def copy_history(self, dest, job_ids=None): """Copy history logs from master. Args: dest (str): The path of the local dir where the logs will be copied. job_ids (list of str, optional): A list with the ids of the jobs for which the history should be copied. If nothing is passed, the history of all jobs is copied. """ if not os.path.exists(dest): logger.warning("Destination directory " + dest + " does not exist. It will be created") os.makedirs(dest) history_dir = os.path.join(self.logs_dir, "history") if job_ids: pattern = " -o ".join("-name " + jid + "*" for jid in job_ids) list_dirs = SshProcess("find " + history_dir + " " + pattern, self.master) list_dirs.run() else: list_dirs = SshProcess("find " + history_dir + " -name job_*", self.master) list_dirs.run() remote_files = [] for line in list_dirs.stdout.splitlines(): remote_files.append(line) action = Get([self.master], remote_files, dest) action.run()
def install_os(reconfigure, tags = None): update_config_state() # Clone or pull Kolla if os.path.isdir('kolla'): logger.info("Remove previous Kolla installation") kolla_path = os.path.join(SCRIPT_PATH, "kolla") call("rm -rf %s" % kolla_path, shell=True) logger.info("Cloning Kolla") call("cd %s ; git clone %s -b %s > /dev/null" % (SCRIPT_PATH, KOLLA_REPO, KOLLA_BRANCH), shell=True) logger.warning("Patching kolla, this should be \ deprecated with the new version of Kolla") playbook = os.path.join(SCRIPT_PATH, "ansible/patches.yml") inventory_path = os.path.join(SYMLINK_NAME, 'multinode') run_ansible([playbook], inventory_path, STATE['config']) kolla_cmd = [os.path.join(SCRIPT_PATH, "kolla", "tools", "kolla-ansible")] if reconfigure: kolla_cmd.append('reconfigure') else: kolla_cmd.append('deploy') kolla_cmd.extend(["-i", "%s/multinode" % SYMLINK_NAME, "--passwords", "%s/passwords.yml" % SYMLINK_NAME, "--configdir", "%s" % SYMLINK_NAME]) if tags is not None: kolla_cmd.extend(["--tags", args]) call(kolla_cmd)
def workflow(self, comb): """ """ comb_ok = False logger.info(slugify(comb) + \ ' starts to compile') try: export = "source /opt/intel/bin/compilervars.sh intel64; " src_directory = "/home/arrouan/workspace/aevol/git/world/aevol/" bin_directory = "/home/arrouan/workspace/aevol/compiled_binary/" configure_option = "--with-tracing --without-x" if comb['parallel'] == 'tbb': configure_option += " --with-tbb" if comb['blas'] == 'openblas': configure_option += " --with-blas" elif comb['blas'] == 'mkl': configure_option += " --with-mkl" elif comb['blas'] == 'atlas': configure_option += " --with-atlas" if comb['experiment'] == 'raevol': configure_option += " --with-raevol" if comb['compilator'] == 'intel': configure_option += " CXX=icc" full_bin_directory = bin_directory + comb['experiment']+'_'+comb['compilator']+'_'+comb['parallel']+'_'+comb['blas'] try: os.mkdir(full_bin_directory) except: for f in os.listdir(full_bin_directory): os.remove(full_bin_directory + "/" + f) p = Process(export+'cd '+src_directory+'; autoreconf; ./configure '+configure_option+'; make clean; make; cp src/aevol_run '+full_bin_directory+'/; cp src/aevol_create '+full_bin_directory+'/') p.shell = True # p.run() print p.stdout comb_ok = True finally: if comb_ok: self.sweeper.done(comb) logger.info(slugify(comb) + \ ' has been done') else: self.sweeper.cancel(comb) logger.warning(slugify(comb) + \ ' has been canceled') logger.info(style.step('%s Remaining'), len(self.sweeper.get_remaining()))
