Istio Versions:
AuthService version: 0.5.2
Hello!
I'm trialling AuthService as a way of delivering auth redirection for n+1 single tenant services that require different clients. I followed the guide on setting up the GRPC endpoint, but when enrolling one service that makes 7 calls to the pod (Static UI assets, so page plus all the js, css, svgs etc) and about 50% of those will fail. This is with one pod (CPU: 100m, Mem: 200mb) and none of the resource utilisation changes. We scaled up the service to 20 pods and the failure rate dropped to ~5%. Is this expected when using GRPC? We believe (Not helped by poor istio gateway upstream error logging) that the GRPC backpressure is causing the gateway to immediately give up and return a 403. We see those 403s come back faster than the successful hits on the authservice.
A second question, are there any docs or instructions for utilising the HTTP endpoint? I would like to try this before I give up.
Istio Versions:
AuthService version: 0.5.2
Hello!
I'm trialling AuthService as a way of delivering auth redirection for n+1 single tenant services that require different clients. I followed the guide on setting up the GRPC endpoint, but when enrolling one service that makes 7 calls to the pod (Static UI assets, so page plus all the js, css, svgs etc) and about 50% of those will fail. This is with one pod (CPU: 100m, Mem: 200mb) and none of the resource utilisation changes. We scaled up the service to 20 pods and the failure rate dropped to ~5%. Is this expected when using GRPC? We believe (Not helped by poor istio gateway upstream error logging) that the GRPC backpressure is causing the gateway to immediately give up and return a 403. We see those 403s come back faster than the successful hits on the authservice.
A second question, are there any docs or instructions for utilising the HTTP endpoint? I would like to try this before I give up.