Per request from Aurelien, make orterun report-pid and report-uri functions work the same as that of ompi-server. Since these are used for ompi-server-like functionality, it makes sense that the report options work the same. Make orte-top take the corresponding input the same way too for consistency.

The modified cmd line options are:

--report-uri x where x is either '-' for stdout, '+' for stderr, or a filename
--report-pid x where x is the same as above

For orte-top, you can now provide either a pid or a uri (which allows connection to remote mpiruns), specified either directly or with a "file:x" option as per mpirun's ompi-server option.

Note: I did not add a report-pid option to ompi-server as it probably wouldn't be useful - the report-uri option works as well, and allows remote access (which is likely the normal way it would be used).

This commit was SVN r20168.

orte/tools/orte-top/help-orte-top.txt

@@ -32,6 +32,44 @@ Pid provided: %d
 
 #
 [orte-top:pid-required]
-This tool requires that you specify the pid of the mpirun executing
+This tool requires that you specify contact info for the mpirun executing
 the specified rank(s). Please use the --help option for more information.
 
+#
+[orte-top:hnp-filename-bad]
+We are unable to parse the filename where contact info for the
+mpirun to be contacted was to be found. The option we were given was:
+
+--%s %s
+
+This appears to be missing the required ':' following the
+keyword "file". Please use the --help option for more information on
+the correct format for this command line option.
+#
+orte-top:hnp-filename-access]
+We are unable to access the filename where contact info for the
+mpirun to be contacted was to be found. The filename we were given was:
+
+File: %s
+
+Please use the --help option for more information on
+the correct format for this command line option.
+#
+[orte-top:hnp-file-bad]
+We are unable to read the mpirun's contact info from the
+given filename. The filename we were given was:
+
+FILE: %s
+
+Please use the --help option for more information on
+the correct format for this command line option.
+#
+[orte-top:hnp-uri-bad]
+We are unable to correctly parse the mpirun's contact info. The uri we were given was:
+
+URI: %s
+
+Please remember that this is *not* a standard uri, but
+a special format used internally by Open MPI for communications. It can
+best be generated by simply directing mpirun to put its
+uri in a file, and then giving us that filename.

orte/tools/orte-top/orte-top.1in

@@ -38,8 +38,16 @@ Display help for this command
 .
 .TP
 .B -pid | --pid \fR<value>\fP
-The pid of the mpirun whose processes you want information about. Note that
+The pid of the mpirun whose processes you want information about, or the name
-the ompi-top command must be executed on the same node as mpirun.
+of the file (specified as file:filename) that contains that info. Note that
+the ompi-top command must be executed on the same node as mpirun to use this option.
+.
+.
+.TP
+.B -uri | --uri \fR<value>\fP
+Specify the URI of the mpirun whose processes you want information about, or the name
+of the file (specified as file:filename) that contains that info. Note that
+the ompi-top command does not have to be executed on the same node as mpirun to use this option.
 .
 .
 .TP

orte/tools/orte-top/orte-top.c

@@ -53,6 +53,7 @@
 #include "orte/util/hnp_contact.h"
 #include "orte/util/name_fns.h"
 #include "orte/runtime/orte_wait.h"
+#include "orte/mca/rml/base/rml_contact.h"
 
 /*
  * Local variables & functions
@@ -69,6 +70,7 @@ static opal_event_t *my_exit_event;
 static FILE *fp = NULL;
 static bool help;
 static char *hnppidstr;
+static char *hnpuristr;
 static char *ranks;
 static orte_hnp_contact_t *target_hnp;
 static int update_rate;
@@ -117,6 +119,12 @@ opal_cmd_line_init_t cmd_line_opts[] = {
       &hnppidstr, OPAL_CMD_LINE_TYPE_STRING,
       "The pid of the mpirun that you wish to query/monitor" },
 
+    { NULL, NULL, NULL, 
+      '\0', "uri", "uri", 
+      1,
+      &hnpuristr, OPAL_CMD_LINE_TYPE_STRING,
+      "The uri of the mpirun that you wish to query/monitor" },
+    
     { NULL, NULL, NULL, 
       '\0', "rank", "rank", 
       1,
@@ -245,28 +253,6 @@ main(int argc, char *argv[])
         return ORTE_ERROR;
     }
     
