Node does not transition from COMPLETING to DOWN state due to node not

responding. Wait for tasks to complete or administrator to set DOWN.

NEWS

@@ -9,6 +9,8 @@ documents those changes that are of interest to users and admins.
  -- Slurmctld does not attempt to recover jobs when the switch plugin 
     value changes (decision reached when any job's switch state recovery 
     fails)
+ -- Node does not transition from COMPLETING to DOWN state due to 
+    not responding. Wait for tasks to complete or admin to set DOWN
 
 * Changes in SLURM 0.3.0.0-pre6
 ===============================

src/slurmctld/agent.c

@@ -370,7 +370,7 @@ static void *_wdog(void *args)
 					       (unsigned long) thread_ptr[i].thread);
 					if (pthread_kill(thread_ptr[i].thread,
 						     SIGALRM) == ESRCH)
-						thread_ptr[i].state = DSH_FAILED;
+						thread_ptr[i].state = DSH_NO_RESP;
 				}
 				break;
 			case DSH_NEW:

src/slurmctld/ping_nodes.c

@@ -166,6 +166,7 @@ void ping_nodes (void)
 		if ((node_ptr->last_response != (time_t)0) &&
 		    (node_ptr->last_response <= node_dead_time) &&
 		    ((base_state != NODE_STATE_DOWN) &&
+		     (base_state != NODE_STATE_COMPLETING) &&
 		     (base_state != NODE_STATE_DRAINED))) {
 			error ("Node %s not responding, setting DOWN", 
 			       node_ptr->name);