1 // Copyright (C) 2007-2023 CEA, EDF, OPEN CASCADE
3 // Copyright (C) 2003-2007 OPEN CASCADE, EADS/CCR, LIP6, CEA/DEN,
4 // CEDRAT, EDF R&D, LEG, PRINCIPIA R&D, BUREAU VERITAS
6 // This library is free software; you can redistribute it and/or
7 // modify it under the terms of the GNU Lesser General Public
8 // License as published by the Free Software Foundation; either
9 // version 2.1 of the License, or (at your option) any later version.
11 // This library is distributed in the hope that it will be useful,
12 // but WITHOUT ANY WARRANTY; without even the implied warranty of
13 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 // Lesser General Public License for more details.
16 // You should have received a copy of the GNU Lesser General Public
17 // License along with this library; if not, write to the Free Software
18 // Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 // See http://www.salome-platform.org/ or email : webmaster.salome@opencascade.com
23 #include "SALOME_ResourcesManager.hxx"
24 #include "Utils_ExceptHandlers.hxx"
25 #include "Utils_CorbaException.hxx"
43 #include <sys/types.h>
45 #include "utilities.h"
47 #define MAX_SIZE_FOR_HOSTNAME 256;
51 const char *SALOME_ResourcesManager::_ResourcesManagerNameInNS = "/ResourcesManager";
53 //=============================================================================
57 //=============================================================================
59 SALOME_ResourcesManager::SALOME_ResourcesManager(CORBA::ORB_ptr orb,
60 PortableServer::POA_var poa,
61 SALOME_NamingService_Abstract *ns,
62 const char *xmlFilePath)
63 : _rm(new ResourcesManager_cpp(xmlFilePath))
65 MESSAGE("SALOME_ResourcesManager constructor");
67 _orb = CORBA::ORB::_duplicate(orb) ;
69 PortableServer::POAManager_var pman = poa->the_POAManager();
70 CORBA::PolicyList policies;
72 PortableServer::ThreadPolicy_var threadPol(poa->create_thread_policy(PortableServer::SINGLE_THREAD_MODEL));
73 policies[0] = PortableServer::ThreadPolicy::_duplicate(threadPol);
74 _poa = poa->create_POA("SingleThreadPOA",pman,policies);
77 PortableServer::ObjectId_var id = _poa->activate_object(this);
78 CORBA::Object_var obj = _poa->id_to_reference(id);
79 Engines::ResourcesManager_var refContMan = Engines::ResourcesManager::_narrow(obj);
80 _NS->Register(refContMan,_ResourcesManagerNameInNS);
81 MESSAGE("SALOME_ResourcesManager constructor end");
84 //=============================================================================
86 * Standard constructor, parse resource file.
87 * - if ${APPLI} exists in environment,
88 * look for ${HOME}/${APPLI}/CatalogResources.xml
89 * - else look for default:
90 * ${KERNEL_ROOT_DIR}/share/salome/resources/kernel/CatalogResources.xml
91 * - parse XML resource file.
93 //=============================================================================
95 SALOME_ResourcesManager::SALOME_ResourcesManager(CORBA::ORB_ptr orb,
96 PortableServer::POA_var poa,
97 SALOME_NamingService_Abstract *ns) : _rm(new ResourcesManager_cpp())
100 _orb = CORBA::ORB::_duplicate(orb) ;
102 PortableServer::POAManager_var pman = poa->the_POAManager();
103 CORBA::PolicyList policies;
105 PortableServer::ThreadPolicy_var threadPol(poa->create_thread_policy(PortableServer::SINGLE_THREAD_MODEL));
106 policies[0] = PortableServer::ThreadPolicy::_duplicate(threadPol);
107 _poa = poa->create_POA("SingleThreadPOA",pman,policies);
108 threadPol->destroy();
110 PortableServer::ObjectId_var id = _poa->activate_object(this);
111 CORBA::Object_var obj = _poa->id_to_reference(id);
112 Engines::ResourcesManager_var refContMan = Engines::ResourcesManager::_narrow(obj);
114 _NS->Register(refContMan,_ResourcesManagerNameInNS);
117 //=============================================================================
119 * Standard Destructor
121 //=============================================================================
123 SALOME_ResourcesManager::~SALOME_ResourcesManager()
125 MESSAGE("SALOME_ResourcesManager destructor");
129 //=============================================================================
131 * shutdown all the containers, then the ContainerManager servant
133 //=============================================================================
135 void SALOME_ResourcesManager::Shutdown()
140 _NS->Destroy_Name(_ResourcesManagerNameInNS);
141 PortableServer::ObjectId_var oid = _poa->servant_to_id(this);
142 _poa->deactivate_object(oid);
146 * Return list of resources available (regarding content of CatalogResources.xml) but select only those with canRunContainers attribute set to true.
