Remove the uggly surf_watched_hosts pimple and allow the autorestart to work properly.

Revalidate the tesh files which were actually wrong since hosts were not restarted at the right time...

diff --git a/examples/msg/masterslave/masterslave_cpu_ti_crosstraffic.tesh b/examples/msg/masterslave/masterslave_cpu_ti_crosstraffic.tesh
index 9ff571d..61ab881 100644 (file)
--- a/examples/msg/masterslave/masterslave_cpu_ti_crosstraffic.tesh
+++ b/examples/msg/masterslave/masterslave_cpu_ti_crosstraffic.tesh
@@ -288,8 +288,9 @@ $ ${bindir:=.}/masterslave_failure$EXEEXT --log=xbt_cfg.thres:critical --log=no_
 > [  0.000000] (1:master@Tremblay) Bourassa
 > [  0.000000] (1:master@Tremblay) Tremblay
 > [  0.000000] (1:master@Tremblay) Got 20 task to process :
-> [  1.000000] (0:@) Restart processes on host: Jupiter
+> [  1.000000] (0:@) Restart processes on host: Fafard
 > [  1.000000] (1:master@Tremblay) Mmh. Something went wrong with 'Jupiter'. Nevermind. Let's keep going!
+> [  2.000000] (0:@) Restart processes on host: Jupiter
 > [ 11.000000] (1:master@Tremblay) Mmh. Got timeouted while speaking to 'Fafard'. Nevermind. Let's keep going!
 > [ 12.082474] (1:master@Tremblay) Send completed
 > [ 12.082474] (4:slave@Ginette) Received "Task"

diff --git a/examples/msg/masterslave/masterslave_failure.tesh b/examples/msg/masterslave/masterslave_failure.tesh
index 00e6ee5..09ac278 100644 (file)
--- a/examples/msg/masterslave/masterslave_failure.tesh
+++ b/examples/msg/masterslave/masterslave_failure.tesh
@@ -12,8 +12,9 @@ $ masterslave/masterslave_failure$EXEEXT --log=xbt_cfg.thres:critical --log=no_l
 > [  0.000000] (1:master@Tremblay) Bourassa
 > [  0.000000] (1:master@Tremblay) Tremblay
 > [  0.000000] (1:master@Tremblay) Got 20 task to process :
-> [  1.000000] (0:@) Restart processes on host: Jupiter
+> [  1.000000] (0:@) Restart processes on host: Fafard
 > [  1.000000] (1:master@Tremblay) Mmh. Something went wrong with 'Jupiter'. Nevermind. Let's keep going!
+> [  2.000000] (0:@) Restart processes on host: Jupiter
 > [ 11.000000] (1:master@Tremblay) Mmh. Got timeouted while speaking to 'Fafard'. Nevermind. Let's keep going!
 > [ 12.030928] (1:master@Tremblay) Send completed
 > [ 12.030928] (4:slave@Ginette) Received "Task"

diff --git a/examples/msg/masterslave/masterslave_failure_crosstraffic.tesh b/examples/msg/masterslave/masterslave_failure_crosstraffic.tesh
index e77fff6..a9b1cb8 100644 (file)
--- a/examples/msg/masterslave/masterslave_failure_crosstraffic.tesh
+++ b/examples/msg/masterslave/masterslave_failure_crosstraffic.tesh
@@ -12,8 +12,9 @@ $ masterslave/masterslave_failure$EXEEXT --log=xbt_cfg.thres:critical --log=no_l
 > [  0.000000] (1:master@Tremblay) Bourassa
 > [  0.000000] (1:master@Tremblay) Tremblay
 > [  0.000000] (1:master@Tremblay) Got 20 task to process :
-> [  1.000000] (0:@) Restart processes on host: Jupiter
+> [  1.000000] (0:@) Restart processes on host: Fafard
 > [  1.000000] (1:master@Tremblay) Mmh. Something went wrong with 'Jupiter'. Nevermind. Let's keep going!
+> [  2.000000] (0:@) Restart processes on host: Jupiter
 > [ 11.000000] (1:master@Tremblay) Mmh. Got timeouted while speaking to 'Fafard'. Nevermind. Let's keep going!
 > [ 12.082474] (1:master@Tremblay) Send completed
 > [ 12.082474] (4:slave@Ginette) Received "Task"

diff --git a/src/include/surf/surf.h b/src/include/surf/surf.h
index 7f16b84..d1c5a58 100644 (file)
--- a/src/include/surf/surf.h
+++ b/src/include/surf/surf.h
@@ -643,6 +643,11 @@ XBT_PUBLIC_DATA(s_surf_model_description_t) surf_new_model_description[];
  */
 XBT_PUBLIC_DATA(xbt_dynar_t) model_list;
 
