Update the routed components as we no longer need to init_routes. Fixes case of direct launch via srun

Signed-off-by: Ralph Castain <[email protected]>

Signed-off-by: Ralph Castain <[email protected]>

Author: Ralph Castain

orte/mca/ess/base/ess_base_std_app.c

@@ -54,6 +54,7 @@
 #include "orte/mca/grpcomm/base/base.h"
 #include "orte/mca/oob/base/base.h"
 #include "orte/mca/rml/rml.h"
+#include "orte/mca/rml/base/rml_contact.h"
 #include "orte/mca/odls/odls_types.h"
 #include "orte/mca/filem/base/base.h"
 #include "orte/mca/errmgr/base/base.h"
@@ -148,6 +149,17 @@ int orte_ess_base_app_setup(bool db_restrict_local)
                                          "output-", NULL, NULL);
     }
     /* Setup the communication infrastructure */
+    /* Routed system */
+    if (ORTE_SUCCESS != (ret = mca_base_framework_open(&orte_routed_base_framework, 0))) {
+        ORTE_ERROR_LOG(ret);
+        error = "orte_routed_base_open";
+        goto error;
+    }
+    if (ORTE_SUCCESS != (ret = orte_routed_base_select())) {
+        ORTE_ERROR_LOG(ret);
+        error = "orte_routed_base_select";
+        goto error;
+    }
     /*
      * OOB Layer
      */
@@ -172,29 +184,40 @@ int orte_ess_base_app_setup(bool db_restrict_local)
         error = "orte_rml_base_select";
         goto error;
     }
+    /* if we have info on the HNP and local daemon, process it */
+    if (NULL != orte_process_info.my_hnp_uri) {
+        /* we have to set the HNP's name, even though we won't route messages directly
+         * to it. This is required to ensure that we -do- send messages to the correct
+         * HNP name
+         */
+        if (ORTE_SUCCESS != (ret = orte_rml_base_parse_uris(orte_process_info.my_hnp_uri,
+                                                            ORTE_PROC_MY_HNP, NULL))) {
+            ORTE_ERROR_LOG(ret);
+            error = "orte_rml_parse_HNP";
+            goto error;
+        }
+    }
+    if (NULL != orte_process_info.my_daemon_uri) {
+        /* extract the daemon's name so we can update the routing table */
+        if (ORTE_SUCCESS != (ret = orte_rml_base_parse_uris(orte_process_info.my_daemon_uri,
+                                                            ORTE_PROC_MY_DAEMON, NULL))) {
+            ORTE_ERROR_LOG(ret);
+            error = "orte_rml_parse_daemon";
+            goto error;
+        }
+        /* Set the contact info in the RML - this won't actually establish
+         * the connection, but just tells the RML how to reach the daemon
+         * if/when we attempt to send to it
+         */
+        orte_rml.set_contact_info(orte_process_info.my_daemon_uri);
+    }
+
     /* setup the errmgr */
     if (ORTE_SUCCESS != (ret = orte_errmgr_base_select())) {
         ORTE_ERROR_LOG(ret);
         error = "orte_errmgr_base_select";
         goto error;
     }
-    /* Routed system */
-    if (ORTE_SUCCESS != (ret = mca_base_framework_open(&orte_routed_base_framework, 0))) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_routed_base_open";
-        goto error;
-    }
-    if (ORTE_SUCCESS != (ret = orte_routed_base_select())) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_routed_base_select";
-        goto error;
-    }
-    /* setup the routed info */
-    if (ORTE_SUCCESS != (ret = orte_routed.init_routes(NULL, ORTE_PROC_MY_NAME->jobid, NULL))) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_routed.init_routes";
-        goto error;
-    }
 
     /* get a conduit for our use - we never route IO over fabric */
     OBJ_CONSTRUCT(&transports, opal_list_t);

orte/mca/ess/base/ess_base_std_orted.c

@@ -47,6 +47,7 @@
 
 #include "orte/mca/rtc/base/base.h"
 #include "orte/mca/rml/base/base.h"
+#include "orte/mca/rml/base/rml_contact.h"
 #include "orte/mca/routed/base/base.h"
 #include "orte/mca/routed/routed.h"
 #include "orte/mca/oob/base/base.h"
@@ -386,6 +387,17 @@ int orte_ess_base_orted_setup(char **hosts)
     }
 
     /* Setup the communication infrastructure */
+    /* Routed system */
+    if (ORTE_SUCCESS != (ret = mca_base_framework_open(&orte_routed_base_framework, 0))) {
+        ORTE_ERROR_LOG(ret);
+        error = "orte_routed_base_open";
+        goto error;
+    }
+    if (ORTE_SUCCESS != (ret = orte_routed_base_select())) {
+        ORTE_ERROR_LOG(ret);
+        error = "orte_routed_base_select";
+        goto error;
+    }
     if (ORTE_SUCCESS != (ret = mca_base_framework_open(&orte_oob_base_framework, 0))) {
         ORTE_ERROR_LOG(ret);
         error = "orte_oob_base_open";
@@ -407,6 +419,21 @@ int orte_ess_base_orted_setup(char **hosts)
         goto error;
     }
 
