Add ORTE ALPS support (Cray XT CNL)

This commit was SVN r17482.

orte/mca/pls/alps/Makefile.am

@@ -0,0 +1,49 @@
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+sources = \
+        pls_alps.h \
+        pls_alps_component.c \
+        pls_alps_module.c
+
+dist_pkgdata_DATA = help-pls-alps.txt
+
+# Make the output library in this directory, and name it either
+# mca_<type>_<name>.la (for DSO builds) or libmca_<type>_<name>.la
+# (for static builds).
+
+if OMPI_BUILD_pls_alps_DSO
+component_noinst =
+component_install = mca_pls_alps.la
+else
+component_noinst = libmca_pls_alps.la
+component_install =
+endif
+
+mcacomponentdir = $(pkglibdir)
+mcacomponent_LTLIBRARIES = $(component_install)
+mca_pls_alps_la_SOURCES = $(sources)
+mca_pls_alps_la_LDFLAGS = -module -avoid-version
+mca_pls_alps_la_LIBADD = \
+    $(top_ompi_builddir)/orte/libopen-rte.la \
+    $(top_ompi_builddir)/opal/libopen-pal.la
+
+noinst_LTLIBRARIES = $(component_noinst)
+libmca_pls_alps_la_SOURCES =$(sources)
+libmca_pls_alps_la_LDFLAGS = -module -avoid-version

orte/mca/pls/alps/configure.m4

@@ -0,0 +1,25 @@
+# -*- shell-script -*-
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+# MCA_pls_alps_CONFIG([action-if-found], [action-if-not-found])
+# -----------------------------------------------------------
+AC_DEFUN([MCA_pls_alps_CONFIG],[
+	OMPI_CHECK_ALPS([pls_alps], [$1], [$2])
+])dnl

orte/mca/pls/alps/configure.params

@@ -0,0 +1,23 @@
+# -*- shell-script -*-
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2007      Los Alamos National Security, LLC.  All rights
+#                         reserved. 
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+PARAM_CONFIG_FILES="Makefile"

orte/mca/pls/alps/help-pls-alps.txt

@@ -0,0 +1,42 @@
+# -*- text -*-
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+[multiple-prefixes]
+The ALPS process starter for Open MPI does not support multiple
+different --prefix options to mpirun.  You can specify at most one
+unique value for the --prefix option (in any of the application
+contexts); it will be applied to all the application contexts of your
+parallel job.
+
+Put simply, you must have Open MPI installed in the same location on
+all of your ALPS nodes.
+
+Multiple different --prefix options were specified to mpirun.  This is
+a fatal error for the ALPS process starter in Open MPI.
+
+The first two prefix values supplied were:
+    %s
+and %s
+#
+[no-hosts-in-list]
+The ALPS process starter for Open MPI didn't find any hosts in
+the map for this application. This can be caused by a lack of
+an allocation, or by an error in the Open MPI code. Please check
+to ensure you have a ALPS allocation. If you do, then please pass
+the error to the Open MPI user's mailing list for assistance.

orte/mca/pls/alps/pls_alps.h

@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ */
+
+#ifndef ORTE_PLS_ALPS_EXPORT_H
+#define ORTE_PLS_ALPS_EXPORT_H
+
+#include "orte_config.h"
+
+#include "opal/mca/mca.h"
+#include "orte/mca/pls/pls.h"
+
+#if defined(c_plusplus) || defined(__cplusplus)
+extern "C" {
+#endif
+
+    struct orte_pls_alps_component_t {
+        orte_pls_base_component_t super;
+        int priority;
+        int debug;
+        bool timing;
+        char *orted;
+        char *custom_args;
+    };
+    typedef struct orte_pls_alps_component_t orte_pls_alps_component_t;
+
+    /*
+     * Globally exported variable
+     */
+    
+    ORTE_MODULE_DECLSPEC extern orte_pls_alps_component_t 
+        mca_pls_alps_component;
+    ORTE_DECLSPEC extern orte_pls_base_module_t
+        orte_pls_alps_module;
+
+#if defined(c_plusplus) || defined(__cplusplus)
+}
+#endif
+#endif /* ORTE_PLS_ALPS_EXPORT_H */

orte/mca/pls/alps/pls_alps_component.c

@@ -0,0 +1,167 @@
+/*
+ * Copyright (c) 2004-2007 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ *
+ * These symbols are in a file by themselves to provide nice linker
+ * semantics.  Since linkers generally pull in symbols by object
+ * files, keeping these symbols as the only symbols in this file
+ * prevents utility programs such as "ompi_info" from having to import
+ * entire components just to query their version and parameters.
+ */
+
+#include "orte_config.h"
+
+#include "opal/util/output.h"
+#include "opal/mca/base/mca_base_param.h"
+#include "orte/orte_constants.h"
+
+#include "orte/util/proc_info.h"
+#include "orte/mca/errmgr/errmgr.h"
+
+#include "orte/mca/pls/pls.h"
+#include "orte/mca/pls/base/base.h"
+#include "orte/mca/pls/base/pls_private.h"
+#include "pls_alps.h"
+
+
+/*
+ * Public string showing the pls ompi_alps component version number
+ */
+const char *mca_pls_alps_component_version_string =
+  "Open MPI alps pls MCA component version " ORTE_VERSION;
+
+
+/*
+ * Local functions
+ */
+static int pls_alps_open(void);
+static int pls_alps_close(void);
+static orte_pls_base_module_t *pls_alps_init(int *priority);
+
+
+/*
+ * Instantiate the public struct with all of our public information
+ * and pointers to our public functions in it
+ */
+
+orte_pls_alps_component_t mca_pls_alps_component = {
+
+    {
+        /* First, the mca_component_t struct containing meta
+           information about the component itself */
+
+        {
+            /* Indicate that we are a pls v1.3.0 component (which also
+               implies a specific MCA version) */
+
+            ORTE_PLS_BASE_VERSION_1_3_0,
+            
+            /* Component name and version */
+            
+            "alps",
+            ORTE_MAJOR_VERSION,
+            ORTE_MINOR_VERSION,
+            ORTE_RELEASE_VERSION,
+            
+            /* Component open and close functions */
+            
+            pls_alps_open,
+            pls_alps_close
+        },
+        
+        /* Next the MCA v1.0.0 component meta data */
+        
+        {
+            /* The component is checkpoint ready */
+            MCA_BASE_METADATA_PARAM_CHECKPOINT
+        },
+        
+        /* Initialization / querying functions */
+        
+        pls_alps_init
+    }
+
+    /* Other orte_pls_alps_component_t items -- left uninitialized
+       here; will be initialized in pls_alps_open() */
+};
+
+
+static int pls_alps_open(void)
+{
+    mca_base_component_t *comp = &mca_pls_alps_component.super.pls_version;
+    int tmp, value;
+
+    mca_base_param_reg_int(comp, "debug", "Enable debugging of alps pls",
+                           false, false, 0, 
+                           &mca_pls_alps_component.debug);
+    if (mca_pls_alps_component.debug == 0) {
+        mca_base_param_reg_int_name("orte", "debug",
+                                    "Whether or not to enable debugging output for all ORTE components (0 or 1)",
+                                    false, false, false, &mca_pls_alps_component.debug);
+    }
+
+    mca_base_param_reg_int(comp, "priority", "Default selection priority",
+                           false, false, 75, 
+                           &mca_pls_alps_component.priority);
+
+    mca_base_param_reg_string(comp, "orted",
+                              "Command to use to start proxy orted",
+                              false, false, "orted",
+                              &mca_pls_alps_component.orted);
+
+    tmp = mca_base_param_reg_int_name("orte", "timing",
+                                      "Request that critical timing loops be measured",
+                                      false, false, 0, &value);
+    if (value != 0) {
+        mca_pls_alps_component.timing = true;
+    } else {
+        mca_pls_alps_component.timing = false;
+    }
+    
+    mca_base_param_reg_string(comp, "args",
+                              "Custom arguments to srun",
+                              false, false, NULL,
+                              &mca_pls_alps_component.custom_args);
+
+    return ORTE_SUCCESS;
+}
+
+
+static orte_pls_base_module_t *pls_alps_init(int *priority)
+{
+    /* if we are NOT an HNP, then don't select us */
+    if (!orte_process_info.seed) {
+        return NULL;
+    }
+    
+    *priority = mca_pls_alps_component.priority;
+    return &orte_pls_alps_module;
+}
+
+
+static int pls_alps_close(void)
+{
+    if (NULL != mca_pls_alps_component.orted) {
+        free(mca_pls_alps_component.orted);
+    }
+
+    if (NULL != mca_pls_alps_component.custom_args) {
+        free(mca_pls_alps_component.custom_args);
+    }
+
+    return ORTE_SUCCESS;
+}

