Make sure to check for the NO_CKPT state while waiting. This means that the target was not able to checkpoint [ever | at this time]. So {{{ompi-checkpoint}}} should exit after printing the error message, instead of hanging and waiting.

Will need to be moved to v1.5 and v1.4. v1.4 will require a custom patch, but should apply cleanly to v1.5. CMRs to follow.

This commit was SVN r22289.

orte/tools/orte-checkpoint/orte-checkpoint.c

@@ -287,11 +287,17 @@ main(int argc, char *argv[])
     if(!orte_checkpoint_globals.nowait) { 
         while( ORTE_SNAPC_CKPT_STATE_FINISHED != orte_checkpoint_globals.ckpt_status &&
                ORTE_SNAPC_CKPT_STATE_STOPPED  != orte_checkpoint_globals.ckpt_status &&
+               ORTE_SNAPC_CKPT_STATE_NO_CKPT  != orte_checkpoint_globals.ckpt_status &&
                ORTE_SNAPC_CKPT_STATE_ERROR    != orte_checkpoint_globals.ckpt_status ) {
             opal_progress();
         }
     }
 
+    if( ORTE_SNAPC_CKPT_STATE_NO_CKPT  == orte_checkpoint_globals.ckpt_status ) {
+        exit_status = ORTE_ERROR;
+        goto cleanup;
+    }
+
     if( ORTE_SNAPC_CKPT_STATE_ERROR == orte_checkpoint_globals.ckpt_status ) {
         opal_show_help("help-orte-checkpoint.txt", "ckpt_failure", true,
                        orte_checkpoint_globals.pid, ORTE_ERROR);