+    if (NULL != orte_process_info.my_hnp_uri) {
+        /* extract the HNP's name so we can update the routing table */
+        if (ORTE_SUCCESS != (ret = orte_rml_base_parse_uris(orte_process_info.my_hnp_uri,
+                                                            ORTE_PROC_MY_HNP, NULL))) {
+            ORTE_ERROR_LOG(ret);
+            error = "orte_rml_parse_HNP";
+            goto error;
+        }
+        /* Set the contact info in the RML - this won't actually establish
+         * the connection, but just tells the RML how to reach the HNP
+         * if/when we attempt to send to it
+         */
+        orte_rml.set_contact_info(orte_process_info.my_hnp_uri);
+    }
+
     /* setup the PMIx server */
     if (ORTE_SUCCESS != (ret = pmix_server_init())) {
         ORTE_ERROR_LOG(ret);
@@ -420,23 +447,6 @@ int orte_ess_base_orted_setup(char **hosts)
         error = "orte_errmgr_base_select";
         goto error;
     }
-    /* Routed system */
-    if (ORTE_SUCCESS != (ret = mca_base_framework_open(&orte_routed_base_framework, 0))) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_rml_base_open";
-        goto error;
-    }
-    if (ORTE_SUCCESS != (ret = orte_routed_base_select())) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_routed_base_select";
-        goto error;
-    }
-    /* setup the routed info */
-    if (ORTE_SUCCESS != (ret = orte_routed.init_routes(NULL, ORTE_PROC_MY_NAME->jobid, NULL))) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_routed.init_routes";
-        goto error;
-    }
 
     /* get a conduit for our use - we never route IO over fabric */
     OBJ_CONSTRUCT(&transports, opal_list_t);
@@ -533,12 +543,6 @@ int orte_ess_base_orted_setup(char **hosts)
         }
     }
 
-    /* setup the routed info */
-    if (ORTE_SUCCESS != (ret = orte_routed.init_routes(NULL, ORTE_PROC_MY_NAME->jobid, NULL))) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_routed.init_routes";
-        goto error;
-    }
     /* setup I/O forwarding system - must come after we init routes */
     if (ORTE_SUCCESS != (ret = mca_base_framework_open(&orte_iof_base_framework, 0))) {
         ORTE_ERROR_LOG(ret);

orte/mca/ess/base/ess_base_std_tool.c

@@ -157,13 +157,6 @@ int orte_ess_base_tool_setup(void)
         goto error;
     }
 
-    /* setup the routed info for all components */
-    if (ORTE_SUCCESS != (ret = orte_routed.init_routes(NULL, ORTE_PROC_MY_NAME->jobid, NULL))) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_routed.init_routes";
-        goto error;
-    }
-
     /* setup I/O forwarding system - must come after we init routes */
     if (NULL != orte_process_info.my_hnp_uri) {
         /* only do this if we were given an HNP */

orte/mca/ess/hnp/ess_hnp_module.c

@@ -633,13 +633,6 @@ static int rte_init(void)
     /* set the event base */
     opal_pmix_base_set_evbase(orte_event_base);
 
-    /* setup the routed info */
-    if (ORTE_SUCCESS != (ret = orte_routed.init_routes(NULL, ORTE_PROC_MY_NAME->jobid, NULL))) {
-        ORTE_ERROR_LOG(ret);
-        error = "orte_routed.init_routes";
-        goto error;
-    }
-
     /* setup the PMIx server */
     if (ORTE_SUCCESS != (ret = pmix_server_init())) {
         /* the server code already barked, so let's be quiet */

orte/mca/routed/base/base.h

@@ -53,8 +53,6 @@ ORTE_DECLSPEC int orte_routed_base_update_route(char *module, orte_process_name_
                                                 orte_process_name_t *route);
 ORTE_DECLSPEC orte_process_name_t orte_routed_base_get_route(char *module,
                                                              orte_process_name_t *target);
-ORTE_DECLSPEC int orte_routed_base_init_routes(char *module,
-                                               orte_jobid_t job, opal_buffer_t *ndat);
 ORTE_DECLSPEC int orte_routed_base_route_lost(char *module,
                                               const orte_process_name_t *route);
 ORTE_DECLSPEC bool orte_routed_base_route_is_defined(char *module,

orte/mca/routed/base/routed_base_fns.c

@@ -125,25 +125,6 @@ orte_process_name_t orte_routed_base_get_route(char *module, orte_process_name_t
     return *ORTE_NAME_INVALID;
 }
 
-int orte_routed_base_init_routes(char *module,
-                                 orte_jobid_t job, opal_buffer_t *ndat)
-{
-    orte_routed_base_active_t *active;
-    int rc;
-
-    OPAL_LIST_FOREACH(active, &orte_routed_base.actives, orte_routed_base_active_t) {
-        if (NULL == module ||
-            0 == strcmp(module, active->component->base_version.mca_component_name)) {
-            if (NULL != active->module->init_routes) {
-                if (ORTE_SUCCESS != (rc = active->module->init_routes(job, ndat))) {
-                    return rc;
-                }
-            }
-        }
-    }
-    return ORTE_SUCCESS;
-}
-
 int orte_routed_base_route_lost(char *module, const orte_process_name_t *route)
 {
     orte_routed_base_active_t *active;

orte/mca/routed/base/routed_base_frame.c

@@ -45,7 +45,6 @@ orte_routed_API_t orte_routed = {
     .delete_route = orte_routed_base_delete_route,
     .update_route = orte_routed_base_update_route,
     .get_route = orte_routed_base_get_route,
-    .init_routes = orte_routed_base_init_routes,
     .route_lost = orte_routed_base_route_lost,
     .route_is_defined = orte_routed_base_route_is_defined,
     .set_lifeline = orte_routed_base_set_lifeline,