Cleanup the pretty-print of odls cmds as some were missing. Add a new cmd to terminate the DVM, which the HNP will use to trun around and issue an xcast to the DVM.

orte/mca/odls/odls_types.h

@@ -58,6 +58,7 @@ typedef uint8_t orte_daemon_cmd_flag_t;
 #define ORTE_DAEMON_SPAWN_JOB_CMD           (orte_daemon_cmd_flag_t) 17
 #define ORTE_DAEMON_TERMINATE_JOB_CMD       (orte_daemon_cmd_flag_t) 18
 #define ORTE_DAEMON_HALT_VM_CMD             (orte_daemon_cmd_flag_t) 19
+#define ORTE_DAEMON_HALT_DVM_CMD            (orte_daemon_cmd_flag_t) 20
     
 /* request proc resource usage */
 #define ORTE_DAEMON_TOP_CMD                 (orte_daemon_cmd_flag_t) 22

orte/orted/orted_comm.c

@@ -114,6 +114,7 @@ void orte_daemon_recv(int status, orte_process_name_t* sender,
     orte_proc_t *cur_proc = NULL, *prev_proc = NULL;
     bool found = false;
     orte_node_t *node;
+    orte_grpcomm_signature_t *sig;
     
     /* unpack the command */
     n = 1;
@@ -459,6 +460,7 @@ void orte_daemon_recv(int status, orte_process_name_t* sender,
         return;
         break;
             
+        /****    HALT VM COMMAND    ****/
     case ORTE_DAEMON_HALT_VM_CMD:
         if (orte_debug_daemons_flag) {
             opal_output(0, "%s orted_cmd: received halt_vm cmd",
@@ -491,6 +493,27 @@ void orte_daemon_recv(int status, orte_process_name_t* sender,
         return;
         break;
 
+        /****    HALT DVM COMMAND    ****/
+    case ORTE_DAEMON_HALT_DVM_CMD:
+        if (orte_debug_daemons_flag) {
+            opal_output(0, "%s orted_cmd: received halt_dvm cmd",
+                        ORTE_NAME_PRINT(ORTE_PROC_MY_NAME));
+        }
+        /* we just need to xcast the HALT_VM cmd out, which will send
+         * it back into us */
+        answer = OBJ_NEW(opal_buffer_t);
+        command = ORTE_DAEMON_HALT_VM_CMD;
+        opal_dss.pack(answer, &command, 1, ORTE_DAEMON_CMD);
+        sig = OBJ_NEW(orte_grpcomm_signature_t);
+        sig->signature = (orte_process_name_t*)malloc(sizeof(orte_process_name_t));
+        sig->signature[0].jobid = ORTE_PROC_MY_NAME->jobid;
+        sig->signature[0].vpid = ORTE_VPID_WILDCARD;
+        orte_grpcomm.xcast(sig, ORTE_RML_TAG_DAEMON, answer);
+        OBJ_RELEASE(answer);
+        OBJ_RELEASE(sig);
+        return;
+        break;
+
         /****    SPAWN JOB COMMAND    ****/
     case ORTE_DAEMON_SPAWN_JOB_CMD:
         if (orte_debug_daemons_flag) {
@@ -1127,8 +1150,8 @@ void orte_daemon_recv(int status, orte_process_name_t* sender,
 static char *get_orted_comm_cmd_str(int command)
 {
     switch(command) {
-    case ORTE_DAEMON_NULL_CMD:
+    case ORTE_DAEMON_CONTACT_QUERY_CMD:
-        return strdup("NULL");
+        return strdup("ORTE_DAEMON_CONTACT_QUERY_CMD");
     case ORTE_DAEMON_KILL_LOCAL_PROCS:
         return strdup("ORTE_DAEMON_KILL_LOCAL_PROCS");
     case ORTE_DAEMON_SIGNAL_LOCAL_PROCS:
@@ -1137,26 +1160,49 @@ static char *get_orted_comm_cmd_str(int command)
         return strdup("ORTE_DAEMON_ADD_LOCAL_PROCS");
     case ORTE_DAEMON_TREE_SPAWN:
         return strdup("ORTE_DAEMON_TREE_SPAWN");
+        
+    case ORTE_DAEMON_HEARTBEAT_CMD:
+        return strdup("ORTE_DAEMON_HEARTBEAT_CMD");
+    case ORTE_DAEMON_EXIT_CMD:
+        return strdup("ORTE_DAEMON_EXIT_CMD");
+    case ORTE_DAEMON_PROCESS_AND_RELAY_CMD:
+        return strdup("ORTE_DAEMON_PROCESS_AND_RELAY_CMD");
     case ORTE_DAEMON_MESSAGE_LOCAL_PROCS:
         return strdup("ORTE_DAEMON_MESSAGE_LOCAL_PROCS");
-     case ORTE_DAEMON_EXIT_CMD:
+    case ORTE_DAEMON_NULL_CMD:
-        return strdup("ORTE_DAEMON_EXIT_CMD");
+        return strdup("NULL");
-    case ORTE_DAEMON_SPAWN_JOB_CMD:
+        
-        return strdup("ORTE_DAEMON_SPAWN_JOB_CMD");
-    case ORTE_DAEMON_CONTACT_QUERY_CMD:
-        return strdup("ORTE_DAEMON_CONTACT_QUERY_CMD");
     case ORTE_DAEMON_REPORT_JOB_INFO_CMD:
         return strdup("ORTE_DAEMON_REPORT_JOB_INFO_CMD");
     case ORTE_DAEMON_REPORT_NODE_INFO_CMD:
         return strdup("ORTE_DAEMON_REPORT_NODE_INFO_CMD");
     case ORTE_DAEMON_REPORT_PROC_INFO_CMD:
         return strdup("ORTE_DAEMON_REPORT_PROC_INFO_CMD");
-    case ORTE_DAEMON_HEARTBEAT_CMD:
+    case ORTE_DAEMON_SPAWN_JOB_CMD:
-        return strdup("ORTE_DAEMON_HEARTBEAT_CMD");
+        return strdup("ORTE_DAEMON_SPAWN_JOB_CMD");
+    case ORTE_DAEMON_TERMINATE_JOB_CMD:
+        return strdup("ORTE_DAEMON_TERMINATE_JOB_CMD");
+        
+    case ORTE_DAEMON_HALT_VM_CMD:
+        return strdup("ORTE_DAEMON_HALT_VM_CMD");
+    case ORTE_DAEMON_HALT_DVM_CMD:
+        return strdup("ORTE_DAEMON_HALT_DVM_CMD");
     case ORTE_DAEMON_TOP_CMD:
         return strdup("ORTE_DAEMON_TOP_CMD");
+    case ORTE_DAEMON_NAME_REQ_CMD:
+        return strdup("ORTE_DAEMON_NAME_REQ_CMD");
+    case ORTE_DAEMON_CHECKIN_CMD:
+        return strdup("ORTE_DAEMON_CHECKIN_CMD");
+        
+    case ORTE_TOOL_CHECKIN_CMD:
+        return strdup("ORTE_TOOL_CHECKIN_CMD");
+    case ORTE_DAEMON_PROCESS_CMD:
+        return strdup("ORTE_DAEMON_PROCESS_CMD");
     case ORTE_DAEMON_ABORT_PROCS_CALLED:
         return strdup("ORTE_DAEMON_ABORT_PROCS_CALLED");
+    case ORTE_DAEMON_NEW_COLL_ID:
+        return strdup("ORTE_DAEMON_NEW_COLL_ID");
+        
     default:
         return strdup("Unknown Command!");
     }

