Merge pull request #3259 from rhc54/topic/launch

Update how we pass the node regex so we pass _all_ nodes, even those without daemons.

Author: Ralph Castain

opal/mca/pmix/pmix2x/pmix/src/server/pmix_server.c

@@ -990,11 +990,11 @@ static void _dmodex_req(int sd, short args, void *cbdata)
      * may not be a contribution */
     if (PMIX_SUCCESS == (rc = pmix_hash_fetch(&nptr->server->myremote, info->rank, "modex", &val)) &&
         NULL != val) {
-    data = val->data.bo.bytes;
-    sz = val->data.bo.size;
-    /* protect the data */
-    val->data.bo.bytes = NULL;
-    val->data.bo.size = 0;
+        data = val->data.bo.bytes;
+        sz = val->data.bo.size;
+        /* protect the data */
+        val->data.bo.bytes = NULL;
+        val->data.bo.size = 0;
         PMIX_VALUE_RELEASE(val);
     }
 

opal/mca/pmix/pmix2x/pmix/src/server/pmix_server_ops.c

@@ -138,7 +138,6 @@ pmix_status_t pmix_server_commit(pmix_peer_t *peer, pmix_buffer_t *buf)
     pmix_nspace_t *nptr;
     pmix_rank_info_t *info;
     pmix_dmdx_remote_t *dcd, *dcdnext;
-    pmix_buffer_t *pbkt;
     pmix_value_t *val;
     char *data;
     size_t sz;
@@ -236,16 +235,19 @@ pmix_status_t pmix_server_commit(pmix_peer_t *peer, pmix_buffer_t *buf)
         if (dcd->cd->proc.rank == info->rank) {
            /* we can now fulfill this request - collect the
              * remote/global data from this proc */
-            pbkt = PMIX_NEW(pmix_buffer_t);
             /* get any remote contribution - note that there
              * may not be a contribution */
+            data = NULL;
+            sz = 0;
             if (PMIX_SUCCESS == pmix_hash_fetch(&nptr->server->myremote, info->rank, "modex", &val) &&
                 NULL != val) {
-                PMIX_LOAD_BUFFER(pbkt, val->data.bo.bytes, val->data.bo.size);
+                data = val->data.bo.bytes;
+                sz = val->data.bo.size;
+                /* protect the data */
+                val->data.bo.bytes = NULL;
+                val->data.bo.size = 0;
                 PMIX_VALUE_RELEASE(val);
             }
-            PMIX_UNLOAD_BUFFER(pbkt, data, sz);
-            PMIX_RELEASE(pbkt);
             /* execute the callback */
             dcd->cd->cbfunc(PMIX_SUCCESS, data, sz, dcd->cd->cbdata);
             if (NULL != data) {

opal/mca/pmix/pmix2x/pmix/src/util/compress.h

@@ -9,7 +9,7 @@
  *                         University of Stuttgart.  All rights reserved.
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
- * Copyright (c) 2015-2016 Intel, Inc.  All rights reserved.
+ * Copyright (c) 2015-2017 Intel, Inc. All rights reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -31,8 +31,8 @@
 
 BEGIN_C_DECLS
 
-/* define a limit for storing raw strings */
-#define PMIX_STRING_LIMIT  512
+/* define a limit of 128k for raw strings */
+#define PMIX_STRING_LIMIT  131072
 
 /* define a macro for quickly checking if a string exceeds the
  * compression limit */

orte/mca/ess/alps/ess_alps_module.c

@@ -12,6 +12,7 @@
  * Copyright (c) 2011      Cisco Systems, Inc.  All rights reserved.
  * Copyright (c) 2011-2013 Los Alamos National Security, LLC.
  *                         All rights reserved.
+ * Copyright (c) 2017      Intel, Inc. All rights reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -58,7 +59,6 @@ static int rte_init(void)
 {
     int ret;
     char *error = NULL;
-    char **hosts = NULL;
 