orte/mca/pls/alps/pls_alps_module.c

@@ -0,0 +1,630 @@
+/*
+ * Copyright (c) 2004-2007 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2006 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2006-2007 Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2007      Los Alamos National Security, LLC.  All rights
+ *                         reserved. 
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ *
+ * Additional copyrights may follow
+ *
+ * $HEADER$
+ *
+ * These symbols are in a file by themselves to provide nice linker
+ * semantics.  Since linkers generally pull in symbols by object
+ * files, keeping these symbols as the only symbols in this file
+ * prevents utility programs such as "ompi_info" from having to import
+ * entire components just to query their version and parameters.
+ */
+
+#include "orte_config.h"
+#include "orte/orte_constants.h"
+#include "orte/orte_types.h"
+
+#include <sys/types.h>
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+#include <signal.h>
+#ifdef HAVE_STDLIB_H
+#include <stdlib.h>
+#endif
+#ifdef HAVE_SYS_TYPES_H
+#include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_TIME_H
+#include <sys/time.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+#include <sys/stat.h>
+#endif
+#ifdef HAVE_FCNTL_H
+#include <fcntl.h>
+#endif
+
+#include "opal/mca/installdirs/installdirs.h"
+#include "opal/util/argv.h"
+#include "opal/util/output.h"
+#include "opal/util/opal_environ.h"
+#include "opal/util/path.h"
+#include "opal/util/show_help.h"
+#include "opal/util/basename.h"
+#include "opal/mca/base/mca_base_param.h"
+
+#include "orte/runtime/params.h"
+#include "orte/runtime/runtime.h"
+#include "orte/runtime/orte_wakeup.h"
+#include "orte/runtime/orte_wait.h"
+#include "orte/mca/ns/base/base.h"
+#include "orte/mca/rml/rml.h"
+#include "orte/mca/errmgr/errmgr.h"
+#include "orte/mca/smr/smr.h"
+#include "orte/mca/rmaps/rmaps.h"
+
+#include "orte/mca/pls/pls.h"
+#include "orte/mca/pls/base/base.h"
+#include "orte/mca/pls/base/pls_private.h"
+#include "pls_alps.h"
+
+
+/*
+ * Local functions
+ */
+static int pls_alps_launch_job(orte_jobid_t jobid);
+static int pls_alps_terminate_job(orte_jobid_t jobid, struct timeval *timeout, opal_list_t *attrs);
+static int pls_alps_terminate_orteds(struct timeval *timeout, opal_list_t *attrs);
+static int pls_alps_terminate_proc(const orte_process_name_t *name);
+static int pls_alps_signal_job(orte_jobid_t jobid, int32_t signal, opal_list_t *attrs);
+static int pls_alps_signal_proc(const orte_process_name_t *name, int32_t signal);
+static int pls_alps_finalize(void);
+
+static int pls_alps_start_proc(int argc, char **argv, char **env,
+                                char *prefix);
+
+
+/*
+ * Global variable
+ */
+orte_pls_base_module_1_3_0_t orte_pls_alps_module = {
+    pls_alps_launch_job,
+    pls_alps_terminate_job,
+    pls_alps_terminate_orteds,
+    pls_alps_terminate_proc,
+    pls_alps_signal_job,
+    pls_alps_signal_proc,
+    pls_alps_finalize
+};
+
+/*
+ * Local variables
+ */
+static pid_t alps_pid = 0;
+static orte_jobid_t active_job = ORTE_JOBID_INVALID;
+static bool failed_launch;
+
+
+/* When working in this function, ALWAYS jump to "cleanup" if
+ * you encounter an error so that orterun will be woken up and
+ * the job can cleanly terminate
+ */
+static int pls_alps_launch_job(orte_jobid_t jobid)
+{
+    orte_job_map_t *map = NULL;
+    opal_list_item_t *item;
+    size_t num_nodes;
+    char *jobid_string = NULL;
+    char *param;
+    char **argv = NULL;
+    int argc;
+    int rc;
+    char *tmp;
+    char** env = NULL;
+    char* var;
+    char *nodelist_flat;
+    char **nodelist_argv;
+    int nodelist_argc;
+    orte_process_name_t name;
+    char *name_string;
+    char **custom_strings;
+    int num_args, i;
+    char *cur_prefix;
+    struct timeval joblaunchstart, launchstart, launchstop;
+    int proc_name_index = 0;
+
+    if (mca_pls_alps_component.timing) {
+        if (0 != gettimeofday(&joblaunchstart, NULL)) {
+            opal_output(0, "pls_alps: could not obtain job start time");
+        }        
+    }
+    
+    /* save the active jobid */
+    active_job = jobid;
+    
+    /* indicate the state of the launch */
+    failed_launch = true;
+    
+    /* Query the map for this job.
+     * We need the entire mapping for a couple of reasons:
+     *  - need the prefix to start with.
+     *  - need to know if we are launching on a subset of the allocated nodes
+     * All other mapping responsibilities fall to orted in the fork PLS
+     */
+    rc = orte_rmaps.get_job_map(&map, jobid);
+    if (ORTE_SUCCESS != rc) {
+        ORTE_ERROR_LOG(rc);
+        goto cleanup;
+    }
+    
+    num_nodes = map->num_new_daemons;
+    if (num_nodes == 0) {
+        /* no new daemons required - just launch apps */
+        goto launch_apps;
+    }
+
+    /* need integer value for command line parameter */
+    asprintf(&jobid_string, "%lu", (unsigned long) jobid);
+
+    /*
+     * start building argv array
+     */
+    argv = NULL;
+    argc = 0;
+
+    /*
+     * ALPS aprun  OPTIONS
+     */
+
+    /* add the aprun command */
+    opal_argv_append(&argc, &argv, "aprun");
+
+    /* Append user defined arguments to aprun */
+    if ( NULL != mca_pls_alps_component.custom_args ) {
+        custom_strings = opal_argv_split(mca_pls_alps_component.custom_args, ' ');
+        num_args       = opal_argv_count(custom_strings);
+        for (i = 0; i < num_args; ++i) {
+            opal_argv_append(&argc, &argv, custom_strings[i]);
+        }
+        opal_argv_free(custom_strings);
+    }
+
+    /* number of processors needed */
+    asprintf(&tmp, "-n %lu", (unsigned long) num_nodes);
+    opal_argv_append(&argc, &argv, tmp);
+    free(tmp);
+    opal_argv_append(&argc, &argv, "-N 1");
+
+    /* create nodelist */
+    nodelist_argv = NULL;
+    nodelist_argc = 0;
+
+    for (item =  opal_list_get_first(&map->nodes);
+         item != opal_list_get_end(&map->nodes);
+         item =  opal_list_get_next(item)) {
+        orte_mapped_node_t* node = (orte_mapped_node_t*)item;
+
+        /* if the daemon already exists on this node, then
+         * don't include it
+         */
+        if (node->daemon_preexists) {
+            continue;
+        }
+        
+        /* otherwise, add it to the list of nodes upon which
+         * we need to launch a daemon
+         */
+        opal_argv_append(&nodelist_argc, &nodelist_argv, node->nodename);
+    }
+    if (0 == opal_argv_count(nodelist_argv)) {
+        opal_show_help("help-pls-alps.txt", "no-hosts-in-list", true);
+        rc = ORTE_ERR_FAILED_TO_START;
+        goto cleanup;
+    }