-    /*
-     * Must specify the mpirun pid
-     */
-    if (NULL == hnppidstr) {
-        orte_show_help("help-orte-top.txt", "orte-top:pid-required", true);
-        return ORTE_ERROR;
-    }
-    
-    /* convert the pid */
-    hnppid = strtoul(hnppidstr, NULL, 10);
-    
-    /* if an output file was specified, open it */
-    if (NULL != logfile) {
-        fp = fopen(logfile, "w");
-        if (NULL == fp) {
-            orte_show_help("help-orte-top.txt", "orte-top:cant-open-logfile", true, logfile);
-            return ORTE_ERROR;
-        }
-    } else {
-        fp = stdout;
-    }
-    
     /***************************
      * We need all of OPAL and the TOOL portion of ORTE
      ***************************/
@@ -276,12 +262,15 @@ main(int argc, char *argv[])
     }
     
     OBJ_CONSTRUCT(&orte_exit, orte_trigger_event_t);
-
+    
     if (ORTE_SUCCESS != orte_wait_event(&my_exit_event, &orte_exit, "job_complete", abort_exit_callback)) {
         orte_finalize();
-        return 1;
+        exit(1);
     }
-
+    
+    /* setup the list for recvd stats */
+    OBJ_CONSTRUCT(&recvd_stats, opal_list_t);
+    
     /** setup callbacks for abort signals - from this point
      * forward, we need to abort in a manner that allows us
      * to cleanup
@@ -293,40 +282,172 @@ main(int argc, char *argv[])
                     abort_exit_callback, &int_handler);
     opal_signal_add(&int_handler, NULL);
     
