plm/lsf: Improve error message if lsb_launch fails

Signed-off-by: Joshua Hursey <jhursey@us.ibm.com>

orte/mca/plm/lsf/help-plm-lsf.txt

@@ -10,27 +10,23 @@
 #                         University of Stuttgart.  All rights reserved.
 # Copyright (c) 2004-2005 The Regents of the University of California.
 #                         All rights reserved.
+# Copyright (c) 2017      IBM Corporation.  All rights reserved.
 # $COPYRIGHT$
 #
 # Additional copyrights may follow
 #
 # $HEADER$
 #
-[tm-bad-launchid]
-The TM (PBS / Torque) process starter cannot spawn the specified
-application on a remote node due to an invalid launch_id.
+[lsb_launch-failed]
+The LSF process starter (lsb_launch) failed to start the daemons on
+the nodes in the allocation.
+ Returned : %d
+ lsberrno : (%d) %s
 
-Node name: %s
-Launch id: %d
-
-This is most likely due to use of the "--hostfile" option to the
-command line. At this time, Open MPI/OpenRTE do not support this
-method of operation. Instead, the system expects to directly read
-information regarding the nodes to be used from the environment.
-
-Removing "--hostfile" from the command line will likely allow the
-application to be launched. This will be fixed in a future release
-to support the use of "--hostfile" on the command line.
+This may mean that one or more of the nodes in the LSF allocation is
+not setup properly. Below is a list of the %d nodes that were passed
+to lsb_launch:
+%s
 #
 [multiple-prefixes]
 Multiple different --prefix options were specified to mpirun for the
@@ -40,18 +36,3 @@ starter in Open MPI.
 The first two prefix values supplied for node %s were:
     %s
 and %s
-#
-[tm-spawn-failed]
-The TM (PBS / Torque) process starter failed to spawn a daemon (orted)
-on a remote node.
-
-Command line: %s
-Node name: %s
-Launch id: %d
-
-If you do not understand this error mesage, please try the following:
-
-1. Ensure that the executable "orted" is in your PATH
-2. Use the --prefix option to indicate where we can
-   find that executable
-3. Talk to your local system administrator

orte/mca/plm/lsf/plm_lsf_module.c

@@ -15,6 +15,7 @@
  * Copyright (c) 2008      Institut National de Recherche en Informatique
  *                         et Automatique. All rights reserved.
  * Copyright (c) 2014-2017 Intel, Inc.  All rights reserved.
+ * Copyright (c) 2017      IBM Corporation.  All rights reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -342,9 +343,14 @@ static void launch_daemons(int fd, short args, void *cbdata)
      * orterun can do the rest of its stuff. Instead, we'll catch any
      * failures and deal with them elsewhere
      */
-    if (lsb_launch(nodelist_argv, argv, LSF_DJOB_REPLACE_ENV | LSF_DJOB_NOWAIT, env) < 0) {
+    if ( (rc = lsb_launch(nodelist_argv, argv, LSF_DJOB_REPLACE_ENV | LSF_DJOB_NOWAIT, env)) < 0) {
         ORTE_ERROR_LOG(ORTE_ERR_FAILED_TO_START);
-        opal_output(0, "lsb_launch failed: %d", rc);
+        char *flattened_nodelist = NULL;
+        flattened_nodelist = opal_argv_join(nodelist_argv, '\n');
+        orte_show_help("help-plm-lsf.txt", "lsb_launch-failed",
+                       true, rc, lsberrno, lsb_sysmsg(),
+                       opal_argv_count(nodelist_argv), flattened_nodelist);
+        free(flattened_nodelist);
         rc = ORTE_ERR_FAILED_TO_START;
         orte_wait_enable();  /* re-enable our SIGCHLD handler */
         goto cleanup;