diff --git a/Socket.cpp b/Socket.cpp
index 5a89c1b..2167ed6 100644
--- a/Socket.cpp
+++ b/Socket.cpp
@@ -272,6 +272,26 @@ namespace core
 	       // the confirmed real damage (a 55GB, then 12.9GB debug log)
 	       // this caused before being noticed.
 	       coreutils::Log(coreutils::LOG_EXCEPT) << "Socket " << getDescriptor() << ": " << consecutiveWriteStalls << " consecutive writes made zero progress - giving up and shutting down.";
+	       // shutdown() alone only sets shutDown=true and expects the
+	       // fifo to drain normally before the connection actually
+	       // closes (see eventReceived()'s keepAlive: "!shutDown ||
+	       // (fifo.size() > 0 && !readHangup)") - exactly backwards
+	       // for THIS path, since the whole reason we're here is that
+	       // the front of the fifo can never be sent. Leaving it
+	       // queued meant keepAlive stayed true forever, resetSocket()
+	       // kept re-arming EPOLLWRNORM because fifo.size() > 0, the
+	       // very next write attempt hit this identical stall
+	       // immediately, and shutdown() got called again - forever,
+	       // at whatever rate the kernel could fire EPOLLWRNORM.
+	       // Confirmed live: 2.4 million+ repeats of this exact log
+	       // line for one socket in under 4 milliseconds, pegging a
+	       // worker thread at 100% CPU indefinitely (a real game
+	       // server, brought down by nothing more than an ordinary
+	       // port-scanner connection). Clearing the undeliverable data
+	       // here lets keepAlive correctly evaluate to just !shutDown
+	       // (now false) on the very next cycle, so the connection
+	       // actually tears down instead of spinning forever.
+	       std::queue<std::string>().swap(fifo);
 	       shutdown("stuck write retry loop - no progress after repeated attempts");
 	    }
          }