+/** \ingroup SURF_simulation
+ *  \brief List of hosts that have juste restarted and whose autorestart process should be restarted.
+ */
+XBT_PUBLIC_DATA(xbt_dynar_t) host_that_restart;
+
 /** \ingroup SURF_simulation
  *  \brief List of hosts for which one want to be notified if they ever restart.
  */
@@ -735,8 +740,6 @@ XBT_PUBLIC(xbt_dict_t) get_as_router_properties(const char* name);
 int surf_get_nthreads(void);
 void surf_set_nthreads(int nthreads);
 
-void surf_watched_hosts(void);
-
 /*
  * Returns the initial path. On Windows the initial path is
  * the current directory for the current process in the other

diff --git a/src/simix/smx_global.c b/src/simix/smx_global.c
index 79c866d..352e05f 100644 (file)
--- a/src/simix/smx_global.c
+++ b/src/simix/smx_global.c
@@ -328,6 +328,14 @@ void SIMIX_run(void)
         SIMIX_simcall_post((smx_action_t) action->data);
     }
 
+    /* Autorestart all process */
+    char *hostname = NULL;
+    xbt_dynar_foreach(host_that_restart,iter,hostname) {
+      XBT_INFO("Restart processes on host: %s",hostname);
+      SIMIX_host_autorestart(SIMIX_host_get_by_name(hostname));
+    }
+    xbt_dynar_reset(host_that_restart);
+
     /* Clean processes to destroy */
     SIMIX_process_empty_trash();
 

diff --git a/src/simix/smx_host.c b/src/simix/smx_host.c
index c113725..f519b13 100644 (file)
--- a/src/simix/smx_host.c
+++ b/src/simix/smx_host.c
@@ -322,7 +322,10 @@ void SIMIX_host_restart_processes(smx_host_t host)
 {
   unsigned int cpt;
   smx_process_arg_t arg;
-  xbt_dynar_foreach(SIMIX_host_priv(host)->auto_restart_processes,cpt,arg) {
+  xbt_dynar_t process_list = SIMIX_host_priv(host)->auto_restart_processes;
+  if(!process_list) return;
+
+  xbt_dynar_foreach(process_list,cpt,arg) {
 
     smx_process_t process;
 
@@ -353,7 +356,7 @@ void SIMIX_host_restart_processes(smx_host_t host)
 
     }
   }
-  xbt_dynar_reset(SIMIX_host_priv(host)->auto_restart_processes);
+  xbt_dynar_reset(process_list);
 }
 
 void SIMIX_host_autorestart(smx_host_t host)

diff --git a/src/surf/cpu_cas01.c b/src/surf/cpu_cas01.c
index 264f318..b3aeec0 100644 (file)
--- a/src/surf/cpu_cas01.c
+++ b/src/surf/cpu_cas01.c
@@ -268,8 +268,6 @@ static void cpu_update_resource_state(void *id,
   lmm_variable_t var = NULL;
   lmm_element_t elem = NULL;
 
-  surf_watched_hosts();
-
   if (event_type == cpu->power_event) {
     cpu->power_scale = value;
     lmm_update_constraint_bound(surf_cpu_model->model_private->maxmin_system, cpu->constraint,
@@ -290,9 +288,11 @@ static void cpu_update_resource_state(void *id,
     if (tmgr_trace_event_free(event_type))
       cpu->power_event = NULL;
   } else if (event_type == cpu->state_event) {
-    if (value > 0)
+    if (value > 0) {
+      if(cpu->state_current == SURF_RESOURCE_OFF)
+        xbt_dynar_push_as(host_that_restart, char*, (cpu->generic_resource.name));
       cpu->state_current = SURF_RESOURCE_ON;
-    else {
+    } else {
       lmm_constraint_t cnst = cpu->constraint;
 
       cpu->state_current = SURF_RESOURCE_OFF;

diff --git a/src/surf/cpu_ti.c b/src/surf/cpu_ti.c
index 7084967..144b581 100644 (file)
--- a/src/surf/cpu_ti.c
+++ b/src/surf/cpu_ti.c
@@ -502,8 +502,6 @@ static void cpu_ti_update_resource_state(void *id,
   cpu_ti_t cpu = id;
   surf_action_cpu_ti_t action;
 
-  surf_watched_hosts();
-
   if (event_type == cpu->power_event) {
     tmgr_trace_t power_trace;
     surf_cpu_ti_tgmr_t trace;
@@ -533,9 +531,11 @@ static void cpu_ti_update_resource_state(void *id,
       cpu->power_event = NULL;
 
   } else if (event_type == cpu->state_event) {
-    if (value > 0)
+    if (value > 0) {
+      if(cpu->state_current == SURF_RESOURCE_OFF)
+        xbt_dynar_push_as(host_that_restart, char*, (cpu->generic_resource.name));
       cpu->state_current = SURF_RESOURCE_ON;
-    else {
+    } else {
       cpu->state_current = SURF_RESOURCE_OFF;
 
