Flagger

Flagger aims to be a simple feature flag management system.

Goals

• To allow users to sign-up.
• To allow users to authenticate in to the system.
• To allow authenticated users to create projects, environments and flags.
• To allow authenticated users to toggle the status of the flag i.e. on or off.
• To allow applications to fetch the current status of the provided flag.

Tech Stack

Flagger is built using the Go programming language, MongoDB for database, Redis as the cache, and GRPC for communications.

Development

Flagger requires Docker and Kubernetes for running.

0. Copy the sample env files sample.env and secret_sample.env files to k8s/base using:

   cp sample.env k8s/base/.env
   cp secret_sample.env k8s/base/.env.secret

   Once copied enter the values for the variables in k8s/base/.env.secret.

1. Start a Kubernetes cluster on your machine. I've used minikube and so the instructions here are going to be with reference to minikube. Start a minikube cluster using:

   minikube start

2. Set up the Docker environment for minikube using:

   eval $(minikube docker-env)

   This will allow us to build the images directly into minikube.

3. To build the development image, run:

   make build-dev

4. [Optional] To build the debug image, run:

   make build-debug

5. To start the application in development mode, run:

   kubectl apply -k k8s/overlays/dev

6. [Optional] To start the application in debug mode, run:

   kubectl apply -k k8s/overlays/debug

7. When the application first starts the API server will not be able to run as MongoDB replicaset has not been setup up yet. To do so first, we to spin up a new MongoDB pod using:

   kubectl run mongo --image mongo --rm -it -- bash

8. After a bash session has started, connect to the primary MongoDB instance using:

   mongosh mongodb://mongo-0.mongo-hlsvc

9. Initiate the replicaset using the following mongosh command:

   rs.initiate({
       _id: "rs0",
       members: [
           { _id: 0, host: "mongo-0.mongo-hlsvc" },
           { _id: 1, host: "mongo-1.mongo-hlsvc" },
           { _id: 2, host: "mongo-2.mongo-hlsvc" },
       ],
   });

10. The replicaset will take a little while to initiate. The status of the replicaset can be checked using the following mongosh command:

    rs.status();

    Once you see one primary server and 2 secondary servers, you are ready to go.

11. Now, just wait for the API server to restart automatically and every thing should work.

12. To make requests to the API, you'll first need to get the IP of minikube using:

    minikube ip

13. Next you'll need the port on which the API server is running. To get that run:

    kubectl get service/api-server

    The port that you need will be mapped under the "PORT(S)" column.

14. Now you can make requests to the service on the IP $(minikube ip):<PORT>.

Note: Once you delete your local minikube cluster, you'll have to repeat steps 7 through 11 to setup the replicaset.

Tearing down

To stop the all the deployed units in the cluster, run:

kubectl delete -k k8s/overlays/<overlay>

Here, replace <overlay> with the overlay used to deploy units.

To delete all resources started by minikube, run:

minikube delete