147 * And for each resource the number of proc available of it.
149 * \sa SALOME_ResourcesManager::ListAllResourcesInCatalog
151 void SALOME_ResourcesManager::ListAllAvailableResources(Engines::ResourceList_out machines, Engines::IntegerList_out nbProcsOfMachines)
153 const MapOfParserResourcesType& zeList(_rm->GetList());
154 std::vector<std::string> ret0;
155 std::vector<int> ret1;
156 for(MapOfParserResourcesType::const_iterator it=zeList.begin();it!=zeList.end();it++)
158 const ParserResourcesType& elt((*it).second);
159 if(elt.can_run_containers)
161 ret0.push_back(elt.HostName);
162 ret1.push_back(elt.DataForSort._nbOfNodes*elt.DataForSort._nbOfProcPerNode);
165 machines=new Engines::ResourceList;
166 nbProcsOfMachines=new Engines::IntegerList;
167 std::size_t sz(ret0.size());
168 machines->length((CORBA::ULong)sz); nbProcsOfMachines->length((CORBA::ULong)sz);
169 for(std::size_t j=0;j<sz;j++)
171 (*machines)[(CORBA::ULong)j]=CORBA::string_dup(ret0[j].c_str());
172 (*nbProcsOfMachines)[(CORBA::ULong)j]=ret1[j];
177 * Return list of resources available (regarding content of CatalogResources.xml) whatever canRunContainers attribute value.
179 * \sa SALOME_ResourcesManager::ListAllAvailableResources
181 Engines::ResourceList *SALOME_ResourcesManager::ListAllResourcesInCatalog()
183 const MapOfParserResourcesType& zeList(_rm->GetList());
184 auto sz = zeList.size();
185 Engines::ResourceList *ret(new Engines::ResourceList);
188 for(auto it : zeList)
190 (*ret)[i++] = CORBA::string_dup( it.second.HostName.c_str() );
195 //=============================================================================
196 //! get the name of resources fitting the specified constraints (params)
198 * If hostname specified, check it is local or known in resources catalog.