+    nodelist_flat = opal_argv_join(nodelist_argv, ',');
+    opal_argv_free(nodelist_argv);
+    asprintf(&tmp, "-L %s", nodelist_flat);
+    opal_argv_append(&argc, &argv, tmp);
+    free(tmp);
+
+
+    /*
+     * ORTED OPTIONS
+     */
+
+    /* add the daemon command (as specified by user) */
+    opal_argv_append(&argc, &argv, mca_pls_alps_component.orted);
+    
+    /* ensure we don't lose contact */
+    orte_no_daemonize_flag = true;
+
+    /* Add basic orted command line options, including debug flags */
+    orte_pls_base_orted_append_basic_args(&argc, &argv,
+                                          &proc_name_index,
+                                          NULL);
+
+    /* force orted to use the alps sds */
+    opal_argv_append(&argc, &argv, "--ns-nds");
+    opal_argv_append(&argc, &argv, "alps");
+
+    /* tell the new daemons the base of the name list so they can compute
+     * their own name on the other end
+     */
+    name.jobid = 0;
+    name.vpid = map->daemon_vpid_start;
+    rc = orte_ns.get_proc_name_string(&name_string, &name);
+    if (ORTE_SUCCESS != rc) {
+        opal_output(0, "pls_alps: unable to create process name");
+        goto cleanup;
+    }
+
+    free(argv[proc_name_index]);
+    argv[proc_name_index] = strdup(name_string);
+    free(name_string);
+
+    if (mca_pls_alps_component.debug) {
+        param = opal_argv_join(argv, ' ');
+        if (NULL != param) {
+            opal_output(0, "pls:alps: final top-level argv:");
+            opal_output(0, "pls:alps:     %s", param);
+            free(param);
+        }
+    }
+
+    /* Copy the prefix-directory specified in the
+       corresponding app_context.  If there are multiple,
+       different prefix's in the app context, complain (i.e., only
+       allow one --prefix option for the entire alps run -- we
+       don't support different --prefix'es for different nodes in
+       the ALPS pls) */
+    cur_prefix = NULL;
+    for (i=0; i < map->num_apps; i++) {
+        char * app_prefix_dir = map->apps[i]->prefix_dir;
+         /* Check for already set cur_prefix_dir -- if different,
+           complain */
+        if (NULL != app_prefix_dir) {
+            if (NULL != cur_prefix &&
+                0 != strcmp (cur_prefix, app_prefix_dir)) {
+                opal_show_help("help-pls-alps.txt", "multiple-prefixes",
+                               true, cur_prefix, app_prefix_dir);
+                return ORTE_ERR_FATAL;
+            }
+
+            /* If not yet set, copy it; iff set, then it's the
+               same anyway */
+            if (NULL == cur_prefix) {
+                cur_prefix = strdup(app_prefix_dir);
+                if (mca_pls_alps_component.debug) {
+                    opal_output (0, "pls:alps: Set prefix:%s",
+                                 cur_prefix);
+                }
+            }
+        }
+    }
+
+    /* setup environment */
+    env = opal_argv_copy(environ);
+    
+    /* purge it of any params not for orteds */
+    orte_pls_base_purge_mca_params(&env);
+    
+    /* add the nodelist */
+    var = mca_base_param_environ_variable("orte", "alps", "nodelist");
+    opal_setenv(var, nodelist_flat, true, &env);
+    free(nodelist_flat);
+    free(var);
+
+    if (mca_pls_alps_component.timing) {
+        if (0 != gettimeofday(&launchstart, NULL)) {
+            opal_output(0, "pls_alps: could not obtain start time");
+        }        
+    }
+    
+    /* exec the daemon(s) */
+    if (ORTE_SUCCESS != (rc = pls_alps_start_proc(argc, argv, env, cur_prefix))) {
+        ORTE_ERROR_LOG(rc);
+        goto cleanup;
+    }
+    
+    /* do NOT wait for alps to complete. Alps only completes when the processes
+     * it starts - in this case, the orteds - complete. Instead, we'll catch
+     * any alps failures and deal with them elsewhere
+     */
+    
+    /* wait for daemons to callback */
+    if (ORTE_SUCCESS != (rc = orte_pls_base_daemon_callback(map->num_new_daemons))) {
+        ORTE_ERROR_LOG(rc);
+        goto cleanup;
+    }
+    
+launch_apps:
+    if (ORTE_SUCCESS != (rc = orte_pls_base_launch_apps(map))) {
+        ORTE_ERROR_LOG(rc);
+        goto cleanup;
+    }
+
+    /* declare the launch a success */
+    failed_launch = false;
+    
+    if (mca_pls_alps_component.timing) {
+        if (0 != gettimeofday(&launchstop, NULL)) {
+             opal_output(0, "pls_alps: could not obtain stop time");
+         } else {
+             opal_output(0, "pls_alps: daemon block launch time is %ld usec",
+                         (launchstop.tv_sec - launchstart.tv_sec)*1000000 + 
+                         (launchstop.tv_usec - launchstart.tv_usec));
+             opal_output(0, "pls_alps: total job launch time is %ld usec",
+                         (launchstop.tv_sec - joblaunchstart.tv_sec)*1000000 + 
+                         (launchstop.tv_usec - joblaunchstart.tv_usec));
+         }
+    }
+
+    if (ORTE_SUCCESS != rc) {
+        opal_output(0, "pls:alps: start_procs returned error %d", rc);
+        goto cleanup;
+    }
+
+    /* JMS: short we stash the alps pid in the gpr somewhere for cleanup? */
+
+cleanup:
+    if (NULL != map) {
+        OBJ_RELEASE(map);
+    }
+    if (NULL != argv) {
+        opal_argv_free(argv);
+    }
+    if (NULL != env) {
+        opal_argv_free(env);
+    }
+    
+    if(NULL != jobid_string) {
+        free(jobid_string);
+    }
+    
+    /* check for failed launch - if so, force terminate */
+    if (failed_launch) {
+        orte_pls_base_daemon_failed(jobid, false, -1, 0, ORTE_JOB_STATE_FAILED_TO_START);
+    }
+    
+    return rc;
+}
+
+
+static int pls_alps_terminate_job(orte_jobid_t jobid, struct timeval *timeout, opal_list_t *attrs)
+{
+    int rc;
+    
+    /* order them to kill their local procs for this job */
+    if (ORTE_SUCCESS != (rc = orte_pls_base_orted_kill_local_procs(jobid, timeout, attrs))) {
+        ORTE_ERROR_LOG(rc);
+    }
+    
+    return rc;
+}
+
+
+/**
+* Terminate the orteds for a given job
+ */
+static int pls_alps_terminate_orteds(struct timeval *timeout, opal_list_t *attrs)
+{
+    int rc;
+    
+    /* deregister the waitpid callback to ensure we don't make it look like
+     * alps failed when it didn't. Since the alps may have already completed,
+     * do NOT ERROR_LOG any return code to avoid confusing, duplicate error
+     * messages
+     */
+    orte_wait_cb_cancel(alps_pid);
+    
+    /* tell them to die! */
+    if (ORTE_SUCCESS != (rc = orte_pls_base_orted_exit(timeout, attrs))) {
+        ORTE_ERROR_LOG(rc);
+    }
+    
+    return rc;
+}
+
+
+/*
+ * The way we've used ALPS, we can't kill individual processes --
+ * we'll kill the entire job
+ */
