Dark Mirror

Dark mirror is a set of lambda functions which fetch data from different sources as news feeds and apis, weather and randomically generate playlists from your spotify account to you can listen diffent kind of music everyday.

Made with

• Python 3+
• Kubernetes
• Serverless
• Kubeless

News feed

There are tree major feed news we are going to use here:

- BBC
- The New York Times
- The Guardian

Weather

For wheather, we are going to fetch the data from AccuWeather which seems to be pretty reliable and very detailed source.

Spotify

The idea of this service is to access your account and generate a short playlist with a music spotify suggest you but maybe you never saw it in your service client.

Setup and deploy

There are a few steps you must to execute to maintain this project and also deploy all services available here.

First things first you must have access to kubernetes or running kubernetes locally. If you want, I also have a project that helps you to set uo k8 locally. You can find the project scripts here: https://gitlab.com/bfelipe/k8-local

Also you must have installed serveless framework, which relly on nodejs to handle serverless development and deployment easy.

So. Once you have everything up and running you want to go into the service folder you want to deploy and install the dependencies serverless requires to work with your lambda functions. To do that you just need to execute:

npm install

You going to see a folder called node_modules inside the service folder. Now you just need to execute the command bellow to deploy your service into kubernetes:

serverless deploy

You don't need to worry about packing the dependencies each lambda function needs because we are using kubeless. This guy will take care of everything about download the project dependencies.

The function must be available once serverless finished the deploy process. To check if it is created properlly, you can execute:

kubectl get functions

If you have more than once function it will be listed here. If not you are going to see something like:

NAME            AGE
<function-name>   00s

Now you must check if your function is actually running into a pod. To do that you must execute:

kubectl get pods

The function you just deployed should be listed there with a generated code.

Exposing our function

We already have our function deployed, but still we need to expose them to be able to request from an url. You still can test your function following the same steps described at "Testing serverless function", but let's make it something easy to access and handle. First lets create a trigger for our functions. Righ now kubeless has tree types of triggers(http, cronjob and pubsub), once you create a trigger, kubeless will handle to create an ingress object for your function, so no need to write any yaml file for that. To expose your function you should type in your console:

kubeless trigger <trigger-type> create <trigger-name> -n <namespace> --function-name <function-name> --hostname <ex. my-domain.io> --path <ex. hello>

IMPORTANT: For your path, you do not need to specify / before your path name.

Now you can check if your function has been exposed by typing:

kubectl get ing -n <namespace>

You going to see something similar to that:

NAME                HOSTS           ADDRESS      PORTS   AGE
<function-name>     my-domain.io    000.00.0.0   80      00s

For more information about your function you can use:

kubectl describe ing <trigger-name> -n <namespace>

or kubectl logs -n kubeless -l kubeless=controller -c http-trigger-controller

If you are running locally in your machine, you should configure /etc/hosts with the same info you used to create your trigger.

ex. /etc/hosts

000.000.0.0     my-domain.io

Now you can call your function using the follor curl, or just typing the same url into your browser:

curl --header "Content-Type:application/json" http://<hostname>/<path>

Testing serverless functions

There are two ways to test your function without exposing it.

The first way, you can simply invoke it by using serverless framework, typing into your console:

serverless invoke --function <function-name> --log

If your function is set to return a response you are going to get it after the call.

The second way is by calling your function using kubeless cli.

kubeless function call <function-name>

In both cases if your function is expencting to get an input data, everything you need to do is to add '--data' with the actual data you want to send to the lambda function.

To Do

• Add the guardian get_news function
• Add the new york times get_news function
• Add spotify random playlist function