Merge pull request #1195 from hppritcha/topic/wlm_detect

support Cray nativized slurm environment
2015-12-15 07:58:53 -07:00 · 2015-12-15 07:58:53 -07:00 · 7a82174747
--- a/config/orte_check_alps.m4
+++ b/config/orte_check_alps.m4
@ -129,7 +129,18 @@ AC_DEFUN([ORTE_CHECK_ALPS],[
                                                       AC_MSG_WARN([on the configure line using --with-alps option.])
                                                       AC_MSG_ERROR([Aborting])],[])]
                                                       )
-
+                               PKG_CHECK_MODULES_STATIC([CRAY_WLM_DETECT], [cray-wlm_detect],
+                                               [orte_check_cray_alps_happy="yes"
+                                                AC_DEFINE_UNQUOTED([CRAY_WLM_DETECT],[1],
+                                                                   [defined to 1 if cray wlm available, 0 otherwise])
+                                               ],
+                                               [orte_check_cray_alps_happy="no"]
+                                               [AS_IF([test "$with_alps" = "yes"],
+                                                      [AC_MSG_WARN([ALPS support requested but pkg-config failed.])
+                                                       AC_MSG_WARN([Need to explicitly indicate ALPS directory])
+                                                       AC_MSG_WARN([on the configure line using --with-alps option.])
+                                                       AC_MSG_ERROR([Aborting])],[])]
+                                                       )
                            ],
                            [AC_MSG_WARN([See ./configure --help for how to control Open MPI])
                             AC_MSG_WARN([configuration for ALPS on CLE 5 and higher systems])
@ -146,10 +157,10 @@ AC_DEFUN([ORTE_CHECK_ALPS],[

        AS_IF([test "$orte_check_cray_alps_happy" = "yes"],
              [$1_LDFLAGS="$CRAY_ALPSLLI_LIBS $CRAY_ALPSUTIL_LIBS"
-               $1_CPPFLAGS="$CRAY_ALPSLLI_CFLAGS $CRAY_ALPSUTIL_CFLAGS $CRAY_ALPS_CFLAGS"
-               $1_LIBS="$CRAY_ALPSLLI_LIBS $CRAY_ALPSUTIL_LIBS"
-               $1_WRAPPER_EXTRA_LDFLAGS="$CRAY_ALPSLLI_LIBS $CRAY_ALPSUTIL_LIBS"
-               $1_WRAPPER_EXTRA_LIBS="$CRAY_ALPSLLI_LIBS $CRAY_ALPSUTIL_LIBS"],
+               $1_CPPFLAGS="$CRAY_ALPSLLI_CFLAGS $CRAY_ALPSUTIL_CFLAGS $CRAY_ALPS_CFLAGS $CRAY_WLM_DETECT_CFLAGS"
+               $1_LIBS="$CRAY_ALPSLLI_LIBS $CRAY_ALPSUTIL_LIBS $CRAY_WLM_DETECT_LIBS"
+               $1_WRAPPER_EXTRA_LDFLAGS="$CRAY_ALPSLLI_LIBS $CRAY_ALPSUTIL_LIBS $CRAY_WLM_DETECT_LIBS"
+               $1_WRAPPER_EXTRA_LIBS="$CRAY_ALPSLLI_LIBS $CRAY_ALPSUTIL_LIBS $CRAY_WLM_DETECT_LIBS"],
              [])

    fi
--- a/orte/mca/plm/alps/Makefile.am
+++ b/orte/mca/plm/alps/Makefile.am
@ -10,6 +10,8 @@
 # Copyright (c) 2004-2005 The Regents of the University of California.
 #                         All rights reserved.
 # Copyright (c) 2008-2010 Cisco Systems, Inc.  All rights reserved.
+# Copyright (c) 2015      Los Alamos National Security, LLC.  All rights
+#                         reserved.
 # $COPYRIGHT$
 #
 # Additional copyrights may follow
@ -39,8 +41,13 @@ endif
 mcacomponentdir = $(ortelibdir)
 mcacomponent_LTLIBRARIES = $(component_install)
 mca_plm_alps_la_SOURCES = $(sources)
-mca_plm_alps_la_LDFLAGS = -module -avoid-version
+mca_plm_alps_la_CPPFLAGS = $(plm_alps_CPPFLAGS)
+mca_plm_alps_la_LDFLAGS = -module -avoid-version $(plm_alps_LDFLAGS)
+mca_plm_alps_la_LIBADD = $(plm_alps_LIBS) \
+    $(ORTE_TOP_BUILDDIR)/orte/mca/common/alps/lib@ORTE_LIB_PREFIX@mca_common_alps.la

 noinst_LTLIBRARIES = $(component_noinst)
 libmca_plm_alps_la_SOURCES =$(sources)
-libmca_plm_alps_la_LDFLAGS = -module -avoid-version
+libmca_plm_alps_la_CPPFLAGS = $(plm_alps_CPPFLAGS)
+libmca_plm_alps_la_LDFLAGS = -module -avoid-version $(plm_alps_LDFLAGS)
+libmca_plm_alps_la_LIBADD = $(plm_alps_LIBS)
--- a/orte/mca/plm/alps/configure.m4
+++ b/orte/mca/plm/alps/configure.m4
@ -27,5 +27,11 @@ AC_DEFUN([MCA_orte_plm_alps_CONFIG],[