     OPAL_OUTPUT_VERBOSE((1, orte_ess_base_framework.framework_output,
                          "ess:alps in rte_init"));
@@ -90,23 +90,11 @@ static int rte_init(void)
      * default procedure
      */
     if (ORTE_PROC_IS_DAEMON) {
-        if (NULL != orte_node_regex) {
-            /* extract the nodes */
-            if (ORTE_SUCCESS != (ret =
-                orte_regex_extract_node_names(orte_node_regex, &hosts)) ||
-                NULL == hosts) {
-                error = "orte_regex_extract_node_names";
-                goto fn_fail;
-            }
-        }
-        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
+        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup())) {
             ORTE_ERROR_LOG(ret);
             error = "orte_ess_base_orted_setup";
             goto fn_fail;
         }
-        if (NULL != hosts) {
-            opal_argv_free(hosts);
-        }
 
         /*
          * now synchronize with aprun.

orte/mca/ess/base/base.h

@@ -12,7 +12,7 @@
  * Copyright (c) 2011-2012 Cisco Systems, Inc.  All rights reserved.
  * Copyright (c) 2012      Oak Ridge National Labs.  All rights reserved.
  * Copyright (c) 2013      Los Alamos National Security, LLC.  All rights reserved.
- * Copyright (c) 2013      Intel, Inc. All rights reserved.
+ * Copyright (c) 2013-2017 Intel, Inc. All rights reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -67,7 +67,7 @@ ORTE_DECLSPEC void orte_ess_base_app_abort(int status, bool report);
 ORTE_DECLSPEC int orte_ess_base_tool_setup(void);
 ORTE_DECLSPEC int orte_ess_base_tool_finalize(void);
 
-ORTE_DECLSPEC int orte_ess_base_orted_setup(char **hosts);
+ORTE_DECLSPEC int orte_ess_base_orted_setup(void);
 ORTE_DECLSPEC int orte_ess_base_orted_finalize(void);
 
 /* Detect whether or not this proc is bound - if not,

orte/mca/ess/base/ess_base_std_orted.c

@@ -103,7 +103,7 @@ static void setup_sighandler(int signal, opal_event_t *ev,
 }
 
 
-int orte_ess_base_orted_setup(char **hosts)
+int orte_ess_base_orted_setup(void)
 {
     int ret = ORTE_ERROR;
     int fd;
@@ -113,7 +113,6 @@ int orte_ess_base_orted_setup(char **hosts)
     orte_job_t *jdata;
     orte_proc_t *proc;
     orte_app_context_t *app;
-    orte_node_t *node;
     char *param;
     hwloc_obj_t obj;
     unsigned i, j;
@@ -218,12 +217,9 @@ int orte_ess_base_orted_setup(char **hosts)
      * a specific module to use
      */
     (void) mca_base_var_env_name("plm", &param);
-
     plm_in_use = !!(getenv(param));
     free (param);
-
     if (plm_in_use)  {
-
         if (ORTE_SUCCESS != (ret = mca_base_framework_open(&orte_plm_base_framework, 0))) {
             ORTE_ERROR_LOG(ret);
             error = "orte_plm_base_open";
@@ -332,11 +328,6 @@ int orte_ess_base_orted_setup(char **hosts)
     app = OBJ_NEW(orte_app_context_t);
     opal_pointer_array_set_item(jdata->apps, 0, app);
     jdata->num_apps++;
-    /* create and store a node object where we are */
-    node = OBJ_NEW(orte_node_t);
-    node->name = strdup(orte_process_info.nodename);
-    node->index = ORTE_PROC_MY_NAME->vpid;
-    opal_pointer_array_set_item(orte_node_pool, ORTE_PROC_MY_NAME->vpid, node);
 
