Prometheus-Learnings

Introduction

This repo is built for running an demo instance of prometheus and grafana over an local kind cluster or over Google Cloud Provider.

Pre-Requisites

For running locally

• Install kind cluster on your machine. Installation Guide can be found here
• Install kubectl command line

For running on GCP

• Terraform CLI must be installed.Installation Guide can be found here
• An Google Account with appropriate privilages to create GKE Clusters.
• Gcloud SDK installed on your machine. Installation Guide can be found here

Running Locally

• Clone this repo in your working directory

git clone https://github.com/deepankur797/prometheus-learnings.git

• Create a kind cluster

kind create cluster --name "your_cluster_name"

• Switch context to your cluster using the following

kubectl cluster-info --context kind-your_cluster_name

• Create a namespace "monitoring"

kubectl create ns monitoring

• Switch to prometheus-learning/kubeFiles directory
• Apply all files

kubectl apply -f clusterRole.yaml
kubectl apply -f config-map.yaml
kubectl apply -f prometheus-deployment.yaml
kubectl apply -f prometheus-service.yaml

This will create the prometheus deployment. You can access it through kubectl port-forward or at http://{ip-of-kind-cluster-node}:30001

• For installation of grafana, switch to prometheus-learning/grafanaFiles directory and apply the following commands

kubectl apply -f grafana-datasource-config.yaml
kubectl apply -f deployment.yaml
kubectl apply -f service.yaml

You can access this also using kubectl port-forward or at http://{ip-of-kind-cluster-node}:32001

Running the setup over GCP

• First authenticate terraform using gcloud

gcloud init
gcloud auth application-default login

• After the successful login, switch to the directory prometheus-learning/terraform2/

terraform init
terraform plan
terraform apply

• The following variables will be asked

1. project_id = "your_project_id"
2. region = "mention the zone name or region name"  # use region name to create a regional cluster and zone name to create a zonal cluster 
3. gke_username = "" # creds will be applied to virtual machines
4. gke_password = "" # Length of 16, aplha-numeric required
5. gke_num_nodes = "number of nodes" # make it one 

• After successful completion of terraform plan you can access the services at

1. Prometheus : http://{node-public-ip}:30001
2. Grafana    : http://{node-public-ip}:32001

• As this is a demo cluster hence no load-balancing is used and the services are exposed using node-port type of kubernetes cluster.
• An simple flask appliction is also exposed in the GCP setup with some custom metrics exposed. And a custom dashboard to monitor these custom metrics is also being imported by default by the name of node-info in grafana.

Custom app can be visited at
http://{node-public-ip}:32000

La Fin