Add a new ras module to support bootstrap operations. Additional functionality may eventually be required in the component, but for now all it does is provide a mechanism for ensuring that other allocations don't confuse the system.

Only active if specifically directed to use it

This commit was SVN r22040.

orte/include/orte/constants.h

@@ -99,7 +99,8 @@ enum {
     ORTE_ERR_FILE_NOT_EXECUTABLE            = (ORTE_ERR_BASE - 27),
     ORTE_ERR_HNP_COULD_NOT_START            = (ORTE_ERR_BASE - 28),
     ORTE_ERR_SYS_LIMITS_SOCKETS             = (ORTE_ERR_BASE - 29),
-    ORTE_ERR_SOCKET_NOT_AVAILABLE           = (ORTE_ERR_BASE - 30)
+    ORTE_ERR_SOCKET_NOT_AVAILABLE           = (ORTE_ERR_BASE - 30),
+    ORTE_ERR_SYSTEM_WILL_BOOTSTRAP          = (ORTE_ERR_BASE - 31)
 };
 
 #define ORTE_ERR_MAX                      (ORTE_ERR_BASE - 100)

orte/mca/ras/base/ras_base_allocate.c

@@ -137,6 +137,15 @@ int orte_ras_base_allocate(orte_job_t *jdata)
     if (NULL != orte_ras_base.active_module)  {
         /* read the allocation */
         if (ORTE_SUCCESS != (rc = orte_ras_base.active_module->allocate(&nodes))) {
+            if (ORTE_ERR_SYSTEM_WILL_BOOTSTRAP == rc) {
+                /* this module indicates that nodes will be discovered
+                 * on a bootstrap basis, so there is nothing more
+                 * for us to do
+                 */
+                OBJ_DESTRUCT(&nodes);
+                rc = ORTE_SUCCESS;
+                goto DISPLAY;
+            }
             ORTE_ERROR_LOG(rc);
             OBJ_DESTRUCT(&nodes);
             return rc;

orte/mca/ras/cm/Makefile.am

@@ -0,0 +1,39 @@
+#
+# Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved. 
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+sources = \
+        ras_cm.h \
+        ras_cm_component.c \
+        ras_cm_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_cm_DSO
+lib =
+lib_sources = 
+component = mca_ras_cm.la
+component_sources = $(sources)
+else
+lib = libmca_ras_cm.la
+lib_sources = $(sources)
+component =
+component_sources =
+endif
+
+mcacomponentdir = $(pkglibdir)
+mcacomponent_LTLIBRARIES = $(component)
+mca_ras_cm_la_SOURCES = $(component_sources)
+mca_ras_cm_la_LDFLAGS = -module -avoid-version
+
+noinst_LTLIBRARIES = $(lib)
+libmca_ras_cm_la_SOURCES = $(lib_sources)
+libmca_ras_cm_la_LDFLAGS = -module -avoid-version

orte/mca/ras/cm/configure.params

@@ -0,0 +1,11 @@
+# -*- shell-script -*-
+#
+# Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved. 
+# $COPYRIGHT$
+# 
+# Additional copyrights may follow
+# 
+# $HEADER$
+#
+
+PARAM_CONFIG_FILES="Makefile"

orte/mca/ras/cm/ras_cm.h

@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved. 
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ */
+/**
+ * @file
+ *
+ */
+#ifndef ORTE_RAS_CM_H
+#define ORTE_RAS_CM_H
+
+#include "orte_config.h"
+#include "orte/mca/ras/ras.h"
+#include "orte/mca/ras/base/base.h"
+
+BEGIN_C_DECLS
+
+ORTE_DECLSPEC extern orte_ras_base_component_t mca_ras_cm_component;
+ORTE_DECLSPEC extern orte_ras_base_module_t orte_ras_cm_module;
+
+END_C_DECLS
+
+#endif