     /* create and store a proc object for us */
     proc = OBJ_NEW(orte_proc_t);
@@ -345,19 +336,6 @@ int orte_ess_base_orted_setup(char **hosts)
     proc->pid = orte_process_info.pid;
     proc->state = ORTE_PROC_STATE_RUNNING;
     opal_pointer_array_set_item(jdata->procs, proc->name.vpid, proc);
-    /* record that the daemon (i.e., us) is on this node
-     * NOTE: we do not add the proc object to the node's
-     * proc array because we are not an application proc.
-     * Instead, we record it in the daemon field of the
-     * node object
-     */
-    OBJ_RETAIN(proc);   /* keep accounting straight */
-    node->daemon = proc;
-    ORTE_FLAG_SET(node, ORTE_NODE_FLAG_DAEMON_LAUNCHED);
-    node->state = ORTE_NODE_STATE_UP;
-    /* now point our proc node field to the node */
-    OBJ_RETAIN(node);   /* keep accounting straight */
-    proc->node = node;
     /* record that the daemon job is running */
     jdata->num_procs = 1;
     jdata->state = ORTE_JOB_STATE_RUNNING;
@@ -514,7 +492,6 @@ int orte_ess_base_orted_setup(char **hosts)
     orte_topo_signature = opal_hwloc_base_get_topo_signature(opal_hwloc_topology);
     t->sig = strdup(orte_topo_signature);
     opal_pointer_array_add(orte_node_topologies, t);
-    node->topology = t;
     if (15 < opal_output_get_verbosity(orte_ess_base_framework.framework_output)) {
         opal_output(0, "%s Topology Info:", ORTE_NAME_PRINT(ORTE_PROC_MY_NAME));
         opal_dss.dump(0, opal_hwloc_topology, OPAL_HWLOC_TOPO);
@@ -526,12 +503,25 @@ int orte_ess_base_orted_setup(char **hosts)
      * after we enable_comm as that function determines our
      * own port, which we need in order to construct the nidmap
      */
-    if (NULL != hosts) {
+    if (NULL != orte_node_regex) {
+        if (ORTE_SUCCESS != (ret = orte_util_nidmap_parse(orte_node_regex))) {
+            ORTE_ERROR_LOG(ret);
+            error = "construct nidmap";
+            goto error;
+        }
+    }
+
+    if (orte_static_ports) {
+        if (NULL == orte_node_regex) {
+            /* we didn't get the node info */
+            error = "cannot construct daemon map for static ports - no node map info";
+            goto error;
+        }
         /* extract the node info from the environment and
          * build a nidmap from it - this will update the
          * routing plan as well
          */
-        if (ORTE_SUCCESS != (ret = orte_util_build_daemon_nidmap(hosts))) {
+        if (ORTE_SUCCESS != (ret = orte_util_build_daemon_nidmap())) {
             ORTE_ERROR_LOG(ret);
             error = "construct daemon map from static ports";
             goto error;
@@ -635,6 +625,7 @@ int orte_ess_base_orted_setup(char **hosts)
     }
 
     return ORTE_SUCCESS;
+
  error:
     orte_show_help("help-orte-runtime.txt",
                    "orte_init:startup:internal-failure",

orte/mca/ess/env/ess_env_module.c

@@ -10,7 +10,7 @@
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
  * Copyright (c) 2011-2012 Cisco Systems, Inc.  All rights reserved.
- * Copyright (c) 2013-2015 Intel, Inc. All rights reserved.
+ * Copyright (c) 2013-2017 Intel, Inc. All rights reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -98,7 +98,6 @@ static int rte_init(void)
 {
     int ret;
     char *error = NULL;
-    char **hosts = NULL;
 
