Bug Report: Dedicated Server Fatal Crash on gRPC Stream Disconnect
Version: 0.10.0.3.104-256f9653 | OS: Linux via Wine 11.0 (AMP/CubeCoders)
Summary
The dedicated server calls appError and crashes whenever the gRPC stream to the Windrose backend is interrupted. It does not attempt to reconnect. This causes a crash-restart loop that can render the host machine completely unresponsive.
Error (from server log)
Code:
[R5LogNetCm] Error: Disconnect with error. error_code UNKNOWN.
error_message 'Stream removed'.
[LogWindows] Fatal error:
[R5NetCmClient_GsStream.cpp Line 203] SetBrokenState
GsStream is broken. 'Cm disconnected. Status 2 (Stream removed).
CallNumber 1'. PrevState Registration
Impact
Any gRPC stream interruption (network hiccup, backend timeout) causes an immediate full server crash
Server manager restarts it instantly, creating a rapid crash-loop
Each restart cycle on Linux leaves zombie processes that accumulate and eventually make the host unresponsive to SSH, requiring a physical reset
We observed 7 crashes in a single day, all identical, causing two full host machine outages
Expected Behavior
Treat dropped gRPC streams as transient and reconnect with backoff — not as a fatal unrecoverable error.
Launch Flags Used
-nullrhi -nosound -stdout -FullStdOutLogOutput