orte/mca/ras/cm/ras_cm_component.c

@@ -0,0 +1,73 @@
+/*
+ * Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved. 
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ */
+
+#include "orte_config.h"
+#include "orte/constants.h"
+
+#include "opal/mca/base/base.h"
+
+#include "ras_cm.h"
+
+
+/*
+ * Local functions
+ */
+static int ras_cm_open(void);
+static int orte_ras_cm_component_query(mca_base_module_t **module, int *priority);
+
+
+orte_ras_base_component_t mca_ras_cm_component = {
+    /* First, the mca_base_component_t struct containing meta
+       information about the component itself */
+
+    {
+        ORTE_RAS_BASE_VERSION_2_0_0,
+        
+        /* Component name and version */
+        "cm",
+        ORTE_MAJOR_VERSION,
+        ORTE_MINOR_VERSION,
+        ORTE_RELEASE_VERSION,
+        
+        /* Component open and close functions */
+        ras_cm_open,
+        NULL,
+        orte_ras_cm_component_query
+    },
+    {
+        /* The component is checkpoint ready */
+        MCA_BASE_METADATA_PARAM_CHECKPOINT
+    }
+};
+
+
+static int ras_cm_open(void)
+{
+    return ORTE_SUCCESS;
+}
+
+
+static int orte_ras_cm_component_query(mca_base_module_t **module, int *priority)
+{
+    char *directive;
+    
+    /* determine if we were specified */
+    directive = getenv("OMPI_MCA_ras");
+    
+    if (NULL == directive || 0 != strcmp("cm", directive)) {
+        *priority = 0;
+        *module = NULL;
+        return ORTE_ERROR;
+    }
+    
+    /* we were specified */
+    *priority = 100;
+    *module = (mca_base_module_t *) &orte_ras_cm_module;
+    return ORTE_SUCCESS;
+}

orte/mca/ras/cm/ras_cm_module.c

@@ -0,0 +1,60 @@
+/*
+ * Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved. 
+ * $COPYRIGHT$
+ * 
+ * Additional copyrights may follow
+ * 
+ * $HEADER$
+ */
+#include "orte_config.h"
+#include "orte/constants.h"
+#include "orte/types.h"
+
+#include "opal/class/opal_list.h"
+#include "opal/util/output.h"
+
+#include "orte/util/name_fns.h"
+#include "orte/runtime/orte_globals.h"
+
+#include "ras_cm.h"
+
+
+/*
+ * Local functions
+ */
+static int allocate(opal_list_t *nodes);
+static int finalize(void);
+
+
+/*
+ * Module APIs
+ */
+orte_ras_base_module_t orte_ras_cm_module = {
+    allocate,
+    finalize
+};
+
+/**
+ * Since the system will be bootstrapping, there is
+ * nothing to do here
+ */
+static int allocate(opal_list_t *nodes)
+{
+    OPAL_OUTPUT_VERBOSE((1, orte_ras_base.ras_output,
+                         "%s ras:cm:allocate: success",
+                         ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
+    /* indicate that nodes will be discovered via bootstrap */
+    return ORTE_ERR_SYSTEM_WILL_BOOTSTRAP;
+}
+
+/*
+ * There's really nothing to do here
+ */
+static int finalize(void)
+{
+    
+    OPAL_OUTPUT_VERBOSE((1, orte_ras_base.ras_output,
+                         "%s ras:cm:finalize: success (nothing to do)",
+                         ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
+    return ORTE_SUCCESS;
+}

orte/mca/rmaps/base/rmaps_base_support_fns.c

@@ -267,10 +267,11 @@ PROCESS:
      * in the mapper
      */
     OPAL_OUTPUT_VERBOSE((5, orte_rmaps_base.rmaps_output,
-                         "%s rmaps:base: mapping proc for job %s to node %s",
+                         "%s rmaps:base: mapping proc for job %s to node %s whose daemon is %s",
                          ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
                          ORTE_JOBID_PRINT(proc->name.jobid),
-                         (NULL == node->name) ? "NULL" : node->name));
+                         (NULL == node->name) ? "NULL" : node->name,
+                         (NULL == node->daemon) ? "NULL" : ORTE_NAME_PRINT(&(node->daemon->name))));
     
     if (0 > (rc = opal_pointer_array_add(node->procs, (void*)proc))) {
         ORTE_ERROR_LOG(rc);

orte/util/error_strings.c

@@ -121,6 +121,9 @@ orte_err2str(int errnum)
     case ORTE_ERR_SOCKET_NOT_AVAILABLE:
         retval = "Unable to open a TCP socket for out-of-band communications";
         break;
+    case ORTE_ERR_SYSTEM_WILL_BOOTSTRAP:
+        retval = "System will determine resources during bootstrap of daemons";
+        break;
     default:
         retval = NULL;
     }