-    /* setup the list for recvd stats */
-    OBJ_CONSTRUCT(&recvd_stats, opal_list_t);
-
     /*
-     * Get the list of available hnp's and setup contact info
+     * Must specify the mpirun pid
-     * to them in the RML
      */
-    OBJ_CONSTRUCT(&hnp_list, opal_list_t);
+    if (NULL != hnppidstr) {
-    if (ORTE_SUCCESS != (ret = orte_list_local_hnps(&hnp_list, true) ) ) {
+        if (0 == strncmp(hnppidstr, "file", strlen("file")) ||
-        goto cleanup;
+            0 == strncmp(hnppidstr, "FILE", strlen("FILE"))) {
-    }
+            char input[1024], *filename;
-    
+            FILE *fp;
-    /*
+            
-     * For each hnp in the listing
+            /* it is a file - get the filename */
-     */
+            filename = strchr(hnppidstr, ':');
-    while (NULL != (item  = opal_list_remove_first(&hnp_list))) {
+            if (NULL == filename) {
-        orte_hnp_contact_t *hnp = (orte_hnp_contact_t*)item;
+                /* filename is not correctly formatted */
-        if (hnppid == hnp->pid) {
+                orte_show_help("help-orte-top.txt", "orte-top:hnp-filename-bad", true, "pid", hnppidstr);
-            /* this is the one we want */
+                orte_finalize();
-            target_hnp = hnp;
+                exit(1);
-            break;
+            }
+            ++filename; /* space past the : */
+            
+            if (0 >= strlen(filename)) {
+                /* they forgot to give us the name! */
+                orte_show_help("help-orte-top.txt", "orte-top:hnp-filename-bad", true, "pid", hnppidstr);
+                orte_finalize();
+                exit(1);
+            }
+            
+            /* open the file and extract the pid */
+            fp = fopen(filename, "r");
+            if (NULL == fp) { /* can't find or read file! */
+                orte_show_help("help-orte-top.txt", "orte-top:hnp-filename-access", true, filename);
+                orte_finalize();
+                exit(1);
+            }
+            if (NULL == fgets(input, 1024, fp)) {
+                /* something malformed about file */
+                fclose(fp);
+                orte_show_help("help-orte-top.txt", "orte-top:hnp-file-bad", true, filename);
+                orte_finalize();
+                exit(1);
+            }
+            fclose(fp);
+            input[strlen(input)-1] = '\0';  /* remove newline */
+            /* convert the pid */
+            hnppid = strtoul(input, NULL, 10);
+        } else {
+            /* should just be the pid itself */
+            hnppid = strtoul(hnppidstr, NULL, 10);
         }
-        OBJ_RELEASE(hnp);
+        /*
+         * Get the list of available hnp's and setup contact info
+         * to them in the RML
+         */
+        OBJ_CONSTRUCT(&hnp_list, opal_list_t);
+        if (ORTE_SUCCESS != (ret = orte_list_local_hnps(&hnp_list, true) ) ) {
+            orte_show_help("help-orte-top.txt", "orte-top:pid-not-found", true, hnppid);
+            orte_finalize();
+            exit(1);
+        }
+        
+        /*
+         * For each hnp in the listing
+         */
+        while (NULL != (item  = opal_list_remove_first(&hnp_list))) {
+            orte_hnp_contact_t *hnp = (orte_hnp_contact_t*)item;
+            if (hnppid == hnp->pid) {
+                /* this is the one we want */
+                target_hnp = hnp;
+                /* let it continue to run so we deconstruct the list */
+                continue;
+            }
+            OBJ_RELEASE(hnp);
+        }
+        OBJ_DESTRUCT(&hnp_list);
+        
+        /* if we get here without finding the one we wanted, then abort */
+        if (NULL == target_hnp) {
+            orte_show_help("help-orte-top.txt", "orte-top:pid-not-found", true, hnppid);
+            orte_finalize();
+            exit(1);
+        }
+    } else if (NULL != hnpuristr) {
+        if (0 == strncmp(hnpuristr, "file", strlen("file")) ||
+            0 == strncmp(hnpuristr, "FILE", strlen("FILE"))) {
+            char input[1024], *filename;
+            FILE *fp;
+            
+            /* it is a file - get the filename */
+            filename = strchr(hnpuristr, ':');
+            if (NULL == filename) {
+                /* filename is not correctly formatted */
+                orte_show_help("help-orte-top.txt", "orte-top:hnp-filename-bad", true, "uri", hnpuristr);
+                orte_finalize();
+                exit(1);
+            }
+            ++filename; /* space past the : */
+            
+            if (0 >= strlen(filename)) {
+                /* they forgot to give us the name! */
+                orte_show_help("help-orte-top.txt", "orte-top:hnp-filename-bad", true, "uri", hnpuristr);
+                orte_finalize();
+                exit(1);
+            }
+            
+            /* open the file and extract the uri */
+            fp = fopen(filename, "r");
+            if (NULL == fp) { /* can't find or read file! */
+                orte_show_help("help-orte-top.txt", "orte-top:hnp-filename-access", true, filename);
+                orte_finalize();
+                exit(1);
+            }
+            if (NULL == fgets(input, 1024, fp)) {
+                /* something malformed about file */
+                fclose(fp);
+                orte_show_help("help-orte-top.txt", "orte-top:hnp-file-bad", true, filename);
+                orte_finalize();
+                exit(1);
+            }
+            fclose(fp);
+            input[strlen(input)-1] = '\0';  /* remove newline */
+            /* construct the target hnp info */
+            target_hnp = OBJ_NEW(orte_hnp_contact_t);
+            target_hnp->rml_uri = strdup(input);
+        } else {
+            /* should just be the uri itself - construct the target hnp info */
+            target_hnp = OBJ_NEW(orte_hnp_contact_t);
+            target_hnp->rml_uri = strdup(hnpuristr);
+        }
+        /* set the info in our contact table */
+        if (ORTE_SUCCESS != orte_rml.set_contact_info(target_hnp->rml_uri)) {
+            orte_show_help("help-orte-top.txt", "orte-top:hnp-uri-bad", true, target_hnp->rml_uri);
+            orte_finalize();
+            exit(1);
+        }
+        /* extract the name */
+        if (ORTE_SUCCESS != orte_rml_base_parse_uris(target_hnp->rml_uri, &target_hnp->name, NULL)) {
+            orte_show_help("help-orte-top.txt", "orte-top:hnp-uri-bad", true, target_hnp->rml_uri);
+            orte_finalize();
+            exit(1);
+        }
+        /* set the route to be direct */
+        if (ORTE_SUCCESS != orte_routed.update_route(&target_hnp->name, &target_hnp->name)) {
+            orte_show_help("help-orte-top.txt", "orte-top:hnp-uri-bad", true, target_hnp->rml_uri);
+            orte_finalize();
+            exit(1);
+        }
+    } else {
+        orte_show_help("help-orte-top.txt", "orte-top:no-contact-given", true);
+        orte_finalize();
+        exit(1);
     }
     