+static int pls_alps_terminate_proc(const orte_process_name_t *name)
+{
+    opal_output(0, "pls:alps:terminate_proc: not supported");
+    return ORTE_ERR_NOT_SUPPORTED;
+}
+
+
+/**
+ * Signal all the processes in the child alps by sending the signal directly to it
+ */
+static int pls_alps_signal_job(orte_jobid_t jobid, int32_t signal, opal_list_t *attrs)
+{
+    if (0 != alps_pid) {
+        kill(alps_pid, (int)signal);
+   }
+    return ORTE_SUCCESS;
+}
+
+
+/*
+ * Signal a specific process
+ */
+static int pls_alps_signal_proc(const orte_process_name_t *name, int32_t signal)
+{
+    opal_output(0, "pls:alps:signal_proc: not supported");
+    return ORTE_ERR_NOT_SUPPORTED;
+}
+
+
+static int pls_alps_finalize(void)
+{
+    int rc;
+    
+    /* cleanup any pending recvs */
+    if (ORTE_SUCCESS != (rc = orte_pls_base_comm_stop())) {
+        ORTE_ERROR_LOG(rc);
+    }
+    
+    return ORTE_SUCCESS;
+}
+
+
+static void alps_wait_cb(pid_t pid, int status, void* cbdata){
+    /* According to the ALPS folks, alps always returns the highest exit
+       code of our remote processes. Thus, a non-zero exit status doesn't
+       necessarily mean that alps failed - it could be that an orted returned
+       a non-zero exit status. Of course, that means the orted failed(!), so
+       the end result is the same - the job didn't start.
+    
+       As a result, we really can't do much with the exit status itself - it
+       could be something in errno (if alps itself failed), or it could be
+       something returned by an orted, or it could be something returned by
+       the OS (e.g., couldn't find the orted binary). Somebody is welcome
+       to sort out all the options and pretty-print a better error message. For
+       now, though, the only thing that really matters is that
+       alps failed. Report the error and make sure that orterun
+       wakes up - otherwise, do nothing!
+    */
+    
+    if (0 != status) {
+        if (failed_launch) {
+            /* we have a problem during launch */
+            opal_output(0, "ERROR: alps failed to start the required daemons.");
+            opal_output(0, "ERROR: This could be due to an inability to find the orted binary");
+            opal_output(0, "ERROR: on one or more remote nodes, lack of authority to execute");
+            opal_output(0, "ERROR: on one or more specified nodes, or other factors.");
+            
+            /* report that the daemon has failed so we break out of the daemon
+             * callback receive and exit
+             */
+            orte_pls_base_daemon_failed(active_job, true, pid, status, ORTE_JOB_STATE_FAILED_TO_START);
+            
+        } else {
+            /* an orted must have died unexpectedly after launch - report
+             * that the daemon has failed so we exit
+             */
+            orte_pls_base_daemon_failed(active_job, false, pid, status, ORTE_JOB_STATE_ABORTED);
+        }
+    }
+    
+}
+
+
+static int pls_alps_start_proc(int argc, char **argv, char **env,
+                                char *prefix)
+{
+    int fd;
+    char *exec_argv = opal_path_findv(argv[0], 0, env, NULL);
+
+    if (NULL == exec_argv) {
+        return ORTE_ERR_NOT_FOUND;
+    }
+
+    alps_pid = fork();
+    if (-1 == alps_pid) {
+        ORTE_ERROR_LOG(ORTE_ERR_SYS_LIMITS_CHILDREN);
+        return ORTE_ERR_SYS_LIMITS_CHILDREN;
+    }
+    
+    if (0 == alps_pid) {  /* child */
+        char *bin_base = NULL, *lib_base = NULL;
+
+        /* Figure out the basenames for the libdir and bindir.  There
+           is a lengthy comment about this in pls_rsh_module.c
+           explaining all the rationale for how / why we're doing
+           this. */
+
+        lib_base = opal_basename(opal_install_dirs.libdir);
+        bin_base = opal_basename(opal_install_dirs.bindir);
+
+        /* If we have a prefix, then modify the PATH and
+           LD_LIBRARY_PATH environment variables.  */
+        if (NULL != prefix) {
+            char *oldenv, *newenv;
+
+            /* Reset PATH */
+            oldenv = getenv("PATH");
+            if (NULL != oldenv) {
+                asprintf(&newenv, "%s/%s:%s", prefix, bin_base, oldenv);
+            } else {
+                asprintf(&newenv, "%s/%s", prefix, bin_base);
+            }
+            opal_setenv("PATH", newenv, true, &env);
+            if (mca_pls_alps_component.debug) {
+                opal_output(0, "pls:alps: reset PATH: %s", newenv);
+            }
+            free(newenv);
+
+            /* Reset LD_LIBRARY_PATH */
+            oldenv = getenv("LD_LIBRARY_PATH");
+            if (NULL != oldenv) {
+                asprintf(&newenv, "%s/%s:%s", prefix, lib_base, oldenv);
+            } else {
+                asprintf(&newenv, "%s/%s", prefix, lib_base);
+            }
+            opal_setenv("LD_LIBRARY_PATH", newenv, true, &env);
+            if (mca_pls_alps_component.debug) {
+                opal_output(0, "pls:alps: reset LD_LIBRARY_PATH: %s",
+                            newenv);
+            }
+            free(newenv);
+        }
+
+        fd = open("/dev/null", O_CREAT|O_WRONLY|O_TRUNC, 0666);
+        if(fd > 0) {
+            dup2(fd, 0);
+        }
+
+        /* When not in debug mode and --debug-daemons was not passed,
+         * tie stdout/stderr to dev null so we don't see messages from orted */
+        if (0 == mca_pls_alps_component.debug && !orte_debug_daemons_flag) {
+            if (fd >= 0) {
+                if (fd != 1) {
+                    dup2(fd,1);
+                }
+                if (fd != 2) {
+                    dup2(fd,2);
+                }
+            }
+        }
+
+        if (fd > 2) {
+            close(fd);
+        }
+
+        /* get the alps process out of orterun's process group so that
+           signals sent from the shell (like those resulting from
+           cntl-c) don't get sent to alps */
+        setpgid(0, 0);
+         
+        
+        char* param = opal_argv_join(argv, ';');
+        execve(exec_argv, argv, env);
+
+        opal_output(0, "pls:alps:start_proc: exec failed");
+        /* don't return - need to exit - returning would be bad -
+           we're not in the calling process anymore */
+        exit(1);
+    } else {  /* parent */
+        /* just in case, make sure that the alps process is not in our
+        process group any more.  Stevens says always do this on both
+        sides of the fork... */
+        setpgid(alps_pid, alps_pid);
+        
+        /* setup the waitpid so we can find out if alps succeeds! */
+        orte_wait_cb(alps_pid, alps_wait_cb, NULL);
+        free(exec_argv);
+    }
+
+    return ORTE_SUCCESS;
+}

