Save the state of scheduled node reboots so they will not be

lost should the slurmctld restart.

NEWS

@@ -131,6 +131,8 @@ documents those changes that are of interest to users and admins.
  -- When creating reservation with CoreCnt check that Slurm
     uses SelectType=select/cons_res otherwise don't send the
     request to slurmctld and return an error.
+ -- Save the state of scheduled node reboots so they will not be
+    lost should the slurmctld restart.
 
 * Changes in Slurm 14.03.6
 ==========================

src/slurmctld/controller.c

@@ -1300,6 +1300,10 @@ static void _queue_reboot_msg(void)
 			want_nodes_reboot = true;
 			continue;
 		}
+		if (IS_NODE_COMPLETING(node_ptr)) {
+			want_nodes_reboot = true;
+			continue;
+		}
 		if (IS_NODE_IDLE(node_ptr) && !IS_NODE_NO_RESPOND(node_ptr) &&
 		    !IS_NODE_POWER_UP(node_ptr)) /* only active idle nodes */
 			want_reboot = true;
@@ -1322,6 +1326,7 @@ static void _queue_reboot_msg(void)
 		reboot_agent_args->node_count++;
 		node_ptr->node_state = NODE_STATE_FUTURE |
 				(node_ptr->node_state & NODE_STATE_FLAGS);
+		node_ptr->node_state &= ~NODE_STATE_MAINT;
 		bit_clear(avail_node_bitmap, i);
 		bit_clear(idle_node_bitmap, i);
 		node_ptr->last_response = now;
@@ -1334,6 +1339,7 @@ static void _queue_reboot_msg(void)
 		xfree(host_str);
 		agent_queue_request(reboot_agent_args);
 		last_node_update = now;
+		schedule_node_save();
 	}
 }
 

src/slurmctld/node_mgr.c

@@ -529,6 +529,8 @@ extern int load_all_node_state ( bool state_only )
 						mem_spec_limit;
 					node_ptr->tmp_disk      = tmp_disk;
 				}
+				if (node_state & NODE_STATE_MAINT)
+					node_ptr->node_state |= NODE_STATE_MAINT;
 				if (node_state & NODE_STATE_POWER_UP) {
 					if (power_save_mode) {
 						node_ptr->node_state |=

src/slurmctld/proc_req.c

@@ -4674,6 +4674,10 @@ inline static void _slurm_rpc_reboot_nodes(slurm_msg_t * msg)
 		node_ptr->node_state |= NODE_STATE_MAINT;
 		want_nodes_reboot = true;
 	}
+
+	if (want_nodes_reboot == true)
+		schedule_node_save();
+
 	unlock_slurmctld(node_write_lock);
 	FREE_NULL_BITMAP(bitmap);
 	rc = SLURM_SUCCESS;