Merge pull request #3725 from rhc54/topic/cleanup

Ensure we properly cleanup on termination, including when terminating due to ctrl-c

Author: Ralph Castain

orte/mca/ess/base/ess_base_std_app.c

@@ -342,6 +342,8 @@ int orte_ess_base_app_finalize(void)
     (void) mca_base_framework_close(&orte_state_base_framework);
 
     orte_session_dir_finalize(ORTE_PROC_MY_NAME);
+    /* cleanup the process info */
+    orte_proc_info_finalize();
 
     return ORTE_SUCCESS;
 }

orte/orted/orted_main.c

@@ -933,6 +933,10 @@ int orte_daemon(int argc, char *argv[])
     orte_finalize();
     opal_finalize_util();
 
+    orte_session_dir_cleanup(ORTE_JOBID_WILDCARD);
+    /* cleanup the process info */
+    orte_proc_info_finalize();
+
     if (orte_debug_flag) {
         fprintf(stderr, "exiting with status %d\n", orte_exit_status);
     }

orte/runtime/orte_finalize.c

@@ -12,7 +12,7 @@
  * Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved.
  * Copyright (c) 2011-2013 Los Alamos National Security, LLC.
  *                         All rights reserved.
- * Copyright (c) 2014-2016 Intel, Inc. All rights reserved.
+ * Copyright (c) 2014-2017 Intel, Inc. All rights reserved.
  * Copyright (c) 2017      Research Organization for Information Science
  *                         and Technology (RIST). All rights reserved.
  * $COPYRIGHT$
@@ -84,9 +84,6 @@ int orte_finalize(void)
     orte_schizo.finalize();
     (void) mca_base_framework_close(&orte_schizo_base_framework);
 
-    /* cleanup the process info */
-    orte_proc_info_finalize();
-
     /* Close the general debug stream */
     opal_output_close(orte_debug_output);
 

orte/tools/orterun/orterun.c

@@ -86,6 +86,7 @@
 #include "orte/mca/rml/rml.h"
 #include "orte/mca/state/state.h"
 #include "orte/util/proc_info.h"
+#include "orte/util/session_dir.h"
 #include "orte/util/show_help.h"
 #include "orte/util/threads.h"
 
@@ -222,6 +223,9 @@ int orterun(int argc, char *argv[])
     /* cleanup and leave */
     orte_submit_finalize();
     orte_finalize();
+    orte_session_dir_cleanup(ORTE_JOBID_WILDCARD);
+    /* cleanup the process info */
+    orte_proc_info_finalize();
 
     if (orte_debug_flag) {
         fprintf(stderr, "exiting with status %d\n", orte_exit_status);

orte/util/session_dir.c

@@ -370,45 +370,33 @@ int orte_session_dir(bool create, orte_process_name_t *proc)
 int
 orte_session_dir_cleanup(orte_jobid_t jobid)
 {
-    int rc = ORTE_SUCCESS;
-
     if (!orte_create_session_dirs || orte_process_info.rm_session_dirs ) {
         /* we haven't created them or RM will clean them up for us*/
         return ORTE_SUCCESS;
     }
 
-    if (NULL == orte_process_info.job_session_dir ||
+    if (NULL == orte_process_info.jobfam_session_dir ||
         NULL == orte_process_info.proc_session_dir) {
         /* this should never happen - it means we are calling
          * cleanup *before* properly setting up the session
          * dir system. This leaves open the possibility of
          * accidentally removing directories we shouldn't
          * touch
          */
-        rc = ORTE_ERR_NOT_INITIALIZED;
-        goto CLEANUP;
+        return ORTE_ERR_NOT_INITIALIZED;
     }
 
     /* recursively blow the whole session away for our job family,
      * saving only output files
      */
-    opal_os_dirpath_destroy(orte_process_info.job_session_dir,
+    opal_os_dirpath_destroy(orte_process_info.jobfam_session_dir,
                             true, orte_dir_check_file);
 
-    /* now attempt to eliminate the top level directory itself - this
-     * will fail if anything is present, but ensures we cleanup if
-     * we are the last one out
-     */
-    if( NULL != orte_process_info.top_session_dir ){
-        opal_os_dirpath_destroy(orte_process_info.top_session_dir,
-                                false, orte_dir_check_file);
-    }
-
-    if (opal_os_dirpath_is_empty(orte_process_info.job_session_dir)) {
+    if (opal_os_dirpath_is_empty(orte_process_info.jobfam_session_dir)) {
         if (orte_debug_flag) {
-            opal_output(0, "sess_dir_cleanup: found job session dir empty - deleting");
+            opal_output(0, "sess_dir_cleanup: found jobfam session dir empty - deleting");
         }
-        rmdir(orte_process_info.job_session_dir);
+        rmdir(orte_process_info.jobfam_session_dir);
     } else {
         if (orte_debug_flag) {
             if (OPAL_ERR_NOT_FOUND ==
@@ -418,12 +406,10 @@ orte_session_dir_cleanup(orte_jobid_t jobid)
                 opal_output(0, "sess_dir_cleanup: job session dir not empty - leaving");
             }
         }
-        goto CLEANUP;
     }
 
-    if ( NULL != orte_process_info.top_session_dir ){
-
-        if( opal_os_dirpath_is_empty(orte_process_info.top_session_dir) ) {
+    if (NULL != orte_process_info.top_session_dir) {
+        if (opal_os_dirpath_is_empty(orte_process_info.top_session_dir)) {
             if (orte_debug_flag) {
                 opal_output(0, "sess_dir_cleanup: found top session dir empty - deleting");
             }
@@ -440,9 +426,17 @@ orte_session_dir_cleanup(orte_jobid_t jobid)
         }
     }
 
-CLEANUP:
+    /* now attempt to eliminate the top level directory itself - this
+     * will fail if anything is present, but ensures we cleanup if
+     * we are the last one out
+     */
+    if( NULL != orte_process_info.top_session_dir ){
+        opal_os_dirpath_destroy(orte_process_info.top_session_dir,
+                                false, orte_dir_check_file);
+    }
+
 
-    return rc;
+    return ORTE_SUCCESS;
 }