orte/mca/ras/alps/Makefile.am

@@ -0,0 +1,58 @@
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+AM_CPPFLAGS = $(ras_alps_CPPFLAGS)
+
+dist_pkgdata_DATA = help-ras-alps.txt
+
+sources = \
+        ras_alps.h \
+        ras_alps_component.c \
+        ras_alps_module.c
+
+
+# Make the output library in this directory, and name it either
+# mca_<type>_<name>.la (for DSO builds) or libmca_<type>_<name>.la
+# (for static builds).
+
+if OMPI_BUILD_ras_alps_DSO
+lib =
+lib_sources = 
+component = mca_ras_alps.la
+component_sources = $(sources)
+else
+lib = libmca_ras_alps.la
+lib_sources = $(sources)
+component =
+component_sources =
+endif
+
+mcacomponentdir = $(pkglibdir)
+mcacomponent_LTLIBRARIES = $(component)
+mca_ras_alps_la_SOURCES = $(component_sources)
+mca_ras_alps_la_LDFLAGS = -module -avoid-version $(ras_alps_LDFLAGS)
+mca_ras_alps_la_LIBADD = \
+    $(ras_alps_LIBS) \
+    $(top_ompi_builddir)/orte/libopen-rte.la \
+    $(top_ompi_builddir)/opal/libopen-pal.la
+
+noinst_LTLIBRARIES = $(lib)
+libmca_ras_alps_la_SOURCES = $(lib_sources)
+libmca_ras_alps_la_LDFLAGS = -module -avoid-version $(ras_alps_LDFLAGS)
+libmca_ras_alps_la_LIBADD = $(ras_alps_LIBS)

orte/mca/ras/alps/configure.m4

@@ -0,0 +1,25 @@
+# -*- shell-script -*-
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+# MCA_ras_alps_CONFIG([action-if-found], [action-if-not-found])
+# -----------------------------------------------------------
+AC_DEFUN([MCA_ras_alps_CONFIG],[
+	OMPI_CHECK_ALPS([ras_alps], [$1], [$2])
+	])dnl

orte/mca/ras/alps/configure.params

@@ -0,0 +1,23 @@
+# -*- shell-script -*-
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2007      Los Alamos National Security, LLC.  All rights
+#                         reserved. 
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+PARAM_CONFIG_FILES="Makefile"

orte/mca/ras/alps/help-ras-alps.txt

@@ -0,0 +1,43 @@
+# -*- text -*-
+#
+# Copyright (c) 2004-2007 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+# This is the US/English help file for Open MPI MCA error messages.
+#
+[alps-env-var-not-found]
+While trying to determine what resources are available, the ALPS
+resource allocator expects to find the following environment variables:
+
+	 BATCH_PARTITION_ID
+    
+However, it was unable to find the following environment variable:
+
+    %s
+
+#This is a fatal error.
+[alps-env-var-bad-value]
+While trying to determine what resources are available, the ALPS
+resource allocator uses the following environment variables:
+
+    ALPS_NODELIST       value: %s
+    ALPS_TASKS_PER_NODE value: %s
+    
+However, an error was encountered when trying to parse the following variable:
+
+    %s
+
+#This is a fatal error.

