries to help you with some of this. It creates all IAM policies for you based on the resources you have told it you need to access. It creates the IAM execution role for you and associates the policy with it. Kappa will zip up the function and any dependencies and upload them to AWS Lambda. It also sends test data to the uploaded function and finds the related CloudWatch log stream and displays the log events. Finally, it will add the event source to turn your function on.
If you need to make changes, kappa will allow you to easily update your Lambda function with new code or update your event sources as needed.
The quickest way to get kappa is to install the latest stable version via pip:
pip install kappa
Or for the development version:
pip install git+https://github.com/garnaat/kappa.git
To get a feel for how kappa works, let's take a look at a very simple example
contained in the samples/simple
directory of the kappa distribution. This
example is so simple, in fact, that it doesn't really do anything. It's just a
small Lambda function (written in Python) that accepts some JSON input, logs
that input to CloudWatch logs, and returns a JSON document back.
The structure of the directory is:
simple/ ├── _src │ ├── README.md │ ├── requirements.txt │ ├── setup.cfg │ └── simple.py ├── _tests │ └── test_one.json └── kappa.yml.sample
Within the directory we see:
kappa.yml.sample
which is a sample YAML configuration file for the project_src
which is a directory containing the source code for the Lambda function_test
which is a directory containing some test data
The first step is to make a copy of the sample configuration file:
cd simple
cp kappa.yml.sample kappa.yml
Now you will need to edit kappa.yml
slightly for your use. The file looks
like this:
---
name: kappa-simple
environments:
dev:
profile: <your profile here>
region: <your region here>
environment_variables:
<key 1>: <value 1>
<key 2>: <value 2>
policy:
resources:
- arn: arn:aws:logs:*:*:*
actions:
- "*"
prod:
profile: <your profile here>
region: <your region here>
policy:
resources:
- arn: arn:aws:logs:*:*:*
actions:
- "*"
lambda:
description: A very simple Kappa example
handler: simple.handler
runtime: python2.7
memory_size: 128
timeout: 3
log_retention_policy: 7
The name
at the top is just a name used for this Lambda function and other
things we create that are related to this Lambda function (e.g. roles,
policies, etc.).
The environments
section is where we define the different environments into
which we wish to deploy this Lambda function. Each environment is identified
by a profile
(as used in the AWS CLI and other AWS tools) and a
region
. You can define as many environments as you wish but each
invocation of kappa
will deal with a single environment. An environment
can optionally contain environment variables
as key-value pairs. Each
environment section also includes a policy
section. This is where we tell
kappa about AWS resources that our Lambda function needs access to and what
kind of access it requires. For example, your Lambda function may need to
read from an SNS topic or write to a DynamoDB table and this is where you would
provide the ARN (Amazon Resource Name) that identifies those resources.
Since this is a very simple example, the only resource listed here is for
CloudWatch logs so that our Lambda function is able to write to the CloudWatch
log group that will be created for it automatically by AWS Lambda.
The lambda
section contains the configuration information about our Lambda
function. These values are passed to Lambda when we create the function and
can be updated at any time after. log_retention_policy
is an optional
parameter. When supplied, it defines the number of days our Lambda function
Cloudwatch logs kept for. By default, these logs are never removed.
To modify this for your own use, you just need to put in the right values for
profile
and region
in one of the environment sections. You can also
change the names of the environments to be whatever you like but the name
dev
is the default value used by kappa so it's kind of handy to avoid
typing.
Once you have made the necessary modifications, you should be ready to deploy your Lambda function to the AWS Lambda service. To do so, just do this:
kappa deploy
This assumes you want to deploy the default environment called dev
and that
you have named your config file kappa.yml
. If, instead, you called your
environment test
and named your config file foo.yml, you would do this:
kappa --env test --config foo.yml deploy
In either case, you should see output that looks something like this:
kappa deploy
# deploying
# ...deploying policy kappa-simple-dev
# ...creating function kappa-simple-dev
# done
So, what kappa has done is it has created a new Managed Policy called
kappa-simple-dev
that grants access to the CloudWatch Logs service. It has
also created an IAM role called kappa-simple-dev
that uses that policy.
And finally it has zipped up our Python code and created a function in AWS
Lambda called kappa-simple-dev.
To test this out, try this:
kappa invoke _tests/test_one.json
# invoking
# START RequestId: 0f2f9ecf-9df7-11e5-ae87-858fbfb8e85f Version: $LATEST
# [DEBUG] 2015-12-08T22:00:15.363Z 0f2f9ecf-9df7-11e5-ae87-858fbfb8e85f {u'foo': u'bar', u'fie': u'baz'}
# END RequestId: 0f2f9ecf-9df7-11e5-ae87-858fbfb8e85f
# REPORT RequestId: 0f2f9ecf-9df7-11e5-ae87-858fbfb8e85f Duration: 0.40 ms Billed Duration: 100 ms Memory Size: 256 MB Max Memory Used: 23 MB
#
# Response:
# {"status": "success"}
# done
We have just called our Lambda function, passing in the contents of the file
_tests/test_one.json
as input to our function. We can see the output of
the CloudWatch logs for the call and we can see the logging call in the Python
function that prints out the event
(the data) passed to the function. And
finally, we can see the Response from the function which, for now, is just a
hard-coded data structure returned by the function.
Need to make a change in your function, your list of resources, or your
function configuration? Just go ahead and make the change and then re-run the
deploy
command:
kappa deploy
Kappa will figure out what has changed and make the necessary updates for you.
That gives you a quick overview of kappa. To learn more about it, I recommend you check out the tutorial.
Hands up who loves writing IAM policies. Yeah, that's what I thought. With Kappa, there is a simplified way of writing policies and granting your Lambda function the permissions it needs.
The simplified version allows you to specify, in your kappa.yml
file, the
ARN of the resource you want to access, and then a list of the API methods you
want to allow. For example:
policy:
resources:
- arn: arn:aws:logs:*:*:*
actions:
- "*"
To express this using the official IAM policy format, you can instead use a statement:
policy:
statements:
- Effect: Allow
Resource: "*"
Action:
- "logs:*"
Both of these do the same thing.