recover monitoring fd in each sandbox in the chain, otherwise, sledge will stuck at sometimes, the reason needs time to dive in later

runtime/include/sandbox_functions.h

@@ -145,7 +145,6 @@ sandbox_open_http(struct sandbox *sandbox)
 	/* Set the sandbox as the data the http-parser has access to */
 	sandbox->http_parser.data = sandbox; //assign data to sandbox in case to operator it when a callback happended
 
-	if (sandbox->module->next_module == NULL) {
 	/* Freshly allocated sandbox going runnable for first time, so register client socket with epoll */
 	struct epoll_event accept_evt;
 	accept_evt.data.ptr = (void *)sandbox;
@@ -154,7 +153,6 @@ sandbox_open_http(struct sandbox *sandbox)
 	                   &accept_evt);
 	if (unlikely(rc < 0)) panic_err();
 }
-}
 
 /**
  * Prints key performance metrics for a sandbox to runtime_sandbox_perf_log

runtime/src/current_sandbox.c

@@ -116,6 +116,7 @@ current_sandbox_start(void)
                                                            sandbox->request_arrival_timestamp, true, pre_func_output, output_length);
                 /* Add to the Global Sandbox Request Scheduler */
                 global_request_scheduler_add(sandbox_request);
+		sandbox_remove_from_epoll(sandbox);
 		sandbox_set_as_returned(sandbox, SANDBOX_RUNNING);
 	} else {
 		/* Retrieve the result, construct the HTTP response, and send to client */