Kubernetes Practice

Building the project

1. Create namespace kubectl create -f namespaces.yml. This will create dev and logging namespaces. dev is where we will operate our application. logging is where we will operate our logging infrastructure.
2. Run kubectl create -f . --recusive to create the rest of the infrastructure.

PRODUCT MICROSERVICE ./products

Our first microservice in kubernetes is Product ./products/product.yml. It is implemented using Spring Boot. It displays the laptop products in the JSON format.

To run this, we would need to complete its dependencies, such as the database which it uses to store product information. MongoDB is the No-SQL solution for this. It is deployed by mongo.yml

Access Product Service

First we will need to expose the product service to the node. Since it is exposed by the NodePort, we would need to find the port that this service will expose to.

1. Use minikube ip to find the ip address of the minikube. This will be ip address of our port.
2. Use kubectl -n dev get svc product -o wide to find the external port. This is in range 3000.
3. Use curl or browser to access the service.

LOGGING

This is a logging part of the kubernetes infrastructure. We design the microservice based infrastructure to scale as our application grows. And, with the growth of the application, we would have a system where the logging is scattered all over the microservices. This requires centralized logging.

This stack uses ElasticSearch as database, Kibana as visualization tool and fluend as the logging agent deployed in a cluster.

ElasticSearch is deployed as StatefulSet to leverage its stateful cluster. The order is maintained.

Kibana is a stateless Deployment that can scale as it likes.

Fluentd is a stateless DaemonSet that is deployed in each node, pulling out all the logs it can.

Accessing it locally:

1. Figure out the kibana pod first kubectl -n logging get pods | grep kibana
2. Do a port forward: kubectl port-forward ${KIBANA_POD} 5601:5601
3. Open a browser and see your logs.