orte/mca/ras/alps/ras_alps.h

@@ -0,0 +1,41 @@
+/*
+ * Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ */
+/**
+ * @file
+ *
+ * Resource Allocation (ALPS)
+ */
+#ifndef ORTE_RAS_ALPS_H
+#define ORTE_RAS_ALPS_H
+
+#include "orte/mca/ras/ras.h"
+#include "orte/mca/ras/base/base.h"
+
+#if defined(c_plusplus) || defined(__cplusplus)
+extern "C" {
+#endif
+
+    ORTE_DECLSPEC extern orte_ras_base_component_t mca_ras_alps_component;
+    ORTE_DECLSPEC extern orte_ras_base_module_t orte_ras_alps_module;
+
+#if defined(c_plusplus) || defined(__cplusplus)
+}
+#endif
+
+#endif

orte/mca/ras/alps/ras_alps_component.c

@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2004-2007 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ */
+
+#include "orte_config.h"
+
+#include "opal/util/output.h"
+#include "opal/mca/base/base.h"
+#include "opal/mca/base/mca_base_param.h"
+#include "orte/orte_constants.h"
+#include "orte/util/proc_info.h"
+#include "ras_alps.h"
+
+
+/*
+ * Local variables
+ */
+static int param_priority;
+
+
+/*
+ * Local functions
+ */
+static int ras_alps_open(void);
+static orte_ras_base_module_t *ras_alps_init(int*);
+
+
+orte_ras_base_component_t mca_ras_alps_component = {
+    /* First, the mca_base_component_t struct containing meta
+       information about the component itself */
+
+    {
+        /* Indicate that we are a ras v1.3.0 component (which also
+           implies a specific MCA version) */
+        
+        ORTE_RAS_BASE_VERSION_1_3_0,
+        
+        /* Component name and version */
+        
+        "alps",
+        ORTE_MAJOR_VERSION,
+        ORTE_MINOR_VERSION,
+        ORTE_RELEASE_VERSION,
+        
+        /* Component open and close functions */
+        
+        ras_alps_open,
+        NULL
+    },
+    
+    /* Next the MCA v1.0.0 component meta data */
+    {
+        /* The component is checkpoint ready */
+        MCA_BASE_METADATA_PARAM_CHECKPOINT
+    },
+    
+    ras_alps_init
+};
+
+
+static int ras_alps_open(void)
+{
+    param_priority = 
+        mca_base_param_reg_int(&mca_ras_alps_component.ras_version,
+                               "priority",
+                               "Priority of the alps ras component",
+                               false, false, 75, NULL);
+
+    return ORTE_SUCCESS;
+}
+
+
+static orte_ras_base_module_t *ras_alps_init(int* priority)
+{
+    /* if we are not an HNP, then we must not be selected */
+    if (!orte_process_info.seed) {
+        return NULL;
+    }
+    
+    /* Are we running under a ALPS job? */
+
+    if (NULL != getenv("BATCH_PARTITION_ID")) {
+        mca_base_param_lookup_int(param_priority, priority);
+        opal_output(orte_ras_base.ras_output,
+                    "ras:alps: available for selection");
+        return &orte_ras_alps_module;
+    }
+
+    /* Sadly, no */
+
+    opal_output(orte_ras_base.ras_output,
+                "ras:alps: NOT available for selection");
+    return NULL;
+}

orte/mca/ras/alps/ras_alps_module.c

@@ -0,0 +1,140 @@
+/*
+ * Copyright (c) 2004-2007 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ */
+#include "orte_config.h"
+#include "orte/orte_constants.h"
+#include "orte/orte_types.h"
+
+#include <unistd.h>
+#include <string.h>
+#include <ctype.h>
+
+#include "opal/util/argv.h"
+#include "opal/util/output.h"
+#include "opal/util/show_help.h"
+#include "orte/dss/dss.h"
+#include "orte/mca/rmgr/rmgr.h"
+#include "orte/mca/errmgr/errmgr.h"
+
+#include "orte/mca/ras/base/ras_private.h"
+#include "ras_alps.h"
+
+
+/*
+ * Local functions
+ */
+static int orte_ras_alps_allocate(orte_jobid_t jobid, opal_list_t *attributes);
+static int orte_ras_alps_deallocate(orte_jobid_t jobid);
+static int orte_ras_alps_finalize(void);
+
+
+
+
+/*
+ * Global variable
+ */
+orte_ras_base_module_t orte_ras_alps_module = {
+    orte_ras_alps_allocate,
+    orte_ras_base_node_insert,
+    orte_ras_base_node_query,
+    orte_ras_base_node_query_alloc,
+    orte_ras_base_node_lookup,
+    orte_ras_base_proc_query_alloc,
+    orte_ras_alps_deallocate,
+    orte_ras_alps_finalize
+};
+
+/**
+ * Discover available (pre-allocated) nodes.  Allocate the
+ * requested number of nodes/process slots to the job.
+ *  
+ */
+static int orte_ras_alps_allocate(orte_jobid_t jobid, opal_list_t *attributes)
+{
+    int ret;
+    char *alps_batch_id;
+    opal_list_t nodes;
+    opal_list_item_t* item;
+    
+    char *alps_node_cmd_str = "apstat -a `apstat -r | grep $BATCH_PARTITION_ID  | awk '{print $2}'` " 
+        " -r   -v | egrep  \"(nid [0-9]+)\" -o | awk '{print $2}' > ./ompi_ras_alps_node_file";
+            
+    
+    alps_batch_id = getenv("BATCH_PARTITION_ID");
+    if (NULL == alps_batch_id) {
+        opal_show_help("help-ras-alps.txt", "alps-env-var-not-found", 1,
+                       "BATCH_PARTITION_ID");
+        return ORTE_ERR_NOT_FOUND;
+    }
+
+    if(system(alps_node_cmd_str)) { 
+        opal_output(0, "Error in orte_ras_alps_allocate: system call returned an error, for reference I tried to run: %s", 
+                    alps_node_cmd_str);
+        return ORTE_ERROR;
+    }
+
+    OBJ_CONSTRUCT(&nodes, opal_list_t);
+
+    if (ORTE_SUCCESS != (ret = orte_ras_base_read_nodename_file(&nodes, "./ompi_ras_alps_node_file"))) {
+        ORTE_ERROR_LOG(ret);
+        goto cleanup;
+    }
+
+    ret = orte_ras_base_allocate_nodes(jobid, &nodes);
+
+    ret = orte_ras_base_node_insert(&nodes);
+
+cleanup:
+    while (NULL != (item = opal_list_remove_first(&nodes))) {
+        OBJ_RELEASE(item);
+    }
+    OBJ_DESTRUCT(&nodes);
+
+    /* All done */
+
+    if (ORTE_SUCCESS == ret) {
+        opal_output(orte_ras_base.ras_output, 
+                    "ras:alps:allocate: success");
+    } else {
+        opal_output(orte_ras_base.ras_output, 
+                    "ras:alps:allocate: failure (base_allocate_nodes=%d)", ret);
+    }
+    return ret;
+}
+
+/*
+ * There's really nothing to do here
+ */
+static int orte_ras_alps_deallocate(orte_jobid_t jobid)
+{
+    opal_output(orte_ras_base.ras_output, 
+                "ras:alps:deallocate: success (nothing to do)");
+    return ORTE_SUCCESS;
+}
+
+
+/*
+ * There's really nothing to do here
+ */
+static int orte_ras_alps_finalize(void)
+{
+    opal_output(orte_ras_base.ras_output, 
+                "ras:alps:finalize: success (nothing to do)");
+    return ORTE_SUCCESS;
+}
+

