TL; DR: I made a shell script which prints out the CPU and memory usage vs allocation for each pod in a Kubernetes cluster. Then it’s easy to tell where to optimise. Here it is:
# list all pods and read namespace and pod name as input
kubectl get pods -A --no-headers|while read ns pod a; do
# gather cpu and memory real time usage using kubectl top command
# gather cpu and memory allocations using kubectl get pod with jsonpath selected
# join the 2 line by line by pod name using paste command
paste <(kubectl top -n $ns pod $pod --containers --no-headers|sort) <(kubectl get pod -n $ns $pod -o jsonpath='{range .spec.containers[*]}{.name}{"\t"}{.resources.requests.cpu}{"\t"}{.resources.requests.memory}{"\n"}{end}'|sort)
echo ---
done
# sample output
# pod name container. cpu, memory usage cpu, memory allocation
my-service-8d9fb4b98-nbm7s my-service 2m 177Mi my-service 100m 256Mi
my-service-8d9fb4b98-nbm7s istio-proxy 10m 57Mi istio-proxy 50m 64MiJust in case, you need the metrics-server for the kubectl top command to function.
🙂