Ouch - opal_progress calls event_loop with a NO_BLOCK flag. So when run without progress threads, the ORTE tools were not blocking in the event lib as they should be. Avoid calling opal_progress inside ORTE by directly using the event_loop call instead of ORTE_WAIT_FOR_COMPLETION as parts of the OMPI layer are using that macro.

Thanks to George for spotting the problem.

This commit was SVN r27815.

orte/orted/orted_main.c

@@ -821,7 +821,19 @@ int orte_daemon(int argc, char *argv[])
     }
 
     /* loop the event lib until an exit event is detected */
-    ORTE_WAIT_FOR_COMPLETION(orte_event_base_active);
+#if ORTE_ENABLE_PROGRESS_THREADS
+    while (orte_event_base_active) {
+        /* provide a very short quiet period so we
+         * don't hammer the cpu while
+         */
+        struct timespec tp = {0, 100};
+        nanosleep(&tp, NULL);
+    }
+#else
+    while (orte_event_base_active) {
+        opal_event_loop(orte_event_base, OPAL_EVLOOP_ONCE);
+    }
+#endif
 
     /* ensure all local procs are dead */
     orte_odls.kill_local_procs(NULL);

orte/tools/orterun/orterun.c

@@ -1020,7 +1020,19 @@ int orterun(int argc, char *argv[])
     rc = orte_plm.spawn(jdata);
 
     /* loop the event lib until an exit event is detected */
-    ORTE_WAIT_FOR_COMPLETION(orte_event_base_active);
+#if ORTE_ENABLE_PROGRESS_THREADS
+    while (orte_event_base_active) {
+        /* provide a very short quiet period so we
+         * don't hammer the cpu while
+         */
+        struct timespec tp = {0, 100};
+        nanosleep(&tp, NULL);
+    }
+#else
+    while (orte_event_base_active) {
+        opal_event_loop(orte_event_base, OPAL_EVLOOP_ONCE);
+    }
+#endif
 
     /* ensure all local procs are dead */
     orte_odls.kill_local_procs(NULL);