201 * - select first machines with corresponding OS (all machines if
202 * parameter OS empty),
203 * - then select the sublist of machines on which the component is known
204 * (if the result is empty, that probably means that the inventory of
205 * components is probably not done, so give complete list from previous step)
207 //=============================================================================
209 Engines::ResourceList *
210 SALOME_ResourcesManager::GetFittingResources(const Engines::ResourceParameters& params)
212 //MESSAGE("ResourcesManager::GetFittingResources");
213 Engines::ResourceList_var ret;
216 resourceParams p = resourceParameters_CORBAtoCPP(params);
220 // Call C++ ResourceManager
221 std::vector <std::string> vec = _rm->GetFittingResources(p);
224 ret = resourceList_CPPtoCORBA(vec);
226 catch(const ResourcesException &ex)
228 INFOS("Caught exception in GetFittingResources C++: " << ex.msg);
229 THROW_SALOME_CORBA_EXCEPTION(ex.msg.c_str(),SALOME::BAD_PARAM);
235 //=============================================================================
237 * dynamically obtains the first machines
239 //=============================================================================
242 SALOME_ResourcesManager::FindFirst(const Engines::ResourceList& listOfResources)
245 std::vector<std::string> rl = resourceList_CORBAtoCPP(listOfResources);
247 return CORBA::string_dup(_rm->Find("first", rl).c_str());
251 SALOME_ResourcesManager::Find(const char* policy, const Engines::ResourceList& listOfResources)
254 std::vector<std::string> rl = resourceList_CORBAtoCPP(listOfResources);
256 return CORBA::string_dup(_rm->Find(policy, rl).c_str());
259 Engines::ResourceDefinition*
260 SALOME_ResourcesManager::GetResourceDefinition(const char * name)
262 Engines::ResourceDefinition_var resDef;
264 ParserResourcesType resource = _rm->GetResourcesDescr(name);
265 resDef = resourceDefinition_CPPtoCORBA(resource);
266 } catch (const exception & ex) {
267 INFOS("Caught exception in GetResourceDefinition: " << ex.what());
268 THROW_SALOME_CORBA_EXCEPTION(ex.what(), SALOME::BAD_PARAM);
271 return resDef._retn();
275 SALOME_ResourcesManager::AddResource(const Engines::ResourceDefinition& new_resource,
276 CORBA::Boolean write,
277 const char * xml_file)
281 ParserResourcesType resource = resourceDefinition_CORBAtoCPP(new_resource);
282 _rm->AddResourceInCatalog(resource);
286 _rm->WriteInXmlFile(std::string(xml_file));
287 _rm->ParseXmlFiles();
290 catch (const SALOME_Exception & e)
292 INFOS("Error in AddResourceInCatalog: " << e);
293 THROW_SALOME_CORBA_EXCEPTION(e.what(), SALOME::BAD_PARAM);
295 catch (const ResourcesException & e)
297 INFOS("Error in AddResourceInCatalog: " << e.msg);
298 THROW_SALOME_CORBA_EXCEPTION(e.msg.c_str(), SALOME::BAD_PARAM);
303 SALOME_ResourcesManager::RemoveResource(const char * resource_name,
304 CORBA::Boolean write,
305 const char * xml_file)
309 _rm->DeleteResourceInCatalog(resource_name);
311 catch (const SALOME_Exception & e)
313 INFOS("Error in DeleteResourceInCatalog: " << e);
314 THROW_SALOME_CORBA_EXCEPTION(e.what(), SALOME::BAD_PARAM);
319 _rm->WriteInXmlFile(std::string(xml_file));
320 _rm->ParseXmlFiles();
325 SALOME_ResourcesManager::getMachineFile(const char * resource_name,
326 CORBA::Long nb_procs,
327 const char * parallelLib)
329 std::string machine_file_name("");
331 if (std::string(parallelLib) == "Dummy")
333 MESSAGE("[getMachineFile] parallelLib is Dummy");
334 MapOfParserResourcesType resourcesList = _rm->GetList();
335 if (resourcesList.find(std::string(resource_name)) != resourcesList.end())
337 ParserResourcesType resource = resourcesList[std::string(resource_name)];
339 // Check if resource is cluster or not
340 if (resource.ClusterMembersList.empty())
342 //It is not a cluster so we create a cluster with one machine
343 ParserResourcesType fake_node;
344 fake_node.HostName = resource.HostName;
345 fake_node.Protocol = resource.Protocol;
346 fake_node.ClusterInternalProtocol = resource.ClusterInternalProtocol;