-    /* if we get here without finding the one we wanted, then abort */
-    if (NULL == target_hnp) {
-        orte_show_help("help-orte-top.txt", "orte-top:pid-not-found", true, hnppid);
-        goto cleanup;
-    }
-
     /* set the target hnp as our lifeline so we will terminate if it exits */
     orte_routed.set_lifeline(&target_hnp->name);
     
+    /* if an output file was specified, open it */
+    if (NULL != logfile) {
+        fp = fopen(logfile, "w");
+        if (NULL == fp) {
+            orte_show_help("help-orte-top.txt", "orte-top:cant-open-logfile", true, logfile);
+            orte_finalize();
+            exit(1);
+        }
+    } else {
+        fp = stdout;
+    }
+    
     /* setup a non-blocking recv to get answers - we don't know how
      * many daemons are going to send replies, so we just have to
      * accept whatever comes back
@@ -415,10 +536,6 @@ cleanup:
     opal_signal_del(&term_handler);
     opal_signal_del(&int_handler);
 
-    while (NULL != (item  = opal_list_remove_first(&hnp_list))) {
-        OBJ_RELEASE(item);
-    }
-    OBJ_DESTRUCT(&hnp_list);
     while (NULL != (item  = opal_list_remove_first(&recvd_stats))) {
         OBJ_RELEASE(item);
     }
@@ -440,10 +557,6 @@ static void abort_exit_callback(int fd, short ign, void *arg)
     opal_signal_del(&term_handler);
     opal_signal_del(&int_handler);
     
-    while (NULL != (item  = opal_list_remove_first(&hnp_list))) {
-        OBJ_RELEASE(item);
-    }
-    OBJ_DESTRUCT(&hnp_list);
     while (NULL != (item  = opal_list_remove_first(&recvd_stats))) {
         OBJ_RELEASE(item);
     }

orte/tools/orterun/help-orterun.txt

@@ -420,4 +420,9 @@ Please remember that the correct format for this command line option is:
 --ompi-server PID:pid-of-%s
 
 where PID can be either "PID" or "pid".
+#
+[orterun:write_file]
+%s was unable to open a file to printout %s as requested. The file
+name given was:
 
+File: %s

orte/tools/orterun/orterun.1in

@@ -306,11 +306,21 @@ Print out mpirun's URI during startup.
 .
 .
 .TP
+.B -report-uri-file\fR,\fP --report-uri-file <filename>
+Print out mpirun's URI to the specified file during startup.
+.
+.
+.TP
 .B -report-pid\fR,\fP --report-pid
 Print out mpirun's PID during startup.
 .
 .
 .TP
+.B -report-pid-file\fR,\fP --report-pid-file <filename>
+Print out mpirun's PID to the specified file during startup.
+.
+.
+.TP
 .B -rf \f |--rankfile <arg0>\fP.
 Provide a rankfile file.
 .

orte/tools/orterun/orterun.c

@@ -137,13 +137,14 @@ static opal_cmd_line_init_t cmd_line_init[] = {
     { NULL, NULL, NULL, 'q', NULL, "quiet", 0,
       &orterun_globals.quiet, OPAL_CMD_LINE_TYPE_BOOL,
       "Suppress helpful messages" },
-    { NULL, NULL, NULL, '\0', "report-pid", "report-pid", 0,
+    { NULL, NULL, NULL, '\0', "report-pid", "report-pid", 1,
-      &orterun_globals.report_pid, OPAL_CMD_LINE_TYPE_BOOL,
+      &orterun_globals.report_pid, OPAL_CMD_LINE_TYPE_STRING,
-      "Printout pid" },
+      "Printout pid on stdout [-], stderr [+], or a file [anything else]" },
-    { NULL, NULL, NULL, '\0', "report-uri", "report-uri", 0,
+    { NULL, NULL, NULL, '\0', "report-uri", "report-uri", 1,
-      &orterun_globals.report_uri, OPAL_CMD_LINE_TYPE_BOOL,
+      &orterun_globals.report_uri, OPAL_CMD_LINE_TYPE_STRING,
-      "Printout URI" },
+      "Printout URI on stdout [-], stderr [+], or a file [anything else]" },
-
+    
+    
     /* hetero apps */
     { "orte", "hetero", "apps", '\0', NULL, "hetero", 0,
         NULL, OPAL_CMD_LINE_TYPE_BOOL,
@@ -495,13 +496,29 @@ int orterun(int argc, char *argv[])
     }    
     
