Scheduling fix for jobs with specific nodes required

If a job requires specific nodes and can not run due to those nodes being
busy, the main scheduling loop will block those specific nodes rather than
the entire queue/partition.
bug 595

NEWS

@@ -11,8 +11,11 @@ documents those changes that are of interest to users and admins.
  -- Fix issue where if using munge and munge wasn't running and a slurmd
     needed to forward a message, the slurmd would core dump.
  -- Update srun.1 man page documenting the PMI2 support.
- -- Fix slurmctld core dump when a jobs gets its qos updated but there
+ -- Fix slurmctld core dump when a jobs gets its QOS updated but there
     is not a corresponding association.
+ -- If a job requires specific nodes and can not run due to those nodes being
+    busy, the main scheduling loop will block those specific nodes rather than
+    the entire queue/partition.
 
 * Changes in Slurm 2.6.6
 ========================

src/slurmctld/job_scheduler.c

@@ -1008,6 +1008,21 @@ next_part:			part_ptr = (struct part_record *)
 			 * scheduler is configured. */
 			if (!backfill_sched)
 				fail_by_part = false;
+#else
+			if (job_ptr->details &&
+			    job_ptr->details->req_node_bitmap &&
+			    (bit_set_count(job_ptr->details->
+					   req_node_bitmap)>=
+			     job_ptr->details->min_nodes)) {
+				fail_by_part = false;
+				/* Do not schedule more jobs on nodes required
+				 * by this job, but don't block the entirecd srcccccc
+				 * queue/partition. */
+				bit_not(job_ptr->details->req_node_bitmap);
+				bit_and(avail_node_bitmap,
+					job_ptr->details->req_node_bitmap);
+				bit_not(job_ptr->details->req_node_bitmap);
+			}
 #endif
 			if (fail_by_part) {
 		 		/* do not schedule more jobs in this partition