[cig-commits] r4879 - in long/3D/Gale/trunk/src/StGermain: . Base/IO/src

walter at geodynamics.org walter at geodynamics.org
Wed Oct 11 13:48:18 PDT 2006


Author: walter
Date: 2006-10-11 13:48:18 -0700 (Wed, 11 Oct 2006)
New Revision: 4879

Modified:
   long/3D/Gale/trunk/src/StGermain/
   long/3D/Gale/trunk/src/StGermain/Base/IO/src/Journal.c
Log:
 r2936 at earth:  boo | 2006-10-11 13:42:48 -0700
  r2852 at earth (orig r3840):  AlanLo | 2006-10-06 00:16:38 -0700
  
  // SGI MPI on the ess does not always print out everything before it exits
  // To ensure all output is displayed before the program quits on firewall, sleep for one second
  // Alan & Kath 20061006
  
  This is why some tests has been failing intermittently on the ess.
  Sleeping for one second to ensure all relevant info is flushed seems justifible
  (rather than chasing a random last output)
  Its only on firewall so the program was going to exit anyway
  
  
  
 



Property changes on: long/3D/Gale/trunk/src/StGermain
___________________________________________________________________
Name: svk:merge
   - 1ef209d2-b310-0410-a72d-e20c9eb0015c:/cig:2935
afb6c753-b9d0-0310-b4e7-dbd8d91cdd35:/trunk/StGermain:3839
   + 1ef209d2-b310-0410-a72d-e20c9eb0015c:/cig:2936
afb6c753-b9d0-0310-b4e7-dbd8d91cdd35:/trunk/StGermain:3840

Modified: long/3D/Gale/trunk/src/StGermain/Base/IO/src/Journal.c
===================================================================
--- long/3D/Gale/trunk/src/StGermain/Base/IO/src/Journal.c	2006-10-11 20:48:15 UTC (rev 4878)
+++ long/3D/Gale/trunk/src/StGermain/Base/IO/src/Journal.c	2006-10-11 20:48:18 UTC (rev 4879)
@@ -604,6 +604,11 @@
 		/* TODO: Don't use FAILURE until Pat beef's up the test scripts to do .error checks
 			exit( EXIT_FAILURE );
 		*/
+
+		// SGI MPI on the ess does not always print out everything before it exits
+		// To ensure all output is displayed before the program quits on firewall, sleep for one second
+		// Alan & Kath 20061006
+		sleep( 1 );
 		if ( nProc == 1 ) {
 			exit(EXIT_SUCCESS);
 		}



More information about the cig-commits mailing list