347 fake_node.UserName = resource.UserName;
348 fake_node.AppliPath = resource.AppliPath;
349 fake_node.DataForSort = resource.DataForSort;
351 resource.ClusterMembersList.push_front(fake_node);
354 // Creating list of machines for creating the machine file
355 std::list<std::string> list_of_machines;
356 std::list<ParserResourcesType>::iterator cluster_it =
357 resource.ClusterMembersList.begin();
358 while (cluster_it != resource.ClusterMembersList.end())
360 // For each member of the cluster we add a nbOfNodes * nbOfProcPerNode in the list
361 unsigned int number_of_proc = (*cluster_it).DataForSort._nbOfNodes *
362 (*cluster_it).DataForSort._nbOfProcPerNode;
363 for (unsigned int i = 0; i < number_of_proc; i++)
364 list_of_machines.push_back((*cluster_it).HostName);
368 // Creating machine file
369 machine_file_name = tmpnam(NULL);
370 std::ofstream machine_file(machine_file_name.c_str(), std::ios_base::out);
372 CORBA::Long machine_number = 0;
373 std::list<std::string>::iterator it = list_of_machines.begin();
374 while (machine_number != nb_procs)
376 // Adding a new node to the machine file
377 machine_file << *it << std::endl;
381 if (it == list_of_machines.end())
382 it = list_of_machines.begin();
387 INFOS("[getMachineFile] Error resource_name not found in resourcesList -> " << resource_name);
389 else if (std::string(parallelLib) == "Mpi")
391 MESSAGE("[getMachineFile] parallelLib is Mpi");
393 MapOfParserResourcesType resourcesList = _rm->GetList();
394 if (resourcesList.find(std::string(resource_name)) != resourcesList.end())
396 ParserResourcesType resource = resourcesList[std::string(resource_name)];
397 // Check if resource is cluster or not
398 if (resource.ClusterMembersList.empty())
400 //It is not a cluster so we create a cluster with one machine
401 ParserResourcesType fake_node;
402 fake_node.HostName = resource.HostName;
403 fake_node.Protocol = resource.Protocol;
404 fake_node.ClusterInternalProtocol = resource.ClusterInternalProtocol;
405 fake_node.UserName = resource.UserName;
406 fake_node.AppliPath = resource.AppliPath;
407 fake_node.DataForSort = resource.DataForSort;
409 resource.ClusterMembersList.push_front(fake_node);
412 // Choose mpi implementation -> each MPI implementation has is own machinefile...
413 if (resource.mpi == lam)
415 // Creating machine file
416 machine_file_name = tmpnam(NULL);
417 std::ofstream machine_file(machine_file_name.c_str(), std::ios_base::out);
419 // We add all cluster machines to the file
420 std::list<ParserResourcesType>::iterator cluster_it =
421 resource.ClusterMembersList.begin();
422 while (cluster_it != resource.ClusterMembersList.end())
424 unsigned int number_of_proc = (*cluster_it).DataForSort._nbOfNodes *
425 (*cluster_it).DataForSort._nbOfProcPerNode;
426 machine_file << (*cluster_it).HostName << " cpu=" << number_of_proc << std::endl;
430 else if ((resource.mpi == openmpi) || (resource.mpi == ompi))
432 // Creating machine file
433 machine_file_name = tmpnam(NULL);
434 std::ofstream machine_file(machine_file_name.c_str(), std::ios_base::out);
436 // We add all cluster machines to the file
437 std::list<ParserResourcesType>::iterator cluster_it =
438 resource.ClusterMembersList.begin();
439 while (cluster_it != resource.ClusterMembersList.end())
441 unsigned int number_of_proc = (*cluster_it).DataForSort._nbOfNodes *
442 (*cluster_it).DataForSort._nbOfProcPerNode;
443 machine_file << (*cluster_it).HostName << " slots=" << number_of_proc << std::endl;
447 else if (resource.mpi == nompi)
449 INFOS("[getMachineFile] Error resource_name MPI implementation was defined for " << resource_name);
452 INFOS("[getMachineFile] Error resource_name MPI implementation not currently handled for " << resource_name);
455 INFOS("[getMachineFile] Error resource_name not found in resourcesList -> " << resource_name);
458 INFOS("[getMachineFile] Error parallelLib is not handled -> " << parallelLib);
460 return CORBA::string_dup(machine_file_name.c_str());