       /* put all action running on cpu to failed */

diff --git a/src/surf/surf.c b/src/surf/surf.c
index 4e9992a..a53a97f 100644 (file)
--- a/src/surf/surf.c
+++ b/src/surf/surf.c
@@ -114,6 +114,7 @@ xbt_dynar_t model_list = NULL;
 tmgr_history_t history = NULL;
 lmm_system_t maxmin_system = NULL;
 xbt_dynar_t surf_path = NULL;
+xbt_dynar_t host_that_restart = NULL;
 xbt_dict_t watched_hosts_lib;
 
 /* Don't forget to update the option description in smx_config when you change this */
@@ -545,6 +546,9 @@ double surf_solve(double max_date)
   tmgr_trace_event_t event = NULL;
   unsigned int iter;
 
+  if(!host_that_restart)
+    host_that_restart = xbt_dynar_new(sizeof(char*), NULL);
+
   if (max_date != -1.0 && max_date != NOW) {
     min = max_date - NOW;
   }
@@ -606,7 +610,9 @@ double surf_solve(double max_date)
             tmgr_history_get_next_event_leq(history, next_event_date,
                                             &value,
                                             (void **) &resource))) {
-      if (resource->model->model_private->resource_used(resource)) {
+      if (resource->model->model_private->resource_used(resource) ||
+          xbt_dict_get_or_null(watched_hosts_lib,resource->name)
+          ) {
         min = next_event_date - NOW;
         XBT_DEBUG
             ("This event will modify model state. Next event set to %f",
@@ -615,7 +621,8 @@ double surf_solve(double max_date)
       /* update state of model_obj according to new value. Does not touch lmm.
          It will be modified if needed when updating actions */
       XBT_DEBUG("Calling update_resource_state for resource %s with min %lf",
-             resource->model->name, min);
+             resource->name, min);
+
       resource->model->model_private->update_resource_state(resource,
                                                             event, value,
                                                             next_event_date);
@@ -669,58 +676,3 @@ static void surf_update_actions_state(surf_model_t model)
   model->model_private->update_actions_state(NOW, min);
 }
 
-
-/* This function is a pimple that we ought to fix. But it won't be easy.
- *
- * The surf_solve() function does properly return the set of actions that changed.
- * Instead, each model change a global data, and then the caller of surf_solve must
- * pick into these sets of action_failed and action_done.
- *
- * This was not clean but ok as long as we didn't had to restart the processes when the resource comes back up.
- * We worked by putting sentinel actions on every resources we are interested in,
- * so that surf informs us if/when the corresponding resource fails.
- *
- * But this does not work to get Simix informed of when a resource comes back up, and this is where this pimple comes.
- * We have a set of resources that are currently down and for which simix needs to know when it comes back up.
- * And the current function is called *at every simulation step* to sweep over that set, searching for a resource
- * that was turned back up in the meanwhile. This is UGLY and slow.
- *
- * The proper solution would be to not rely on globals for the action_failed and action_done swags.
- * They must be passed as parameter by the caller (the handling of these actions in simix may let you
- * think that these two sets can be merged, but their handling in SimDag induce the contrary unless this
- * simdag code can check by itself whether the action is done of failed -- seems very doable, but yet more
- * cleanup to do).
- *
- * Once surf_solve() is passed the set of actions that changed, you want to add a new set of resources back up
- * as parameter to this function. You also want to add a boolean field "restart_watched" to each resource, and
- * make sure that whenever a resource with this field enabled comes back up, it's added to that set so that Simix
- * sees it and react accordingly. This would kill that need for surf to call simix.
- *
- */
-
-static void remove_watched_host(void *key)
-{
-  xbt_dict_remove(watched_hosts_lib, *(char**)key);
-}
-
-void surf_watched_hosts(void)
-{
-  char *key;
-  void *host;
-  xbt_dict_cursor_t cursor;
-  xbt_dynar_t hosts = xbt_dynar_new(sizeof(char*), NULL);
-
-  XBT_DEBUG("Check for host SURF_RESOURCE_ON on watched_hosts_lib");
-  xbt_dict_foreach(watched_hosts_lib,cursor,key,host)
-  {
-    if(SIMIX_host_get_state(host) == SURF_RESOURCE_ON){
-      XBT_INFO("Restart processes on host: %s",SIMIX_host_get_name(host));
-      SIMIX_host_autorestart(host);
-      xbt_dynar_push_as(hosts, char*, key);
-    }
-    else
-      XBT_DEBUG("See SURF_RESOURCE_OFF on host: %s",key);
-  }
-  xbt_dynar_map(hosts, remove_watched_host);
-  xbt_dynar_free(&hosts);
-}