def workflow(self, comb): """ Compute one application launch using a given parameter group """ comb_ok = False try: # Generate configuration file needed by MPI processes logger.info("Generating assembly file...") py = comb['cores'] / comb['px'] prepare = Process('cd %s && python %s %d %d %d %d %d %s app.lad' % (self.workingPath, self.genLadScript, comb['datasize'], comb['datasize'], comb['datasize'], comb['px'], py, comb['transposition'])) prepare.shell = True prepare.run() # Generate the MPI host file mfile = self.generate_machine_file() # Start L2C lad = "./app.lad" logger.info("Computing...") res = Process("export OAR_JOB_KEY_FILE=~/.oar_key ; cd %s && l2c_loader -M,-machinefile,%s --mpi -c %d %s" % (self.workingPath, mfile, comb['cores'], lad)) res.shell = True res.stdout_handlers.append(os.path.join(self.result_dir, slugify(comb) + '.out')) res.stdout_handlers.append(sys.stdout) res.stderr_handlers.append(os.path.join(self.result_dir, slugify(comb) + '.err')) res.stderr_handlers.append(sys.stderr) res.run() if not res.ok: logger.error('Bad L2C termination') raise Exception('Bad L2C termination') if len(res.stderr) > 0: # WARNING: when L2C cannot find the LAD file or something strange like this logger.warning('Not empty error output') # Clean configuration files logger.info("Removing assembly files...") res = Process('cd %s && rm -f app.lad*' % self.workingPath) res.shell = True res.run() comb_ok = True except Exception: pass finally: if comb_ok: self.sweeper.done(comb) logger.info(style.host(slugify(comb)) + ' has been done') else: self.sweeper.cancel(comb) logger.warning(style.host(slugify(comb)) + ' has been canceled') logger.info(style.step('%s Remaining'), len(self.sweeper.get_remaining()))
def workflow(self, comb, host, comb_dir): """ """ comb_ok = False thread_name = style.Thread(host.split('.')[0]) + ': ' logger.info(thread_name + 'Starting combination ' + slugify(comb)) try: logger.info(thread_name + 'Generate conf file') param_str = self.create_string(comb) Remote( "python /home/Work/sgcbntier/paasage_demo/xml_gen_execo.py --cb " + param_str, [host]).run() logger.info(thread_name + 'Run code') Remote( "cd /home/Work/sgcbntier/paasage_demo/ ; python run_all_execo.py --cb %s" % param_str, [host]).run() logger.info(thread_name + 'Get results') traceFile = "ntier_" + param_str get_results = Get([host], [ "/home/Work/sgcbntier/paasage_demo/csv/REQTASK_" + traceFile + ".csv" ], local_location=comb_dir).run() for p in get_results.processes: if not p.ok: logger.error( host + ': Unable to retrieve the files for combination %s', slugify(comb)) exit() comb_ok = True finally: if comb_ok: self.sweeper.done(comb) logger.info(thread_name + ': ' + slugify(comb) + \ ' has been done') else: self.sweeper.cancel(comb) logger.warning(thread_name + ': ' + slugify(comb) + \ ' has been canceled') logger.info(style.step('%s Remaining'), len(self.sweeper.get_remaining()))
def copy_history(self, dest, job_ids=None): """Copy history logs from dfs. Args: dest (str): The path of the local dir where the logs will be copied. job_ids (list of str, optional): A list with the ids of the jobs for which the history should be copied. If nothing is passed, the history of all jobs is copied. """ if not os.path.exists(dest): logger.warning("Destination directory " + dest + " does not exist. It will be created") os.makedirs(dest) # Dirs used user_login = getpass.getuser() hist_dfs_dir = "/tmp/hadoop-yarn/staging/history/done_intermediate/" + \ user_login hist_tmp_dir = "/tmp/hadoop_hist" # Remove file in tmp dir if exists proc = SshProcess("rm -rf " + hist_tmp_dir, self.master) proc.run() # Get files in master if job_ids: proc = SshProcess("mkdir " + hist_tmp_dir, self.master) proc.run() for jid in job_ids: self.execute("fs -get " + hist_dfs_dir + "/" + jid + "* " + hist_tmp_dir, verbose=False) else: self.execute("fs -get " + hist_dfs_dir + " " + hist_tmp_dir, verbose=False) # Copy files from master action = Get([self.master], [hist_tmp_dir], dest) action.run()
def run(self): """ """ token = 'bRIJb9jp5igAAAAAAAAACc5QzQ619Vp0pYa2PdIrt0q2y0qFyJgwrKvtzuTp3Sz_' client = dropbox.client.DropboxClient(token) parameters = {'size': igeom(128, 2048, 5), 'db_if': ['rest', 'sdk']} combs = sweep(parameters) sweeper = ParamSweeper(self.result_dir + "/sweeps", combs) f = open(self.result_dir + '/results.txt', 'w') while len(sweeper.get_remaining()) > 0: comb = sweeper.get_next() logger.info('Treating combination %s', pformat(comb)) comb_dir = self.result_dir + '/' + slugify(comb) try: os.mkdir(comb_dir) except: pass fname = self.create_file(comb['size']) timer = Timer() if comb['db_if'] == 'sdk': self.upload_file_sdk(client, fname, fname.split('/')[-1]) up_time = timer.elapsed() self.download_file_sdk(client, fname.split('/')[-1], comb_dir + fname.split('/')[-1]) dl_time = timer.elapsed() - up_time sweeper.done(comb) elif comb['db_if'] == 'rest': logger.warning('REST interface not implemented') sweeper.skip(comb) continue os.remove(fname) f.write("%f %i %f %f \n" % (timer.start_date(), comb['size'], up_time, dl_time)) f.close()
