Transient upstream retries with gRPC

[Michal Sabo]

Hi,

I would appreciate any help to make retry logic for upstream gRPC services working.

Envoy gRPC retry policy is based on grpc-status header, however grpc-status is propagated via trailers and Envoy's retry logic seems not to recognize that.

More info can be found in this open issue - https://github.com/envoyproxy/envoy/issues/11049

Nevertheless, is here anybody who have working configuration with gRPC retry policy defined? Any help would be appreciated.

Thank you,

Michal Sabo