Reputation: 155
I found something wrong at HPA for istio gateway.
Why did 10m equal 10%? Wasn't 10m 1%?
Kubernetes version is 1.18.5.
# kubectl get hpa --all-namespaces
NAMESPACE NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
istio-system istio-egressgateway Deployment/istio-egressgateway 7%/80% 2 10 2 13d
istio-system istio-ingressgateway Deployment/istio-ingressgateway 10%/80% 2 10 2 21d
istio-system istiod Deployment/istiod 0%/80% 1 5 1 21d
qa2 graph Deployment/graph 2%/50% 1 10 1 7h35m
qa2 member Deployment/member 0%/50% 1 10 1 7h38m
# kubectl describe hpa istio-ingressgateway -n istio-system | grep "resource cpu"
resource cpu on pods (as a percentage of request): 10% (10m) / 80%
# kubectl describe hpa istio-egressgateway -n istio-system | grep "resource cpu"
resource cpu on pods (as a percentage of request): 7% (7m) / 80%
# kubectl describe hpa istiod -n istio-system | grep "resource cpu"
resource cpu on pods (as a percentage of request): 0% (3m) / 80%
# kubectl describe hpa graph -n qa2 | grep "resource cpu"
resource cpu on pods (as a percentage of request): 2% (24m) / 50%
# kubectl describe hpa member -n qa2 | grep "resource cpu"
resource cpu on pods (as a percentage of request): 1% (12m) / 50%
Upvotes: 2
Views: 586
Reputation: 7810
These values are not the same, and they are not directly calculated from each other.
The value in percents is the target average utilization (corresponding to the targetAverageUtilization
parameter), which is relative to the requested value.
The value in the brackets is the target average value (targetAverageValue
), which is not measured in percents - this is an absolute raw value for the metric.
Upvotes: 3