Cloud Native approach have solved many single units of complexities, but it have created tools sprawl within for teams. This demo show how to implement single pane of glass for observability tools based on Kuma envoy based service mesh (https://kuma.io/)
Components
| Component | Purpose |
|---|---|
| Kuma | Envoy based service mesh |
| Loki | Centralized logging service |
| Grafana | Dashboards |
| Tempo | Traces storage solution |
All is being deployed on GKE for sake of simplicity
1. Install GKE Cluster
gcloud container clusters create sample-cluster \
--release-channel stable \
--zone europe-west4-a \
--node-locations europe-west4-a
Don't forget to change Kubernetes context before moving further
- Add helm repos
helm repo add grafana https://grafana.github.io/helm-charts
- Setup tracing namespace
kubectl apply -f yamls/0-tracing.yaml
- Install Grafana
helm install grafana grafana/grafana -n tracing --version 6.13.5 -f yamls/5-grafana.yaml
- Install Tempo
helm install tempo grafana/tempo --version 0.7.4 -n tracing -f yamls/1-tempo.yaml
- Install Open Telemetry Collector
kubectl apply -f yamls/3-otel.yaml
- Get binary
curl -L https://kuma.io/installer.sh | sh -- Install Kuma
./kumactl install control-plane | kubectl apply -f -
- Enable logging
kumactl install logging | kubectl apply -f -
- Enable collectors
kubectl apply -f yamls/6-kuma-collector.yaml
- Install Kong
kubectl apply -f https://bit.ly/k4k8s
kubectl annotate ns kong kuma.io/sidecar-injection=enabled
kubectl delete pod --all -n kong
- Install book info sample app
kubectl apply -f yamls/7-ns.yaml
kubectl apply -f yamls/8-bookinfo.yaml
#Access To Kuma control plane
kubectl port-forward svc/kuma-control-plane -n kuma-system 5681:5681
To Grafana
kubectl port-forward svc/grafana -n tracing 8081:80
To Bookinfo
- Find ingress and access public IP
Searching in loki:
{namespace="bookinfo"} |="TraceId"
Screens:
Overview from loggin side
From Tempo side
