scwhittle
scwhittle
From discussion on beam-dev it sounds like there is agreement that adding a ErrorReason to the state request response instead of just error string would be useful for improving logging...
I still plan on doing this once I get a chance
Sorry! I was waiting for tests to complete originally but I lost track of it
It was closed automatically since PR was annotated with fixing it. But reopening for verification.
Thanks for your response. There is still something that doesn't make sense to me. When `keepAliveWithoutCalls=true`, the KeepAliveManager is started earlier, during [onTransportStart](https://github.com/grpc/grpc-java/blob/226b1e5ecc7ddcbb819cfc893308ac46973d16d5/core/src/main/java/io/grpc/internal/KeepAliveManager.java#L152). AFAICT this may also be before TCP...
Thanks! So it sounds like there are some places that are perhaps missing timeouts that could lead to this stuck behavior. If keep alive without calls isn't the proper way...
In this instance the connection was stuck for 20 minutes following a high cpu period when it was created, not 30 seconds. This is the original observed instance without much...