Add an API to the errmgr so that apps can register for a callback to warn them of an impending migration - this gives apps a chance to cleanly terminate prior to being migrated for external reasons (e.g., impending failures). The timeout provided indicates to the daemon how long it should wait before proceeding to kill/migrate the process - if the process fails to exit before that time, the daemon will kill it.

This commit was SVN r24412.

orte/mca/errmgr/app/errmgr_app.c

@@ -56,7 +56,8 @@ orte_errmgr_base_module_t orte_errmgr_app_module = {
     update_state,
     NULL,
     NULL,
-    NULL
+    NULL,
+    orte_errmgr_base_register_migration_warning
 };
 
 /************************

orte/mca/errmgr/base/errmgr_base_fns.c

@@ -250,6 +250,11 @@ int orte_errmgr_base_update_state(orte_jobid_t job,
      */
     return ORTE_SUCCESS;
 }
+void orte_errmgr_base_register_migration_warning(struct timeval *tv)
+{
+    /* stub function - ignore */
+    return;
+}
 
 /********************
  * Utility functions

orte/mca/errmgr/base/errmgr_base_open.c

@@ -63,7 +63,8 @@ orte_errmgr_base_module_t orte_errmgr = {
     orte_errmgr_base_update_state,
     NULL, /* predicted_fault     */
     NULL, /* suggest_map_targets */
-    NULL  /* ft_event            */
+    NULL, /* ft_event            */
+    orte_errmgr_base_register_migration_warning
 };
 
 /**

orte/mca/errmgr/base/errmgr_private.h

@@ -28,6 +28,10 @@
 #include "orte/constants.h"
 #include "orte/types.h"
 
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif  /* HAVE_UNISTD_H */
+
 #include "opal/dss/dss_types.h"
 #include "orte/mca/plm/plm_types.h"
 #include "orte/runtime/orte_globals.h"
@@ -71,5 +75,7 @@ ORTE_DECLSPEC int orte_errmgr_base_update_state(orte_jobid_t job,
                                                 pid_t pid,
                                                 orte_exit_code_t exit_code);
 
+ORTE_DECLSPEC void orte_errmgr_base_register_migration_warning(struct timeval *tv);
+
 END_C_DECLS
 #endif

orte/mca/errmgr/errmgr.h

@@ -231,27 +231,42 @@ typedef int (*orte_errmgr_base_module_suggest_map_targets_fn_t)(orte_proc_t *pro
  */
 typedef int  (*orte_errmgr_base_module_ft_event_fn_t)(int state);
 
+/**
+ * Register a callback to alert caller when ORTE is preparing to
+ * migrate the process to another location. This provides an
+ * opportunity for the process to checkpoint any required state,
+ * and to cleanly shutdown.
+ *
+ * @param[in] delay Time to delay before assuming process is stuck
+ *                  and cannot exit on its own - and thus, go
+ *                  ahead and migrate it
+ */
+typedef void (*orte_errmgr_base_module_register_migration_warning_fn_t)(struct timeval *tv);
+
 /*
  * Module Structure
  */
 struct orte_errmgr_base_module_2_3_0_t {
     /** Initialization Function */
-    orte_errmgr_base_module_init_fn_t                   init;
+    orte_errmgr_base_module_init_fn_t                        init;
     /** Finalization Function */
-    orte_errmgr_base_module_finalize_fn_t               finalize;
+    orte_errmgr_base_module_finalize_fn_t                    finalize;
 
-    orte_errmgr_base_module_log_fn_t                    log;
-    orte_errmgr_base_module_abort_fn_t                  abort;
+    orte_errmgr_base_module_log_fn_t                         log;
+    orte_errmgr_base_module_abort_fn_t                       abort;
 
     /** Actual process failure notification */
-    orte_errmgr_base_module_update_state_fn_t           update_state;
+    orte_errmgr_base_module_update_state_fn_t                update_state;
     /** Predicted process/node failure notification */
-    orte_errmgr_base_module_predicted_fault_fn_t        predicted_fault;
+    orte_errmgr_base_module_predicted_fault_fn_t             predicted_fault;
     /** Suggest a node to map a restarting process onto */
-    orte_errmgr_base_module_suggest_map_targets_fn_t    suggest_map_targets;
+    orte_errmgr_base_module_suggest_map_targets_fn_t         suggest_map_targets;
 
     /** Handle any FT Notifications */
-    orte_errmgr_base_module_ft_event_fn_t               ft_event;
+    orte_errmgr_base_module_ft_event_fn_t                    ft_event;
+
+    /* Register to be warned of impending migration */
+    orte_errmgr_base_module_register_migration_warning_fn_t  register_migration_warning;
 };
 typedef struct orte_errmgr_base_module_2_3_0_t orte_errmgr_base_module_2_3_0_t;
 typedef orte_errmgr_base_module_2_3_0_t orte_errmgr_base_module_t;

orte/mca/errmgr/hnp/errmgr_hnp.c

@@ -72,7 +72,8 @@ static orte_errmgr_base_module_t global_module = {
     /* Suggest proc to node mapping */
     orte_errmgr_hnp_global_suggest_map_targets,
     /* FT Event hook  */
-    orte_errmgr_hnp_global_ft_event
+    orte_errmgr_hnp_global_ft_event,
+    orte_errmgr_base_register_migration_warning
 };
 
 

orte/mca/errmgr/orted/errmgr_orted.c

@@ -92,7 +92,8 @@ orte_errmgr_base_module_t orte_errmgr_orted_module = {
     update_state,
     predicted_fault,
     suggest_map_targets,
-    ft_event
+    ft_event,
+    orte_errmgr_base_register_migration_warning
 };
 
 /************************