1 // Copyright (C) 2007-2012 CEA/DEN, EDF R&D, 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.
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 * BatchManager_eSGE.cxx : emulation of SGE client
25 * Auteur : Bernard SECHER - CEA DEN
26 * Mail : mailto:bernard.secher@cea.fr
27 * Date : Thu Apr 24 10:17:22 2008
49 #include <Batch_Constants.hxx>
50 #include <Batch_Utils.hxx>
51 #include <Batch_NotYetImplementedException.hxx>
53 #include "Batch_BatchManager_eSGE.hxx"
54 #include "Batch_JobInfo_eSGE.hxx"
60 BatchManager_eSGE::BatchManager_eSGE(const FactBatchManager * parent, const char * host,
61 const char * username,
62 CommunicationProtocolType protocolType, const char * mpiImpl)
63 : BatchManager(parent, host, username, protocolType, mpiImpl)
69 BatchManager_eSGE::~BatchManager_eSGE()
74 // Methode pour le controle des jobs : soumet un job au gestionnaire
75 const JobId BatchManager_eSGE::submitJob(const Job & job)
77 Parametre params = job.getParametre();
78 const std::string workDir = params[WORKDIR];
79 const string fileToExecute = params[EXECUTABLE];
80 string::size_type p1 = fileToExecute.find_last_of("/");
81 string::size_type p2 = fileToExecute.find_last_of(".");
82 std::string fileNameToExecute = fileToExecute.substr(p1+1,p2-p1-1);
84 // export input files on cluster
85 exportInputFiles(job);
87 // build batch script for job
88 buildBatchScript(job);
90 // define command to submit batch
91 string subCommand = string("bash -l -c \\\"cd ") + workDir + "; qsub " + fileNameToExecute + "_Batch.sh\\\"";
92 string command = _protocol.getExecCommand(subCommand, _hostname, _username);
94 cerr << command.c_str() << endl;
98 int status = Utils::getCommandOutput(command, output);
100 if (status != 0) throw RunTimeException("Can't submit job, error was: " + output);
102 // find id of submitted job in output
104 istringstream iss(output);
105 iss >> strjob >> strjob >> strjob;
107 JobId id(this, strjob);
111 // Ce manager permet de faire de la reprise
113 BatchManager_eSGE::addJob(const Batch::Job & job, const std::string reference)
115 return JobId(this, reference);
118 // Methode pour le controle des jobs : retire un job du gestionnaire
119 void BatchManager_eSGE::deleteJob(const JobId & jobid)
123 istringstream iss(jobid.getReference());
126 // define command to delete batch
127 string subCommand = string("bash -l -c \\\"qdel ") + iss.str() + string("\\\"");
128 string command = _protocol.getExecCommand(subCommand, _hostname, _username);
129 cerr << command.c_str() << endl;
130 status = system(command.c_str());
132 throw RunTimeException("Error of connection on remote host");
134 cerr << "jobId = " << ref << "killed" << endl;
137 // Methode pour le controle des jobs : suspend un job en file d'attente
138 void BatchManager_eSGE::holdJob(const JobId & jobid)
140 throw NotYetImplementedException("BatchManager_eSGE::holdJob");
143 // Methode pour le controle des jobs : relache un job suspendu
144 void BatchManager_eSGE::releaseJob(const JobId & jobid)
146 throw NotYetImplementedException("BatchManager_eSGE::releaseJob");
150 // Methode pour le controle des jobs : modifie un job en file d'attente
151 void BatchManager_eSGE::alterJob(const JobId & jobid, const Parametre & param, const Environnement & env)
153 throw NotYetImplementedException("BatchManager_eSGE::alterJob");
156 // Methode pour le controle des jobs : modifie un job en file d'attente
157 void BatchManager_eSGE::alterJob(const JobId & jobid, const Parametre & param)
159 alterJob(jobid, param, Environnement());
162 // Methode pour le controle des jobs : modifie un job en file d'attente
