Merge branch 'master' of https://github.com/wrench-project/wrench

doc/getting_started.md

@@ -12,11 +12,11 @@ follow the instructions to [install it](@ref install).
 
 # Running a First Example #         {#getting-started-example}
 
-The WRENCH distribution provides an example WMS implementation (`simple-wms`) available 
+The WRENCH distribution provides an example WMS implementation (`SimpleWMS`) available 
 in the `examples` folder. Note that a simple installation via `make && make install`
 compiles all examples.
 
-WRENCH provides two implementations for the `simple-wms` example: a cloud-based 
+WRENCH provides two implementations for the `SimpleWMS` example: a cloud-based 
 implementation `wrench-simple-wms-cloud`, and an implementation to run workflows 
 in a batch system (e.g., SLURM) `wrench-simple-wms-batch`.
 
@@ -34,7 +34,7 @@ wrench-simple-wms-batch <PATH-TO-WRENCH-SRC-FOLDER>/examples/two_hosts.xml <PATH
 
 ## Understanding the Simple-WMS Examples      {#getting-started-example-simplewms}
 
-The `simple-wms` example requires two arguments: (1) a [SimGrid virtual platform 
+The `SimpleWMS` example requires two arguments: (1) a [SimGrid virtual platform 
 description file](http://simgrid.gforge.inria.fr/simgrid/3.17/doc/platform.html); and
 (2) a WRENCH workflow file.
 
@@ -48,7 +48,25 @@ A detailed description on how to create a platform description file can be found
 
 **WRENCH workflow file:**
 WRENCH provides native parsers for [DAX](http://workflowarchive.org) (DAG in XML) 
-and [JSON](http://workflowhub.org/traces/) worfklow description file formats. Refer to their respective Web sites for detailed documentation.
+and [JSON](http://workflowhub.org/traces/) worfklow description file formats. Refer to 
+their respective Web sites for detailed documentation.
+
+The `SimpleWMS` example implementations (either cloud or batch) are structured as follows:
+
+- The first step is to read and parse the workflow and the SimGrid platform files, and
+  create a simulation object (wrench::Simulation).
+- A storage service (wrench::SimpleStorageService) is created and deployed on a host.
+- A cloud (wrench::CloudService) or a batch (wrench::BatchService) service is created and 
+  deployed on a host. Both services are seen by the simulation engine as a compute service 
+  (wrench::ComputeService) – jobs can then be scheduled to these resources. 
+- A WMS (wrench::WMS) is instantiated (in this case the `SimpleWMS`) with a reference to 
+  the workflow object (wrench::Workflow) and a scheduler (wrench::Scheduler). For the 
+  cloud example, a cloud scheduler is required to decide when to spawn VMs on hosts. The
+  batch service does not require a specific scheduler, since resources are fixed. In 
+  such case, a regular scheduler can be used.
+- A file registry (wrench::FileRegistryService), or a file replica catalog keeps track
+  of files stored in different storage services.
+- Workflow input files are staged into the storage service, and the simulation is launched. 
 
 
 # Preparing the Environment #         {#getting-started-prep}

include/wrench/services/compute/batch/BatchServiceProperty.h

@@ -17,14 +17,6 @@ namespace wrench {
     public:
         /** @brief The overhead to start a thread execution, in seconds **/
         DECLARE_PROPERTY_NAME(THREAD_STARTUP_OVERHEAD);
-//        /** @brief The number of bytes in the control message sent by the service to notify a submitter that a job has completed.  **/
-//        DECLARE_PROPERTY_NAME(STANDARD_JOB_DONE_MESSAGE_PAYLOAD);
-//        /** @brief The number of bytes in the control message sent by the service to notify a submitter that a job has failed.  **/
-//        DECLARE_PROPERTY_NAME(STANDARD_JOB_FAILED_MESSAGE_PAYLOAD);
-//        /** @brief The number of bytes in the control message sent to the service to submit a job.  **/
-//        DECLARE_PROPERTY_NAME(SUBMIT_BATCH_JOB_ANSWER_MESSAGE_PAYLOAD);
-//        /** @brief The number of bytes in the control message sent by the service in answer to a job submission.  **/
-//        DECLARE_PROPERTY_NAME(SUBMIT_BATCH_JOB_REQUEST_MESSAGE_PAYLOAD);
         /** @brief The host selection algorithm. Can be:
          *    - FIRSTFIT
          *    - BESTFIT

include/wrench/workflow/execution_events/FailureCause.h

@@ -26,9 +26,11 @@ namespace wrench {
     /** \cond DEVELOPER    */
     /***********************/
 