     /* run the prolog */
     if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
@@ -112,19 +111,11 @@ static int rte_init(void)
     /* if I am a daemon, complete my setup using the
      * default procedure
      */
-    if (NULL != orte_node_regex) {
-        /* extract the nodes */
-        if (ORTE_SUCCESS != (ret = orte_regex_extract_node_names(orte_node_regex, &hosts))) {
-            error = "orte_regex_extract_node_names";
-            goto error;
-        }
-    }
-    if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
+    if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup())) {
         ORTE_ERROR_LOG(ret);
         error = "orte_ess_base_orted_setup";
         goto error;
     }
-    opal_argv_free(hosts);
     return ORTE_SUCCESS;
 
  error:

orte/mca/ess/lsf/ess_lsf_module.c

@@ -10,7 +10,7 @@
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
  * Copyright (c) 2007-2011 Cisco Systems, Inc.  All rights reserved.
- * Copyright (c) 2013      Intel, Inc. All rights reserved.
+ * Copyright (c) 2013-2017 Intel, Inc. All rights reserved.
  * Copyright (c) 2016      Research Organization for Information Science
  *                         and Technology (RIST). All rights reserved.
  * $COPYRIGHT$
@@ -68,7 +68,6 @@ static int rte_init(void)
 {
     int ret;
     char *error = NULL;
-    char **hosts = NULL;
 
     /* run the prolog */
     if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
@@ -83,19 +82,11 @@ static int rte_init(void)
      * default procedure
      */
     if (ORTE_PROC_IS_DAEMON) {
-        if (NULL != orte_node_regex) {
-            /* extract the nodes */
-            if (ORTE_SUCCESS != (ret = orte_regex_extract_node_names(orte_node_regex, &hosts))) {
-                error = "orte_regex_extract_node_names";
-                goto error;
-            }
-        }
-        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
+        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup())) {
             ORTE_ERROR_LOG(ret);
             error = "orte_ess_base_orted_setup";
             goto error;
         }
-        opal_argv_free(hosts);
         return ORTE_SUCCESS;
     }
 

orte/mca/ess/slurm/ess_slurm_module.c

@@ -10,7 +10,7 @@
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
  * Copyright (c) 2008-2011 Cisco Systems, Inc.  All rights reserved.
- * Copyright (c) 2013      Intel, Inc. All rights reserved.
+ * Copyright (c) 2013-2017 Intel, Inc. All rights reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -62,7 +62,6 @@ static int rte_init(void)
 {
     int ret;
     char *error = NULL;
-    char **hosts = NULL;
 
     /* run the prolog */
     if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
@@ -77,23 +76,11 @@ static int rte_init(void)
      * default procedure
      */
     if (ORTE_PROC_IS_DAEMON) {
-        if (NULL != orte_node_regex) {
-            /* extract the nodes */
-            if (ORTE_SUCCESS != (ret =
-                orte_regex_extract_node_names(orte_node_regex, &hosts)) ||
-                NULL == hosts) {
-                error = "orte_regex_extract_node_names";
-                goto error;
-            }
-        }
-        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
+        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup())) {
             ORTE_ERROR_LOG(ret);
             error = "orte_ess_base_orted_setup";
             goto error;
         }
-        if (NULL != hosts) {
-            opal_argv_free(hosts);
-        }
         return ORTE_SUCCESS;
     }
 

orte/mca/ess/tm/ess_tm_module.c

@@ -67,7 +67,6 @@ static int rte_init(void)
 {
     int ret;
     char *error = NULL;
-    char **hosts = NULL;
 
     /* run the prolog */
     if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
@@ -82,21 +81,11 @@ static int rte_init(void)
      * default procedure
      */
     if (ORTE_PROC_IS_DAEMON) {
-        if (NULL != orte_node_regex) {
-            /* extract the nodes */
-            if (ORTE_SUCCESS != (ret =
-                orte_regex_extract_node_names(orte_node_regex, &hosts)) ||
-                NULL == hosts) {
-                error = "orte_regex_extract_node_names";
-                goto error;
-            }
-        }
-        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
+        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup())) {
             ORTE_ERROR_LOG(ret);
             error = "orte_ess_base_orted_setup";
             goto error;
         }
-        opal_argv_free(hosts);
         return ORTE_SUCCESS;
     }
 
@@ -194,4 +183,3 @@ static int tm_set_name(void)
 
     return ORTE_SUCCESS;
 }
-