Fix comm_spawn by registering nspace info only when needed - either when we have local procs, or when job-level info is required by connecting jobs

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

Author: Ralph Castain

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

@@ -612,10 +612,10 @@ static void _process_dmdx_reply(int fd, short args, void *cbdata)
     }
 
     if (NULL == nptr) {
-/*
- * We may not have this namespace because someone asked about this namespace
- * but there are not processses from it running on this host
- */
+        /*
+         * We may not have this namespace because someone asked about this namespace
+         * but there are not processses from it running on this host
+         */
         nptr = PMIX_NEW(pmix_nspace_t);
         (void)strncpy(nptr->nspace, caddy->lcd->proc.nspace, PMIX_MAX_NSLEN);
         nptr->server = PMIX_NEW(pmix_server_nspace_t);

orte/mca/odls/base/odls_base_default_fns.c

@@ -481,16 +481,16 @@ int orte_odls_base_default_construct_child_list(opal_buffer_t *buffer,
         }
     }
 
- COMPLETE:
+  COMPLETE:
     /* register this job with the PMIx server - need to wait until after we
      * have computed the #local_procs before calling the function */
-    if (ORTE_SUCCESS != (rc = orte_pmix_server_register_nspace(jdata))) {
+    if (ORTE_SUCCESS != (rc = orte_pmix_server_register_nspace(jdata, false))) {
         ORTE_ERROR_LOG(rc);
         goto REPORT_ERROR;
     }
     return ORTE_SUCCESS;
 
- REPORT_ERROR:
+  REPORT_ERROR:
     /* we have to report an error back to the HNP so we don't just
      * hang. Although there shouldn't be any errors once this is
      * all debugged, it is still good practice to have a way

orte/orted/orted_main.c

@@ -573,7 +573,7 @@ int orte_daemon(int argc, char *argv[])
         orte_pre_condition_transports(jdata);
 
         /* register the singleton's nspace with our PMIx server */
-        if (ORTE_SUCCESS != (ret = orte_pmix_server_register_nspace(jdata))) {
+        if (ORTE_SUCCESS != (ret = orte_pmix_server_register_nspace(jdata, false))) {
           ORTE_ERROR_LOG(ret);
           goto DONE;
         }

orte/orted/pmix/pmix_server.h

@@ -12,7 +12,7 @@
  * Copyright (c) 2006-2013 Los Alamos National Security, LLC.
  *                         All rights reserved.
  * Copyright (c) 2010-2011 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
@@ -34,9 +34,8 @@ ORTE_DECLSPEC void pmix_server_finalize(void);
 ORTE_DECLSPEC void pmix_server_register_params(void);
 
 
-ORTE_DECLSPEC int orte_pmix_server_register_nspace(orte_job_t *jdata);
+ORTE_DECLSPEC int orte_pmix_server_register_nspace(orte_job_t *jdata, bool force);
 
 END_C_DECLS
 
 #endif /* PMIX_SERVER_H_ */
-

orte/orted/pmix/pmix_server_dyn.c

@@ -13,7 +13,7 @@
  *                         All rights reserved.
  * Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved.
  * Copyright (c) 2011      Oak Ridge National Labs.  All rights reserved.
- * Copyright (c) 2013-2016 Intel, Inc.  All rights reserved.
+ * Copyright (c) 2013-2017 Intel, Inc. All rights reserved.
  * Copyright (c) 2014      Mellanox Technologies, Inc.
  *                         All rights reserved.
  * Copyright (c) 2014-2016 Research Organization for Information Science
@@ -377,7 +377,7 @@ static void _cnlk(int status, opal_list_t *data, void *cbdata)
         goto release;
     }
     OBJ_DESTRUCT(&buf);
-    if (ORTE_SUCCESS != (rc = orte_pmix_server_register_nspace(jdata))) {
+    if (ORTE_SUCCESS != (rc = orte_pmix_server_register_nspace(jdata, true))) {
         OBJ_RELEASE(jdata);
         goto release;
     }
@@ -441,7 +441,7 @@ static void _cnct(int sd, short args, void *cbdata)
          * registered with the local PMIx server */
         if (!orte_get_attribute(&jdata->attributes, ORTE_JOB_NSPACE_REGISTERED, NULL, OPAL_BOOL)) {
             /* it hasn't been registered yet, so register it now */
-            if (ORTE_SUCCESS != (rc = orte_pmix_server_register_nspace(jdata))) {
+            if (ORTE_SUCCESS != (rc = orte_pmix_server_register_nspace(jdata, true))) {
                 goto release;
             }
         }

orte/orted/pmix/pmix_server_register_fns.c

@@ -51,7 +51,7 @@
 #include "pmix_server.h"
 
 /* stuff proc attributes for sending back to a proc */
-int orte_pmix_server_register_nspace(orte_job_t *jdata)
+int orte_pmix_server_register_nspace(orte_job_t *jdata, bool force)
 {
     int rc;
     orte_proc_t *pptr;
@@ -74,8 +74,9 @@ int orte_pmix_server_register_nspace(orte_job_t *jdata)
                         ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
                         ORTE_JOBID_PRINT(jdata->jobid));
 
-    /* if this job has no local procs, then no need to register them */
-    if (0 == jdata->num_local_procs) {
+    /* if this job has no local procs, then no need to register
+     * it unless the job info is needed by connecting jobs */
+    if (!force && 0 == jdata->num_local_procs) {
         return ORTE_SUCCESS;
     }