orte/mca/ras/base/ras_base_support_fns.c

@@ -9,6 +9,7 @@
  *                         University of Stuttgart.  All rights reserved.
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
  * $COPYRIGHT$
  * 
  * Additional copyrights may follow
@@ -118,3 +119,65 @@ CLEANUP:
     
     return ORTE_SUCCESS;
 }
+
+#define RAS_BASE_FILE_MAX_LINE_LENGTH   512
+
+static char *ras_getline(FILE *fp)
+{
+    char *ret, *buff = NULL;
+    char input[RAS_BASE_FILE_MAX_LINE_LENGTH];
+    
+    ret = fgets(input, RAS_BASE_FILE_MAX_LINE_LENGTH, fp);
+    if (NULL != ret) {
+        input[strlen(input)-1] = '\0';  /* remove newline */
+        buff = strdup(input);
+    }
+    
+    return buff;
+}
+    
+int orte_ras_base_read_nodename_file(opal_list_t *nodes, char *filename)
+{
+    FILE *fp;
+    int32_t nodeid=0;
+    orte_ras_node_t *node=NULL;
+    char *hostname;
+    
+    fp = fopen(filename, "r");
+    if (NULL == fp) {
+        ORTE_ERROR_LOG(ORTE_ERR_FILE_OPEN_FAILURE);
+        return ORTE_ERR_FILE_OPEN_FAILURE;
+    }
+    
+    while (NULL != (hostname = ras_getline(fp))) {
+        opal_output(orte_ras_base.ras_output, 
+                    "ras:base:read_nodename: got hostname %s", hostname);
+        
+        /* if this matches the prior nodename, then just add
+         * to the slot count
+         */
+        if (NULL != node &&
+            0 == strcmp(node->node_name, hostname)) {
+            ++node->node_slots;
+            /* free the hostname that came back since we don't need it */
+            free(hostname);
+            continue;
+        }
+        
+        /* must be a new name, so add a new item to the list */
+        opal_output(orte_ras_base.ras_output, 
+                    "ras:base:read_nodename: not found -- added to list");
+        node = OBJ_NEW(orte_ras_node_t);
+        node->node_name = hostname;
+        node->launch_id = nodeid;
+        node->node_slots_inuse = 0;
+        node->node_slots_max = 0;
+        node->node_slots = 1;
+        opal_list_append(nodes, &node->super);
+        /* up the nodeid */
+        nodeid++;
+    }
+    fclose(fp);
+    
+    return ORTE_SUCCESS;
+}

orte/mca/ras/base/ras_private.h

@@ -9,6 +9,7 @@
  *                         University of Stuttgart.  All rights reserved.
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
  * $COPYRIGHT$
  * 
  * Additional copyrights may follow
@@ -86,6 +87,8 @@ ORTE_DECLSPEC int orte_ras_base_set_oversubscribe_override(orte_jobid_t job);
 
 ORTE_DECLSPEC int orte_ras_base_get_oversubscribe_override(orte_jobid_t job, bool *flag);
 
+ORTE_DECLSPEC int orte_ras_base_read_nodename_file(opal_list_t *nodes, char *filename);
+
 /*
  * Query the registry for all available nodes 
  */

orte/mca/sds/alps/Makefile.am

@@ -0,0 +1,51 @@
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+# Use the top-level Makefile.options
+
+
+
+sources = \
+        sds_alps.h \
+        sds_alps_component.c \
+        sds_alps_module.c
+
+# Make the output library in this directory, and name it either
+# mca_<type>_<name>.la (for DSO builds) or libmca_<type>_<name>.la
+# (for static builds).
+
+if OMPI_BUILD_sds_alps_DSO
+component_noinst =
+component_install = mca_sds_alps.la
+else
+component_noinst = libmca_sds_alps.la
+component_install =
+endif
+
+mcacomponentdir = $(pkglibdir)
+mcacomponent_LTLIBRARIES = $(component_install)
+mca_sds_alps_la_SOURCES = $(sources)
+mca_sds_alps_la_LDFLAGS = -module -avoid-version
+mca_sds_alps_la_LIBADD = \
+    $(top_ompi_builddir)/orte/libopen-rte.la \
+    $(top_ompi_builddir)/opal/libopen-pal.la
+
+noinst_LTLIBRARIES = $(component_noinst)
+libmca_sds_alps_la_SOURCES =$(sources)
+libmca_sds_alps_la_LDFLAGS = -module -avoid-version

orte/mca/sds/alps/configure.m4

@@ -0,0 +1,27 @@
+# -*- shell-script -*-
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+# MCA_sds_alps_CONFIG([action-if-found], [action-if-not-found])
+# -----------------------------------------------------------
+AC_DEFUN([MCA_sds_alps_CONFIG],[
+	OMPI_CHECK_ALPS([sds_alps],
+	    [AC_CHECK_FUNC([cnos_get_rank], [$1], [$2])],
+	    [$2])
+])dnl

