A couple verbose additions. Poll the event engine while waiting for the

named pipe.

This commit was SVN r17787.

opal/runtime/opal_cr.c

@@ -330,6 +330,9 @@ int opal_cr_init(void )
 
 #if OPAL_ENABLE_FT_THREAD == 1
     if( !opal_cr_is_tool && opal_cr_thread_use_if_avail) {
+        opal_output_verbose(10, opal_cr_output,
+                            "opal_cr: init: starting the thread\n");
+
         opal_set_using_threads(true);
         /*
          * Start the thread
@@ -347,6 +350,10 @@ int opal_cr_init(void )
         opal_thread_start(&opal_cr_thread);
 
     } /* End opal_cr_is_tool = true */
+    else {
+        opal_output_verbose(10, opal_cr_output,
+                            "opal_cr: init: *Not* Using C/R thread\n");
+    }
 #endif /* OPAL_ENABLE_FT_THREAD == 1 */
 
  cleanup:

orte/mca/snapc/full/snapc_full_app.c

@@ -94,19 +94,22 @@ int app_coord_init() {
     asprintf(&app_comm_pipe_r, "%s/%s.%s", opal_cr_pipe_dir, OPAL_CR_NAMED_PROG_R, tmp_pid);
     asprintf(&app_comm_pipe_w, "%s/%s.%s", opal_cr_pipe_dir, OPAL_CR_NAMED_PROG_W, tmp_pid);
 
-    OPAL_OUTPUT_VERBOSE((15, mca_snapc_full_component.super.output_handle,
-                         "app) Named Pipes (%s) (%s)", 
-                         app_comm_pipe_r, app_comm_pipe_w));
-
     /*
      * Setup a signal handler to catch and start the proper thread
      * to handle the checkpoint
      */
     if( SIG_ERR == signal(opal_cr_entry_point_signal, snapc_full_app_signal_handler) ) {
+        opal_output(mca_snapc_full_component.super.output_handle,
+                    "App) init: Error: Failed to register signal %d\n",
+                    opal_cr_entry_point_signal);
         exit_status = OPAL_ERROR;
         goto cleanup;
     }
 
+    OPAL_OUTPUT_VERBOSE((15, mca_snapc_full_component.super.output_handle,
+                         "app) Named Pipes (%s) (%s), Signal (%d)", 
+                         app_comm_pipe_r, app_comm_pipe_w, opal_cr_entry_point_signal));
+
  cleanup:
     if( NULL != tmp_pid) {
         free(tmp_pid);

orte/mca/snapc/full/snapc_full_local.c

@@ -1124,13 +1124,19 @@ static int snapc_full_local_start_checkpoint_all(size_t ckpt_state)
         }
 
         OPAL_OUTPUT_VERBOSE((20, mca_snapc_full_component.super.output_handle,
-                             "Local) Signal process (%d)\n",
-                             (int) vpid_snapshot->super.process_pid));
+                             "Local) Signal process (%d) with signal %d\n",
+                             (int) vpid_snapshot->super.process_pid,
+                             opal_cr_entry_point_signal));
 
         /*
          * Signal the application
          */
         if( 0 != (ret = kill(vpid_snapshot->super.process_pid, opal_cr_entry_point_signal) ) ) {
+            opal_output(mca_snapc_full_component.super.output_handle,
+                        "local) Error: Failed to signal process %d with signal %d. %d\n", 
+                        (int) vpid_snapshot->super.process_pid,
+                        opal_cr_entry_point_signal,
+                        ret);
             exit_status = ret;
             goto cleanup;
         }
@@ -1239,6 +1245,7 @@ static int snapc_full_local_start_ckpt_open_comm(orte_snapc_full_local_snapshot_
                                      vpid_snapshot->comm_pipe_r, ret, s_time, max_wait_time));
             }
             usleep(usleep_time);
+            opal_event_loop(OPAL_EVLOOP_NONBLOCK);
             continue;
         }
         else if( 0 > (ret = access(vpid_snapshot->comm_pipe_w, F_OK) )) {
@@ -1249,6 +1256,7 @@ static int snapc_full_local_start_ckpt_open_comm(orte_snapc_full_local_snapshot_
                                      vpid_snapshot->comm_pipe_w, ret, s_time, max_wait_time));
             }
             usleep(usleep_time);
+            opal_event_loop(OPAL_EVLOOP_NONBLOCK);
             continue;
         }
         else {