Merge pull request #3287 from rhc54/topic/ht

Provide further (hopefully) helpful messages about the hotel size

Author: Ralph Castain

orte/orted/help-orted.txt

@@ -68,3 +68,15 @@ A request has timed out and will therefore fail:
 
 Your job may terminate as a result of this problem. You may want to
 adjust the MCA parameter pmix_server_max_wait and try again.
+#
+[noroom]
+A request for an asynchronous runtime operation cannot be fulfilled
+because of a lack of room in the tracking array:
+
+  Operation:  %s
+  Number of rooms:  %d
+
+This is usually caused by a large job that encounters significant
+delays across the cluster when starting the application processes.
+Your job may terminate as a result of this problem. You may want to
+adjust the MCA parameter pmix_server_max_reqs and try again.

orte/orted/pmix/pmix_server.c

@@ -556,7 +556,7 @@ static void pmix_server_dmdx_recv(int status, orte_process_name_t* sender,
          * amount of time to start the job */
         ORTE_ADJUST_TIMEOUT(req);
         if (OPAL_SUCCESS != (rc = opal_hotel_checkin(&orte_pmix_server_globals.reqs, req, &req->room_num))) {
-            ORTE_ERROR_LOG(rc);
+            orte_show_help("help-orted.txt", "noroom", true, req->operation, orte_pmix_server_globals.num_rooms);
             OBJ_RELEASE(req);
             send_error(rc, &idreq, sender);
         }
@@ -583,7 +583,7 @@ static void pmix_server_dmdx_recv(int status, orte_process_name_t* sender,
      * amount of time to start the job */
     ORTE_ADJUST_TIMEOUT(req);
     if (OPAL_SUCCESS != (rc = opal_hotel_checkin(&orte_pmix_server_globals.reqs, req, &req->room_num))) {
-        ORTE_ERROR_LOG(rc);
+        orte_show_help("help-orted.txt", "noroom", true, req->operation, orte_pmix_server_globals.num_rooms);
         OBJ_RELEASE(req);
         send_error(rc, &idreq, sender);
         return;

orte/orted/pmix/pmix_server_dyn.c

@@ -105,7 +105,7 @@ static void spawn(int sd, short args, void *cbdata)
 
     /* add this request to our tracker hotel */
     if (OPAL_SUCCESS != (rc = opal_hotel_checkin(&orte_pmix_server_globals.reqs, req, &req->room_num))) {
-        ORTE_ERROR_LOG(rc);
+        orte_show_help("help-orted.txt", "noroom", true, req->operation, orte_pmix_server_globals.num_rooms);
         goto callback;
     }
 

orte/orted/pmix/pmix_server_fence.c

@@ -37,6 +37,7 @@
 
 #include "orte/mca/errmgr/errmgr.h"
 #include "orte/util/name_fns.h"
+#include "orte/util/show_help.h"
 #include "orte/runtime/orte_globals.h"
 #include "orte/mca/grpcomm/grpcomm.h"
 #include "orte/mca/rml/rml.h"
@@ -164,7 +165,7 @@ static void dmodex_req(int sd, short args, void *cbdata)
             /* save the request in the hotel until the
              * data is returned */
             if (OPAL_SUCCESS != (rc = opal_hotel_checkin(&orte_pmix_server_globals.reqs, req, &req->room_num))) {
-                ORTE_ERROR_LOG(rc);
+                orte_show_help("help-orted.txt", "noroom", true, req->operation, orte_pmix_server_globals.num_rooms);
                 /* can't just return as that would cause the requestor
                  * to hang, so instead execute the callback */
                 goto callback;
@@ -180,7 +181,7 @@ static void dmodex_req(int sd, short args, void *cbdata)
          * that we don't know about yet. In this case, just
          * record the request and we will process it later */
         if (OPAL_SUCCESS != (rc = opal_hotel_checkin(&orte_pmix_server_globals.reqs, req, &req->room_num))) {
-            ORTE_ERROR_LOG(rc);
+            orte_show_help("help-orted.txt", "noroom", true, req->operation, orte_pmix_server_globals.num_rooms);
             /* can't just return as that would cause the requestor
              * to hang, so instead execute the callback */
             goto callback;
@@ -209,7 +210,7 @@ static void dmodex_req(int sd, short args, void *cbdata)
     /* track the request so we know the function and cbdata
      * to callback upon completion */
     if (OPAL_SUCCESS != (rc = opal_hotel_checkin(&orte_pmix_server_globals.reqs, req, &req->room_num))) {
-        ORTE_ERROR_LOG(rc);
+        orte_show_help("help-orted.txt", "noroom", true, req->operation, orte_pmix_server_globals.num_rooms);
         goto callback;
     }
 

orte/orted/pmix/pmix_server_pub.c

@@ -38,6 +38,7 @@
 
 #include "orte/mca/errmgr/errmgr.h"
 #include "orte/util/name_fns.h"
+#include "orte/util/show_help.h"
 #include "orte/runtime/orte_data_server.h"
 #include "orte/runtime/orte_globals.h"
 #include "orte/mca/rml/rml.h"
@@ -52,7 +53,7 @@ static void execute(int sd, short args, void *cbdata)
 
     /* add this request to our tracker hotel */
     if (OPAL_SUCCESS != (rc = opal_hotel_checkin(&orte_pmix_server_globals.reqs, req, &req->room_num))) {
-        ORTE_ERROR_LOG(rc);
+        orte_show_help("help-orted.txt", "noroom", true, req->operation, orte_pmix_server_globals.num_rooms);
         goto callback;
     }