orte/mca/sds/alps/configure.params

@@ -0,0 +1,23 @@
+# -*- shell-script -*-
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2007      Los Alamos National Security, LLC.  All rights
+#                         reserved. 
+# Copyright (c) 2008      UT-Battelle, LLC
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+PARAM_CONFIG_FILES="Makefile"

orte/mca/sds/alps/sds_alps.h

@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ */
+
+#ifndef ORTE_SDS_ALPS_H
+#define ORTE_SDS_ALPS_H
+
+#if defined(c_plusplus) || defined(__cplusplus)
+extern "C" {
+#endif
+
+    /*
+     * Module open / close
+     */
+    int orte_sds_alps_component_open(void);
+    int orte_sds_alps_component_close(void);
+    orte_sds_base_module_t* orte_sds_alps_component_init(int *priority);
+
+    /*
+     * Startup / Shutdown
+     */
+    int orte_sds_alps_finalize(void);
+
+    /*
+     * Module functions
+     */
+    int orte_sds_alps_set_name(void);
+    int orte_sds_alps_contact_universe(void);
+
+
+#if defined(c_plusplus) || defined(__cplusplus)
+}
+#endif
+
+#endif /* ORTE_SDS_ALPS_H */

orte/mca/sds/alps/sds_alps_component.c

@@ -0,0 +1,106 @@
+/*
+ * Copyright (c) 2004-2007 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ *
+ * These symbols are in a file by themselves to provide nice linker
+ * semantics.  Since linkers generally pull in symbols by object
+ * files, keeping these symbols as the only symbols in this file
+ * prevents utility programs such as "ompi_info" from having to import
+ * entire components just to query their version and parameters.
+ */
+
+#include "orte_config.h"
+
+#include "orte/orte_constants.h"
+#include "orte/mca/sds/sds.h"
+#include "orte/mca/sds/alps/sds_alps.h"
+#include "opal/mca/base/mca_base_param.h"
+
+extern orte_sds_base_module_t orte_sds_alps_module;
+
+/*
+ * Instantiate the public struct with all of our public information
+ * and pointers to our public functions in it
+ */
+orte_sds_base_component_t mca_sds_alps_component = {
+    /* First, the mca_component_t struct containing meta information
+       about the component itself */
+    {
+        /* Indicate that we are a sds v1.0.0 component (which also
+           implies a specific MCA version) */
+        ORTE_SDS_BASE_VERSION_1_0_0,
+
+        /* Component name and version */
+        "alps",
+        ORTE_MAJOR_VERSION,
+        ORTE_MINOR_VERSION,
+        ORTE_RELEASE_VERSION,
+
+        /* Component open and close functions */
+        orte_sds_alps_component_open,
+        orte_sds_alps_component_close
+    },
+
+    /* Next the MCA v1.0.0 component meta data */
+    {
+        /* The component is not checkpoint ready */
+        MCA_BASE_METADATA_PARAM_NONE
+    },
+
+    /* Initialization / querying functions */
+    orte_sds_alps_component_init
+};
+
+
+int
+orte_sds_alps_component_open(void)
+{
+    return ORTE_SUCCESS;
+}
+
+
+orte_sds_base_module_t *
+orte_sds_alps_component_init(int *priority)
+{
+    int id;
+    char *mode;
+
+    /* okay, not seed/singleton attempt another approach */
+    id = mca_base_param_register_string("ns", "nds", NULL, NULL, NULL);
+    mca_base_param_lookup_string(id, &mode);
+    
+    if (NULL == mode || 0 != strcmp("alps", mode)) {
+        if (NULL != mode) {
+            free(mode);
+        }
+        return NULL; 
+    }
+    if (NULL != mode) {
+        free(mode);
+    }
+
+    *priority = 35;
+    return &orte_sds_alps_module;
+}
+
+
+int
+orte_sds_alps_component_close(void)
+{
+    return ORTE_SUCCESS;
+}
+

orte/mca/sds/alps/sds_alps_module.c

@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ *
+ */
+
+#include "orte_config.h"
+
+#include <catamount/cnos_mpi_os.h>
+
+#include "orte/orte_constants.h"
+#include "orte/mca/sds/sds.h"
+#include "orte/mca/sds/base/base.h"
+#include "orte/mca/sds/alps/sds_alps.h"
+#include "orte/util/proc_info.h"
+#include "orte/mca/ns/base/base.h"
+#include "orte/mca/errmgr/base/base.h"
+
+orte_sds_base_module_t orte_sds_alps_module = {
+    orte_sds_base_basic_contact_universe,
+    orte_sds_alps_set_name,
+    orte_sds_alps_finalize,
+};
+
+int
+orte_sds_alps_set_name(void)
+{
+    int rc;
+    orte_jobid_t jobid;
+    orte_vpid_t vpid;
+
+    if(orte_process_info.seed) { 
+        if (ORTE_SUCCESS != (rc = orte_ns.create_my_name())) {
+            ORTE_ERROR_LOG(rc);
+            return rc;
+        }
+        
+        orte_process_info.num_procs = 1;
+
+        return rc;
+    }
+
+    /* Get our process information
+     *
+     * we're going to make up the jobid.  find our vpid,
+     * assuming range starts at 0 
+     */
+    jobid = 0; /* not 0, since it has special meaning */
+    
+
+    vpid = (orte_vpid_t) cnos_get_rank() + 1;
+    
+    if (ORTE_SUCCESS != (rc = orte_ns.create_process_name(&(orte_process_info.my_name),
+                                                          jobid,
+                                                          vpid))) {
+        ORTE_ERROR_LOG(rc);
+        return rc;
+    }
+    
+    orte_process_info.num_procs = (orte_std_cntr_t) cnos_get_size();
+
+    return ORTE_SUCCESS;
+}
+
+
+int 
+orte_sds_alps_finalize(void)
+{
+    return ORTE_SUCCESS;
+}

orte/mca/sds/cnos/sds_cnos_component.c

@@ -9,6 +9,7 @@
  *                         University of Stuttgart.  All rights reserved.
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
+ * Copyright (c) 2008      UT-Battelle, LLC
  * $COPYRIGHT$
  * 
  * Additional copyrights may follow
@@ -89,7 +90,7 @@ orte_sds_cnos_component_init(int *priority)
         return NULL;
     }
 
-    *priority = 60;
+    *priority = 30;
     return &orte_sds_cnos_module;
 }