Log build job queue timeout less ofter

Rather than generating loads of log messages about too much time
being used to build the job queue every few seconds, log it only
every 10 minutes.
bug 1827

src/slurmctld/job_scheduler.c

@@ -294,6 +294,7 @@ static int _delta_tv(struct timeval *tv)
  */
 extern List build_job_queue(bool clear_start, bool backfill)
 {
+	static time_t last_log_time = 0;
 	List job_queue;
 	ListIterator job_iterator, part_iterator;
 	struct job_record *job_ptr = NULL;
@@ -308,10 +309,15 @@ extern List build_job_queue(bool clear_start, bool backfill)
 	while ((job_ptr = (struct job_record *) list_next(job_iterator))) {
 		if (((tested_jobs % 100) == 0) &&
 		    (_delta_tv(&start_tv) >= build_queue_timeout)) {
-			info("build_job_queue has been running for %d usec, "
-			     "exiting with %d of %d jobs tested, %d job-partition pairs added",
-			     build_queue_timeout, tested_jobs,
-			     list_count(job_list), job_part_pairs);
+			time_t now = time(NULL);
+			if (difftime(now, last_log_time) > 600) {
+				/* Log at most once every 10 minutes */
+				info(
+"%s has run for %d usec, exiting with %d of %d jobs tested, %d job-partition pairs added",
+				     __func__, build_queue_timeout, tested_jobs,
+				     list_count(job_list), job_part_pairs);
+				last_log_time = now;
+			}
 			break;
 		}
 		tested_jobs++;