     /* check for request to report uri */
-    if (orterun_globals.report_uri) {
+    if (NULL != orterun_globals.report_uri) {
-        char *uri;
+        FILE *fp;
-        uri = orte_rml.get_contact_info();
+        char *rml_uri;
-        printf("%s uri: %s\n", orterun_basename, (NULL == uri) ? "NULL" : uri);
+        rml_uri = orte_rml.get_contact_info();
-        if (NULL != uri) {
+        if (0 == strcmp(orterun_globals.report_uri, "-")) {
-            free(uri);
+            /* if '-', then output to stdout */
+            printf("%s\n",  (NULL == rml_uri) ? "NULL" : rml_uri);
+        } else if (0 == strcmp(orterun_globals.report_uri, "+")) {
+            /* if '+', output to stderr */
+            fprintf(stderr, "%s\n",  (NULL == rml_uri) ? "NULL" : rml_uri);
+        } else {
+            fp = fopen(orterun_globals.report_uri, "w");
+            if (NULL == fp) {
+                orte_show_help("help-orterun.txt", "orterun:write_file", false,
+                               orterun_basename, "uri", orterun_globals.report_uri);
+                exit(0);
+            }
+            fprintf(fp, "%s\n", (NULL == rml_uri) ? "NULL" : rml_uri);
+            fclose(fp);
         }
+        if (NULL != rml_uri) {
+            free(rml_uri);
+        }        
     }
     
     /* Change the default behavior of libevent such that we want to
@@ -1179,6 +1196,8 @@ static int init_globals(void)
         orterun_globals.wait_for_server = false;
         orterun_globals.server_wait_timeout = 10;
         orterun_globals.stdin_target = "0";
+        orterun_globals.report_pid        = NULL;
+        orterun_globals.report_uri        = NULL;
     }
 
     /* Reset the other fields every time */
@@ -1187,8 +1206,6 @@ static int init_globals(void)
     orterun_globals.version                    = false;
     orterun_globals.verbose                    = false;
     orterun_globals.quiet                      = false;
-    orterun_globals.report_pid                 = false;
-    orterun_globals.report_uri                 = false;
     orterun_globals.by_node                    = false;
     orterun_globals.by_slot                    = false;
     orterun_globals.debugger                   = false;
@@ -1258,8 +1275,24 @@ static int parse_globals(int argc, char* argv[], opal_cmd_line_t *cmd_line)
     }
 
     /* check for request to report pid */
-    if (orterun_globals.report_pid) {
+    if (NULL != orterun_globals.report_pid) {
-        printf("%s pid: %d\n", orterun_basename, (int)getpid());
+        FILE *fp;
+        if (0 == strcmp(orterun_globals.report_pid, "-")) {
+            /* if '-', then output to stdout */
+            printf("%d\n", (int)getpid());
+        } else if (0 == strcmp(orterun_globals.report_pid, "+")) {
+            /* if '+', output to stderr */
+            fprintf(stderr, "%d\n", (int)getpid());
+        } else {
+            fp = fopen(orterun_globals.report_pid, "w");
+            if (NULL == fp) {
+                orte_show_help("help-orterun.txt", "orterun:write_file", false,
+                               orterun_basename, "pid", orterun_globals.report_pid);
+                exit(0);
+            }
+            fprintf(fp, "%d\n", (int)getpid());
+            fclose(fp);
+        }
     }
     
     /* Do we want a user-level debugger? */

orte/tools/orterun/orterun.h

@@ -42,8 +42,8 @@ struct orterun_globals_t {
     bool version;
     bool verbose;
     bool quiet;
-    bool report_pid;
+    char *report_pid;
-    bool report_uri;
+    char *report_uri;
     bool exit;
     bool by_node;
     bool by_slot;