Hello, I have a python service and a c++ client. The client produces data and is very fast. The service is writing data to some external db and is relatively slow. I am using a client-to-server stream method (called SendData). So it is easy to overwhelm the service.
The question is is there any protection that prevents the service to go out of memory? I did a small experiment and it seems that if the client is flooding the service (at least if it uses the synchronous c++ API), gRPC eventually blocks the client (in ClientWriter::Write) and lets the server catch up. After the server does some progress, the client is unblocked again and continues. I am happy, that something like this is in place, but I have no idea where it comes from. Does anyone know any details about how this works? Is it documented? On what layer does it work? Is it configurable? I read about retries and server pushback, but it seems with streaming they are not usable, so it is probably not it. Vojtech -- You received this message because you are subscribed to the Google Groups "grpc.io" group. To unsubscribe from this group and stop receiving emails from it, send an email to grpc-io+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/grpc-io/435ea361-4045-4f20-b87a-421dd0cc9884n%40googlegroups.com.