 	ORTE_CHECK_ALPS([plm_alps], [plm_alps_happy="yes"], [plm_alps_happy="no"])

-    AS_IF([test "$plm_alps_happy" = "yes"], [$1], [$2])
+    AS_IF([test "$plm_alps_happy" = "yes"],
+          [$1
+           AC_SUBST([plm_alps_CPPFLAGS])
+           AC_SUBST([plm_alps_LDFLAGS])
+           AC_SUBST([plm_alps_LIBS])],
+          [$2])
+
 ])dnl
--- a/orte/mca/plm/alps/plm_alps.h
+++ b/orte/mca/plm/alps/plm_alps.h
@ -23,6 +23,9 @@

 #include "orte/mca/mca.h"
 #include "orte/mca/plm/plm.h"
+#if CRAY_WLM_DETECT
+#include "wlm_detect.h"
+#endif

 BEGIN_C_DECLS

--- a/orte/mca/plm/alps/plm_alps_component.c
+++ b/orte/mca/plm/alps/plm_alps_component.c
@ -136,8 +136,29 @@ static int plm_alps_open(void)

 static int orte_plm_alps_component_query(mca_base_module_t **module, int *priority)
 {
-    *priority = mca_plm_alps_component.priority;
-    *module = (mca_base_module_t *) &orte_plm_alps_module;
+    int alps_wlm_active = 1;
+#if CRAY_WLM_DETECT
+    char slurm[]="SLURM";
+
+    if(!strcmp(slurm,wlm_detect_get_active())) {
+        alps_wlm_active = 0;
+    }
+#endif
+
+    if (alps_wlm_active) {
+        *priority = mca_plm_alps_component.priority;
+        *module = (mca_base_module_t *) &orte_plm_alps_module;
+        OPAL_OUTPUT_VERBOSE((1, orte_plm_base_framework.framework_output,
+                            "%s plm:alps: available for selection",
+                             ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
+    } else {
+        *priority = 0;
+        *module = NULL;
+        OPAL_OUTPUT_VERBOSE((1, orte_plm_base_framework.framework_output,
+                            "%s plm:alps: not available, slurm present",
+                             ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
+    }
+
    return ORTE_SUCCESS;
 }

--- a/orte/mca/plm/slurm/configure.m4
+++ b/orte/mca/plm/slurm/configure.m4
@ -11,7 +11,7 @@
 # Copyright (c) 2004-2005 The Regents of the University of California.
 #                         All rights reserved.
 # Copyright (c) 2009-2010 Cisco Systems, Inc.  All rights reserved.
-# Copyright (c) 2011-2013 Los Alamos National Security, LLC.
+# Copyright (c) 2011-2015 Los Alamos National Security, LLC.
 #                         All rights reserved.
 # $COPYRIGHT$
 #
@ -37,4 +37,11 @@ AC_DEFUN([MCA_orte_plm_slurm_CONFIG],[
    AC_SUBST([plm_slurm_CPPFLAGS])
    AC_SUBST([plm_slurm_LDFLAGS])
    AC_SUBST([plm_slurm_LIBS])
+
+    # check to see if this is a Cray nativized slurm env.
+
+    ORTE_CHECK_ALPS([plm_slurm_cray],
+                    [AC_DEFINE_UNQUOTED([SLURM_CRAY_ENV],[1],
+                    [defined to 1 if slurm cray env, 0 otherwise])])
+
 ])dnl
--- a/orte/mca/plm/slurm/plm_slurm_module.c
+++ b/orte/mca/plm/slurm/plm_slurm_module.c
@ -10,7 +10,7 @@
 * Copyright (c) 2004-2005 The Regents of the University of California.
 *                         All rights reserved.
 * Copyright (c) 2006-2014 Cisco Systems, Inc.  All rights reserved.
- * Copyright (c) 2007-2012 Los Alamos National Security, LLC.  All rights
+ * Copyright (c) 2007-2015 Los Alamos National Security, LLC.  All rights
 *                         reserved.
 * Copyright (c) 2014      Intel Corporation.  All rights reserved.
 * $COPYRIGHT$
@ -27,6 +27,7 @@
 */

 #include "orte_config.h"
+#include "orte/runtime/orte_globals.h"

 #include <string.h>
 #include <sys/types.h>
@ -275,6 +276,18 @@ static void launch_daemons(int fd, short args, void *cbdata)
     */
    opal_argv_append(&argc, &argv, "--cpu_bind=none");

+#if SLURM_CRAY_ENV
+    /*
+     * If in a SLURM/Cray env. make sure that Cray PMI is not pulled in,
+     * neither as a constructor run when orteds start, nor selected
+     * when pmix components are registered
+     */
+
+    opal_setenv("PMI_NO_PREINITIALIZE", "1", false, &orte_launch_environ);
+    opal_setenv("PMI_NO_FORK", "1", false, &orte_launch_environ);
+    opal_setenv("OMPI_NO_USE_CRAY_PMI", "1", false, &orte_launch_environ);
+#endif
+
    /* Append user defined arguments to srun */
    if ( NULL != mca_plm_slurm_component.custom_args ) {
        custom_strings = opal_argv_split(mca_plm_slurm_component.custom_args, ' ');