fix: drain the connection socket on a protocol error (#1422)

Signed-off-by: Andy Dunstall <andydunstall@hotmail.co.uk>

src/facade/dragonfly_connection.cc

@@ -505,8 +505,24 @@ void Connection::ConnectionFlow(FiberSocketBase* peer) {
       DCHECK(memcache_parser_);
       orig_builder->SendProtocolError("bad command line format");
     }
-    error_code ec2 = peer->Shutdown(SHUT_RDWR);
+
+    // Shut down the servers side of the socket to send a FIN to the client
+    // then keep draining the socket (discarding any received data) until
+    // the client closes the connection.
+    //
+    // Otherwise the clients write could fail (or block), so they would never
+    // read the above protocol error (see issue #1327).
+    error_code ec2 = peer->Shutdown(SHUT_WR);
     LOG_IF(WARNING, ec2) << "Could not shutdown socket " << ec2;
+    if (!ec2) {
+      while (true) {
+        // Discard any received data.
+        io_buf_.Clear();
+        if (!peer->Recv(io_buf_.AppendBuffer())) {
+          break;
+        }
+      }
+    }
 
     FetchBuilderStats(stats_, orig_builder);
   }

src/facade/redis_parser.cc

@@ -252,7 +252,7 @@ auto RedisParser::ConsumeArrayLen(Buffer str) -> Result {
       return BAD_ARRAYLEN;
     case OK:
       if (len < -1 || len > kMaxArrayLen) {
-        VLOG_IF(1, len > kMaxArrayLen) << "Multi bulk len is too big " << len;
+        LOG_IF(WARNING, len > kMaxArrayLen) << "Multi bulk len is too big " << len;
 
         return BAD_ARRAYLEN;
       }