+
     /**
      * @brief A top-level class to describe all simulation-valid failures that can occur during
      *        workflow execution
+     *
      */
     class FailureCause {
 

src/wrench/services/compute/standard_job_executor/StandardJobExecutor.cpp

@@ -141,6 +141,7 @@ namespace wrench {
 
       // Kill the StandardJobExecutor
       this->kill_actor();
+
     }
 
     /**

src/wrench/services/compute/standard_job_executor/WorkunitMulticoreExecutor.cpp

@@ -44,7 +44,7 @@ namespace wrench {
      * @param num_cores: the number of cores available to the executor
      * @param callback_mailbox: the callback mailbox to which the worker
      *        thread can send "work done" messages
-     * @param workunit: the workinit to perform
+     * @param workunit: the workunit to perform
      * @param default_storage_service: the default storage service from which to read/write data (if any)
      * @param thread_startup_overhead: the thread_startup overhead, in seconds
      */

src/wrench/simgrid_S4U_util/S4U_Daemon.cpp

@@ -40,20 +40,29 @@ namespace wrench {
      * @param process_name: the name of the simulated process/actor
      */
     S4U_Daemon::S4U_Daemon(std::string process_name)
-            : process_name(process_name),
-              mailbox_name("") {
+            {
+      this->process_name = process_name; // TODO: Why does this leak?
+      this->mailbox_name="";
       this->terminated = false;
     }
 
     S4U_Daemon::~S4U_Daemon() {
 //        WRENCH_INFO("In the Daemon Destructor");
     }
 
-    int daemonGoodbye(void *x, void*y) {
+
+    /**
+     * \cond
+     */
+    static int daemon_goodbye(void *x, void *y) {
       WRENCH_INFO("Terminating");
       return 0;
     }
 
+    /**
+     * \endcond
+     */
+
     /**
      * @brief Start the daemon
      *
@@ -79,7 +88,7 @@ namespace wrench {
       if (daemonized)
         this->s4u_actor->daemonize();
 
-      this->s4u_actor->onExit(daemonGoodbye, (void *)(this->process_name.c_str()));
+      this->s4u_actor->onExit(daemon_goodbye, (void *) (this->process_name.c_str()));
 
 
       // Set the mailbox receiver

src/wrench/simulation/Simulation.cpp

@@ -22,10 +22,7 @@ XBT_LOG_NEW_DEFAULT_CATEGORY(simulation, "Log category for Simulation");
 namespace wrench {
 
     /**
-     * @brief Exception handler to catch SIGABRT signals from SimGrid (which should
-     * probably throw exceptions at some point)
-     *
-     * @param signal: the signal number
+     * \cond
      */
     void signal_handler(int signal) {
       if (signal == SIGABRT) {
@@ -35,6 +32,9 @@ namespace wrench {
         std::cerr << "Unexpected signal " << signal << " received\n";
       }
     }
+    /**
+     * \endcond
+     */
 
     /**
      * @brief  Constructor

test/simulation/StandardJobExecutorTest.cpp

@@ -908,7 +908,13 @@ class OneMultiCoreTaskTestWMS : public wrench::WMS {
 
         this->test->storage_service1->deleteFile(workflow->getFileById("output_file"));
 
+
         workflow->removeTask(task);
+
+//        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+//        // TODO: (this is a general design issue)
+//        delete executor;
+
       }
 