163 void BatchManager_eSGE::alterJob(const JobId & jobid, const Environnement & env)
165 alterJob(jobid, Parametre(), env);
168 // Methode pour le controle des jobs : renvoie l'etat du job
169 JobInfo BatchManager_eSGE::queryJob(const JobId & jobid)
172 istringstream iss(jobid.getReference());
175 // define command to query batch
176 string subCommand = string("bash -l -c \\\"qstat | grep ") + iss.str() + string("\\\"");
177 string command = _protocol.getExecCommand(subCommand, _hostname, _username);
178 cerr << command.c_str() << endl;
181 int status = Utils::getCommandOutput(command, output);
182 if (status && status != 256)
183 throw RunTimeException("Error of connection on remote host");
185 JobInfo_eSGE ji = JobInfo_eSGE(id, output);
189 // Methode pour le controle des jobs : teste si un job est present en machine
190 bool BatchManager_eSGE::isRunning(const JobId & jobid)
192 throw NotYetImplementedException("BatchManager_eSGE::isRunning");
195 void BatchManager_eSGE::buildBatchScript(const Job & job)
198 //TODO porting on Win32 platform
199 std::cerr << "BuildBatchScript" << std::endl;
200 Parametre params = job.getParametre();
204 string fileToExecute = "";
210 // Mandatory parameters
211 if (params.find(WORKDIR) != params.end())
212 workDir = params[WORKDIR].str();
214 throw RunTimeException("params[WORKDIR] is not defined ! Please defined it, cannot submit this job");
215 if (params.find(EXECUTABLE) != params.end())
216 fileToExecute = params[EXECUTABLE].str();
218 throw RunTimeException("params[EXECUTABLE] is not defined ! Please defined it, cannot submit this job");
220 // Optional parameters
221 if (params.find(NBPROC) != params.end())
222 nbproc = params[NBPROC];
223 if (params.find(MAXWALLTIME) != params.end())
224 edt = params[MAXWALLTIME];
225 if (params.find(MAXRAMSIZE) != params.end())
226 mem = params[MAXRAMSIZE];
227 if (params.find(QUEUE) != params.end())
228 queue = params[QUEUE].str();
230 string::size_type p1 = fileToExecute.find_last_of("/");
231 string::size_type p2 = fileToExecute.find_last_of(".");
232 string rootNameToExecute = fileToExecute.substr(p1+1,p2-p1-1);
233 string fileNameToExecute = fileToExecute.substr(p1+1);
235 // Create batch submit file
236 ofstream tempOutputFile;
237 std::string TmpFileName = Utils::createAndOpenTemporaryFile("SGE-script", tempOutputFile);
239 tempOutputFile << "#! /bin/sh -f" << endl;
241 tempOutputFile << "#$ -q " << queue << endl;
242 tempOutputFile << "#$ -pe " << _mpiImpl->name() << " " << nbproc << endl;
244 tempOutputFile << "#$ -l h_rt=" << getWallTime(edt) << endl ;
246 tempOutputFile << "#$ -l h_vmem=" << mem << "M" << endl ;
247 tempOutputFile << "#$ -o " << workDir << "/logs/output.log." << rootNameToExecute << endl ;
248 tempOutputFile << "#$ -e " << workDir << "/logs/error.log." << rootNameToExecute << endl ;
250 // Abstraction of PBS_NODEFILE - TODO
251 tempOutputFile << "export LIBBATCH_NODEFILE=$TMPDIR/machines" << endl;
253 // Launch the executable
254 tempOutputFile << "cd " << workDir << endl ;
255 tempOutputFile << "./" + fileNameToExecute << endl;
256 tempOutputFile.flush();
257 tempOutputFile.close();
259 BATCH_CHMOD(TmpFileName.c_str(), 0x1ED);
260 cerr << "Batch script file generated is: " << TmpFileName.c_str() << endl;
262 int status = _protocol.copyFile(TmpFileName, "", "",
263 workDir + "/" + rootNameToExecute + "_Batch.sh",
264 _hostname, _username);
266 throw RunTimeException("Error of connection on remote host");
271 std::string BatchManager_eSGE::getWallTime(const long edt)
278 oss << h << ":" << m;
280 oss << h << ":0" << m;
281 oss << ":00"; // the seconds