See #112. One of the potential mitigations in reading around that issue is setting grpc.enabled_retries = 0 in the client setup. This means that things are more likely to fail for retryable reasons, but the retries themselves are less likely to fail in an unrecoverable fashion, at least until the underlying problems (which appear to be mostly related to envoy/istio) are resolved.
See #112. One of the potential mitigations in reading around that issue is setting
grpc.enabled_retries = 0in the client setup. This means that things are more likely to fail for retryable reasons, but the retries themselves are less likely to fail in an unrecoverable fashion, at least until the underlying problems (which appear to be mostly related to envoy/istio) are resolved.