Skip to content

Agent Mesh

address review: propagate client write errors immediately instead of waiting out the drain timeout

Fernando Marinoa4821df9d2d1
diff --git a/internal/node/mcp_service.go b/internal/node/mcp_service.go
index 2939bd34..1575ff98 100644
--- a/internal/node/mcp_service.go
+++ b/internal/node/mcp_service.go
@@ -203,7 +203,13 @@ func (m *MCPService) HandleStreamPassThrough(s network.Stream) {
// the final s.Close() at the top of this function. passThroughDrainTimeout
// bounds that wait so a client that never hangs up can't leak the stream
// forever.
- clientErrc := make(chan error, 1)
+ //
+ // clientErrc is buffered for 2, not 1: a client write failure below is
+ // also a client-leg error (the stream to the client is dead, so there is
+ // nothing left to drain for), and with both goroutines able to send, an
+ // unlucky interleaving where the main select has already consumed one
+ // value could otherwise leave the second sender blocked forever.
+ clientErrc := make(chan error, 2)
go func() {
for {
@@ -217,6 +223,7 @@ func (m *MCPService) HandleStreamPassThrough(s network.Stream) {
}
if err := clientConn.Write(ctx, msg); err != nil {
logger.Debugf("[MCPService] %s: client write error: %v", m.info.Name, err)
+ clientErrc <- err
return
}
}