prometheus-metrics-sample-app

A sample prometheus metrics python application

Pre-requisites

1. Docker
2. Kind
3. Kubectl
4. Helm
5. Kustomize

Note: If you are using Docker Desktop on Mac or Windows, please make sure the Docker VM is configured to use a minimum of 4 CPUs and 6GB Memory. (This can be modified under the setting/preferences of Docker Desktop)

Steps

1. Create a kubernetes cluster.

kind create cluster -f kind/config.yaml

2. Install the prometheus-operator using the kube-prometheus-stack helm chart.

helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
helm update

helm install kube-prometheus-stack prometheus-community/kube-prometheus-stack --version 31.0.0 -n monitoring

This will install the prometheus-operator which inturn will create prometheus statefulset, grafana deployment and alertmanager statefulset and a few out of the box exporters in the monitoring namespace

prometheus-metrics-sample-app % kubectl get pods -n monitoring                                  
NAME                                                        READY   STATUS    RESTARTS   AGE
alertmanager-kube-prometheus-stack-alertmanager-0           2/2     Running   0          75s
kube-prometheus-stack-grafana-67f7cb9468-srjbs              3/3     Running   0          79s
kube-prometheus-stack-kube-state-metrics-7f56994f6c-9lp95   1/1     Running   0          79s
kube-prometheus-stack-operator-868bb97b5c-bntw5             1/1     Running   0          79s
kube-prometheus-stack-prometheus-node-exporter-7sgfd        1/1     Running   0          79s
kube-prometheus-stack-prometheus-node-exporter-bg8rg        1/1     Running   0          79s
kube-prometheus-stack-prometheus-node-exporter-h67xq        1/1     Running   0          79s
kube-prometheus-stack-prometheus-node-exporter-nd6jt        1/1     Running   0          79s
prometheus-kube-prometheus-stack-prometheus-0               2/2     Running   0          75s

3. Build the sample python metrics app

cd pythonapp

docker build -t metrics:v0.1 -f dockerfile-distroless .

4. Load the image into the Kubernetes nodes.

kind load docker-image metrics:v0.1 --name monitoring

5. Deploy the python application.

cd ../manifests

kubectl build . | kubectl create -f -

6. Monitor the metrics using prometheus.

kubectl port-forward svc/prometheus-operated -n monitoring 9090

Access localhost:9090 using a browser.

Find the service monitor as a prometheus target.