Scrapping GO- custom metrics using Prometheus and Grafana,,installed within GKE cluster i.e Helm for k8s package manager
Step 1: Create A GKE/ACR/ECR repository to save your docker image Step 2: Create A GKE/EKS/AKS to deploy your app in a managed kubernetes cluster. 2(a): Step to Create a GKE Cluster # gcloud init ( To initialize a gcloud project with automatic authentication) # gcloud container clusters create (name of your cluster) --num-nodes=(required num of nodes in each az) --enable-autoscaling --min-nodes (num) --max-nodes (num) # gcloud container clusters get-credentials (cluster name) (To configure kube config) 2(b) Install Helm (If you haven't) 2(c) Steps to Configure Prometheus and Grafana using Helm # helm repo add prometheus-community https://prometheus-community.github.io/helm-charts (Define public Kubernetes chart repository in the Helm configuration) # helm repo update (Update local repositories) # kubectl create ns prometheus (Create a namespace for Prometheus and Grafana resources) # helm install prometheus prometheus-community/kube-prometheus-stack -n prometheus (Install Prometheus using HELM) # kubectl get all -n prometheus (Check all resources in Prometheus Namespace) # kubectl edit svc prometheus-kube-prometheus-prometheus -n prometheus ( Here we are changing from ClusterIP to LoadBalancer/NodePort) # kubectl edit svc prometheus-grafana - n prometheus ( Here we are changing from ClusterIP to LoadBalancer/NodePort) 2(d) Create a Service Monitor to Scrape your Custom metrics in Prometheus (check template-> deployment.yaml(bottom) 2(e) Install your templates using helm install [][] 2(f) Create a custom_values.yaml to override the ScrapeConfig file of prometheus (Check custom_values.yaml for reference) 2(g) helm upgrade --namespace prometheus -f custom_values.yaml prometheus prometheus-community/kube-prometheus-stack (Run the above command to upgrade helm chart and add custom Scrap Endpoint in prometheus) # kubectl edit svc prometheus-kube-prometheus-prometheus -n prometheus ( Here we are changing from ClusterIP to LoadBalancer/NodePort) # kubectl edit svc prometheus-grafana - n prometheus ( Here we are changing from ClusterIP to LoadBalancer/NodePort) (Since the helm chart is upgraded the Loadbalancer will be reconfigured to default value(ClusterIp)
Points:-