orte/tools/orte-dvm/orte-dvm.c

@@ -138,7 +138,7 @@ int main(int argc, char *argv[])
     char *param, *value;
     orte_job_t *jdata=NULL;
     orte_app_context_t *app;
-    char *uri;
+    char *uri, *ptr;
 
     /* Setup and parse the command line */
     memset(&myglobals, 0, sizeof(myglobals));
@@ -250,6 +250,17 @@ int main(int argc, char *argv[])
         } else if (0 == strcmp(myglobals.report_uri, "+")) {
             /* if '+', output to stderr */
             fprintf(stderr, "VMURI: %s\n", uri);
+        } else if (0 == strncasecmp(myglobals.report_uri, "file:", strlen("file:"))) {
+            ptr = strchr(myglobals.report_uri, ':');
+            ++ptr;
+            fp = fopen(ptr, "w");
+            if (NULL == fp) {
+                orte_show_help("help-orterun.txt", "orterun:write_file", false,
+                               myglobals.basename, "pid", ptr);
+                exit(0);
+            }
+            fprintf(fp, "%s\n", uri);
+            fclose(fp);
         } else {
             fp = fopen(myglobals.report_uri, "w");
             if (NULL == fp) {

orte/tools/orte-submit/orte-submit.c

@@ -487,7 +487,7 @@ int main(int argc, char *argv[])
     /* if this is the terminate command, just send it */
     if (myglobals.terminate) {
         opal_buffer_t *buf;
-        orte_daemon_cmd_flag_t cmd = ORTE_DAEMON_HALT_VM_CMD;
+        orte_daemon_cmd_flag_t cmd = ORTE_DAEMON_HALT_DVM_CMD;
         buf = OBJ_NEW(opal_buffer_t);
         opal_dss.pack(buf, &cmd, 1, ORTE_DAEMON_CMD);
         orte_rml.send_buffer_nb(ORTE_PROC_MY_HNP, buf,