       /** Case 2: Create a multicore task with 50% parallel efficiency that lasts one hour **/
@@ -974,6 +980,10 @@ class OneMultiCoreTaskTestWMS : public wrench::WMS {
 
         this->test->storage_service1->deleteFile(workflow->getFileById("output_file"));
 
+//        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+//        // TODO: (this is a general design issue)
+//        delete executor;
+
       }
 
       /** Case 3: Create a multicore task with 50% parallel efficiency and include thread startup overhead **/
@@ -1037,6 +1047,10 @@ class OneMultiCoreTaskTestWMS : public wrench::WMS {
         workflow->removeTask(task);
 
         this->test->storage_service1->deleteFile(workflow->getFileById("output_file"));
+//
+//        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+//        // TODO: (this is a general design issue)
+//        delete executor;
 
       }
 
@@ -1229,6 +1243,10 @@ class TwoMultiCoreTasksTestWMS : public wrench::WMS {
 
         this->test->storage_service1->deleteFile(workflow->getFileById("output_file"));
 
+//        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+//        // TODO: (this is a general design issue)
+//        delete executor;
+
       }
 
 
@@ -1311,6 +1329,10 @@ class TwoMultiCoreTasksTestWMS : public wrench::WMS {
         workflow->removeTask(task2);
         this->test->storage_service1->deleteFile(workflow->getFileById("output_file"));
 
+//        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+//        // TODO: (this is a general design issue)
+//        delete executor;
+
       }
 
 
@@ -1396,6 +1418,10 @@ class TwoMultiCoreTasksTestWMS : public wrench::WMS {
         workflow->removeTask(task3);
         this->test->storage_service1->deleteFile(workflow->getFileById("output_file"));
 
+//        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+//        // TODO: (this is a general design issue)
+//        delete executor;
+
       }
 
       // Terminate everything
@@ -1584,6 +1610,10 @@ class MultiHostTestWMS : public wrench::WMS {
         workflow->removeTask(task1);
         workflow->removeTask(task2);
         this->test->storage_service1->deleteFile(workflow->getFileById("output_file"));
+
+//        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+//                                                // TODO: (this is a general design issue)
+//        delete executor;
       }
 
       /** Case 2: Create 4 tasks that will run in best fit manner **/
@@ -1637,9 +1667,9 @@ class MultiHostTestWMS : public wrench::WMS {
         wrench::StandardJobExecutorDoneMessage *msg = dynamic_cast<wrench::StandardJobExecutorDoneMessage *>(message.get());
         if (!msg) {
           wrench::StandardJobExecutorFailedMessage *msg = dynamic_cast<wrench::StandardJobExecutorFailedMessage *>(message.get());
-          std::cerr << "----> " << msg->cause->toString() << "\n";
+//          std::cerr << "----> " << msg->cause->toString() << "\n";
 
-          throw std::runtime_error("Unexpected '" + message->getName() + "' message");
+          throw std::runtime_error("Unexpected '" + msg->cause->toString() + "' error");
 
         }
 
@@ -1667,6 +1697,10 @@ class MultiHostTestWMS : public wrench::WMS {
 //          throw std::runtime_error("Case 2: Unexpected task2 end date: " + std::to_string(task2->getEndDate()));
 //        }
 
+//        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+//                                                // TODO: (this is a general design issue)
+//        delete executor;
+
         workflow->removeTask(task1);
         workflow->removeTask(task2);
         workflow->removeTask(task3);
@@ -1837,6 +1871,10 @@ class JobTerminationTestDuringAComputationWMS : public wrench::WMS {
         workflow->removeTask(task2);
         workflow->removeTask(task3);
         workflow->removeTask(task4);
+
+        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+        // TODO: (this is a general design issue)
+        delete executor;
       }
 
 
@@ -2002,6 +2040,11 @@ class JobTerminationTestDuringATransferWMS : public wrench::WMS {
         workflow->removeTask(task2);
         workflow->removeTask(task3);
         workflow->removeTask(task4);
+
+        wrench::S4U_Simulation::sleep(0.00001); // TODO: This is needed to avoid a segfault from the delete
+        // TODO: (this is a general design issue)
+        delete executor;
+
       }