hi, i am testing a performance of my service(myapp), and a little bit confused with the HPA result
"
kubectl get hpa | grep myapp
myapp Deployment/myapp 81%/80% 2 2 2 7d23h
kubectl top pod myapp-85c77fbcd7-7ppxr --containers
myapp-85c77fbcd7-7ppxr myapp 1775m 1091Mi
myapp-85c77fbcd7-7ppxr leader-elector 1m 18Mi
myapp-85c77fbcd7-7ppxr tap-agent 3m 7Mi
myapp-85c77fbcd7-7ppxr istio-proxy 809m 114Mi
cpu request for containers as below:
myapp 3G
leader-elector 50m
tap-agent 100m
istio-proxy 2G (automatically injected, not static defined in deployment yaml)
noticed that the istio-proxy is auto injected, so the cpu request is not defined in myapp deployment yaml
"
according to the above data, the HPA 81% almost equals to (1775m+1m+3m+809m) / (3G+50m+100m), in other words, the istio-proxy’s cpu request is not calculated in HPA
i don’t know why it happens, it’s a bug of k8s API ？