Adding subapp only restart feature

closes idaholab#6297

framework/include/base/FEProblem.h

@@ -1209,6 +1209,7 @@ class FEProblem :
   bool _use_legacy_uo_initialization;
 
   bool _error_on_jacobian_nonzero_reallocation;
+  bool _force_restart;
   bool _fail_next_linear_convergence_check;
 
   /// Whether or not the system is currently computing the Jacobian matrix

framework/include/restart/RestartableDataIO.h

@@ -140,7 +140,7 @@ class RestartableDataIO
   FEProblem & _fe_problem;
 
   /// A vector of file handles, one per thread
-  std::vector<std::ifstream *> _in_file_handles;
+  std::vector<MooseSharedPointer<std::ifstream> > _in_file_handles;
 };
 
 #endif /* RESTARTABLEDATAIO_H */

framework/src/base/FEProblem.C

@@ -94,6 +94,7 @@ InputParameters validParams<FEProblem>()
   params.addParam<bool>("solve", true, "Whether or not to actually solve the Nonlinear system.  This is handy in the case that all you want to do is execute AuxKernels, Transfers, etc. without actually solving anything");
   params.addParam<bool>("use_nonlinear", true, "Determines whether to use a Nonlinear vs a Eigenvalue system (Automatically determined based on executioner)");
   params.addParam<bool>("error_on_jacobian_nonzero_reallocation", false, "This causes PETSc to error if it had to reallocate memory in the Jacobian matrix due to not having enough nonzeros");
+  params.addParam<bool>("force_restart", false, "EXPERIMENTAL: If true, a sub_app may use a restart file instead of using of using the master backup file");
 
   return params;
 }
@@ -153,6 +154,7 @@ FEProblem::FEProblem(const InputParameters & parameters) :
     _use_legacy_uo_aux_computation(_app.legacyUoAuxComputationDefault()),
     _use_legacy_uo_initialization(_app.legacyUoInitializationDefault()),
     _error_on_jacobian_nonzero_reallocation(getParam<bool>("error_on_jacobian_nonzero_reallocation")),
+    _force_restart(getParam<bool>("force_restart")),
     _fail_next_linear_convergence_check(false),
     _currently_computing_jacobian(false)
 {
@@ -314,10 +316,10 @@ void FEProblem::initialSetup()
   // Flush all output to _console that occur during construction and initialization of objects
   _app.getOutputWarehouse().mooseConsole();
 
-  if (_app.isRecovering() && _app.isUltimateMaster())
+  if (_app.isRecovering() && (_app.isUltimateMaster() || _force_restart))
     _resurrector->setRestartFile(_app.getRecoverFileBase());
 
-  if ((_app.isRestarting() || _app.isRecovering()) && _app.isUltimateMaster())
+  if ((_app.isRestarting() || _app.isRecovering()) && (_app.isUltimateMaster() || _force_restart))
     _resurrector->restartFromFile();
   else
   {
@@ -554,6 +556,9 @@ void FEProblem::initialSetup()
     Moose::perf_log.push("execMultiApps()", "Setup");
     //TODO: we did not check the convergence of the multiapps on initial
     execMultiApps(EXEC_INITIAL);
+
+    // We'll backup the Multiapp here
+    backupMultiApps(EXEC_INITIAL);
     Moose::perf_log.pop("execMultiApps()", "Setup");
   }
 

framework/src/restart/RestartableDataIO.C

@@ -30,10 +30,6 @@ RestartableDataIO::RestartableDataIO(FEProblem & fe_problem) :
 
 RestartableDataIO::~RestartableDataIO()
 {
-  unsigned int n_threads = libMesh::n_threads();
-
-  for (unsigned int tid=0; tid<n_threads; tid++)
-    delete _in_file_handles[tid];
 }
 
 void
@@ -236,8 +232,7 @@ RestartableDataIO::readRestartableDataHeader(std::string base_file_name)
 
     const unsigned int file_version = 2;
 
-    mooseAssert(_in_file_handles[tid] == NULL, "Looks like you might be leaking in RestartableDataIO.C");
-    _in_file_handles[tid] = new std::ifstream(file_name.c_str(), std::ios::in | std::ios::binary);
+    _in_file_handles[tid] = MooseSharedPointer<std::ifstream>(new std::ifstream(file_name.c_str(), std::ios::in | std::ios::binary));
 
     // header
     char id[2];
@@ -281,7 +276,7 @@ RestartableDataIO::readRestartableData(const RestartableDatas & restartable_data
   {
     const std::map<std::string, RestartableDataValue *> & restartable_data = restartable_datas[tid];
 
-    if (!_in_file_handles[tid]->is_open())
+    if (!_in_file_handles[tid].get() || !_in_file_handles[tid]->is_open())
       mooseError("In RestartableDataIO: Need to call readRestartableDataHeader() before calling readRestartableData()");
 
     deserializeRestartableData(restartable_data, *_in_file_handles[tid], recoverable_data);

test/tests/restart/restart_subapp_not_master/tests

@@ -13,6 +13,7 @@
     exodiff = 'two_step_solve_master.e'
     cli_args = 'Outputs/file_base=two_step_solve_master
                 MultiApps/full_solve/input_files=two_step_solve_sub.i'
+    recover = false
   [../]
   # split into sub solve 0 <= t <= 2 && master 2 <= t <= 4
   # this time restart the sub but _not_ the master
@@ -23,6 +24,5 @@
     cli_args = 'Outputs/file_base=two_step_solve_master_restart
                 MultiApps/full_solve/input_files=two_step_solve_sub_restart.i'
     prereq = two_step_solve_master
-    skip = 'Segfault #6297'
   [../]
 []

test/tests/restart/restart_subapp_not_master/two_step_solve_master.i

@@ -85,7 +85,7 @@
     type = FullSolveMultiApp
     execute_on = initial
     positions = '0 0 0'
-    ##input_files = two_step_solve_sub.i
+    # input file will come from cli-args
   [../]
 []
 
@@ -100,6 +100,6 @@
 []
 
 [Outputs]
-  ##file_base = two_step_solve_master
+  #file_base will come from cli-args
   exodus = true
 []

test/tests/restart/restart_subapp_not_master/two_step_solve_sub_restart.i

@@ -1,9 +1,10 @@
 [Mesh]
-  file = two_step_solve_master_out_full_solve0_cp/0002_mesh.cpr
+  file = two_step_solve_master_cp/0002_mesh.cpr
 []
 
 [Problem]
-  restart_file_base = two_step_solve_master_out_full_solve0_cp/0002
+  restart_file_base = two_step_solve_master_cp/LATEST
+  force_restart = true
 []
 
 [Functions]
@@ -25,13 +26,7 @@
   [../]
 []
 
-[ICs]
-  [./u_var]
-    type = FunctionIC
-    variable = u
-    function = exact_fn
-  [../]
-[]
+# Initial Condition will come from the restart file
 
 [Kernels]
   [./td]
@@ -75,8 +70,5 @@
 []
 
 [Outputs]
-  [./checkpoint]
-    type = Checkpoint
-    num_files = 3
-  [../]
+  exodus = true
 []