def install_os(reconfigure, tags=None): update_config_state() # Clone or pull Kolla if os.path.isdir('kolla'): logger.info("Remove previous Kolla installation") kolla_path = os.path.join(SCRIPT_PATH, "kolla") call("rm -rf %s" % kolla_path, shell=True) logger.info("Cloning Kolla") call("cd %s ; git clone %s -b %s > /dev/null" % (SCRIPT_PATH, KOLLA_REPO, KOLLA_BRANCH), shell=True) logger.warning("Patching kolla, this should be \ deprecated with the new version of Kolla") playbook = os.path.join(SCRIPT_PATH, "ansible/patches.yml") inventory_path = os.path.join(SYMLINK_NAME, 'multinode') run_ansible([playbook], inventory_path, STATE['config']) kolla_cmd = [os.path.join(SCRIPT_PATH, "kolla", "tools", "kolla-ansible")] if reconfigure: kolla_cmd.append('reconfigure') else: kolla_cmd.append('deploy') kolla_cmd.extend([ "-i", "%s/multinode" % SYMLINK_NAME, "--configdir", "%s" % SYMLINK_NAME ]) if tags is not None: kolla_cmd.extend(["--tags", args]) call(kolla_cmd)
def run(self): sweeper = self.create_paramsweeper() while True: comb = sweeper.get_next() if not comb: break comb_dir = self.result_dir + '/' + slugify(comb) if not os.path.isdir(comb_dir): os.mkdir(comb_dir) comb_file = comb_dir + '/trace' g5k_configuration['kadeploy3'] = comb['version'] logger.info('Treating combination %s', pformat(comb)) get_version = SshProcess( comb['version'] + ' -v', comb['site'], connection_params=default_frontend_connection_params).run() logger.info(get_version.stdout) resources = "" if comb['kavlan']: resources += "{type='kavlan'}/vlan=1+" resources += "nodes=" + str(comb['n_nodes']) sub = OarSubmission(resources=resources, job_type='deploy', walltime="0:30:00", name='Kadeploy_Tests') logger.info('Performing submission of %s on site %s', resources, comb['site']) jobs = oarsub([(sub, comb['site'])]) if jobs[0][0]: try: logger.info('Waiting for job to start') wait_oar_job_start(jobs[0][0], jobs[0][1]) hosts = get_oar_job_nodes(jobs[0][0], jobs[0][1]) logger.info('Deployment of %s', ' '.join([host.address for host in hosts])) kavlan = get_oar_job_kavlan(jobs[0][0], jobs[0][1]) if kavlan: logger.info('In kavlan %s', kavlan) deployment = Deployment(hosts, env_name=comb['env'], vlan=kavlan) deployed, undeployed = deploy(deployment, stdout_handlers=[comb_file], stderr_handlers=[comb_file]) finally: logger.info('Destroying job %s on %s', str(jobs[0][0]), jobs[0][1]) oardel([(jobs[0][0], jobs[0][1])]) else: deployed = [] if len(undeployed) == 0: logger.info('%s is OK', slugify(comb)) elif len(deployed) == 0: logger.error('%s is KO', slugify(comb)) else: logger.warning('%s encountered problems with some hosts', slugify(comb)) sweeper.done(comb)
def run(self): """ run method from engine in order to do our workflow """ mongo = ClientMongo() size = dict if not self.options.file: if not self.options.only: size = { 1, long(self.options.size * 0.25), long(self.options.size * 0.5), long(self.options.size * 0.75), long(self.options.size) } else: size = {long(self.options.size)} else: if self.OnlyDownload: size = getFilSize(self.options.file) else: size = {0} drive = None if self.options.drive: drive = self.options.drive else: drive = self.drive interface = ['rest', 'sdk'] parameters = { 'size': size, 'if': interface, 'drive': drive, 'transfert': self.transfert } p = None for n in range(0, int(self.options.ntest), 1): logger.info('---------------------') logger.info('Round %i', n + 1) combs = sweep(parameters) date = datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S") pathResults = os.getcwd() + '/Results/Bench' + date sweeper = ParamSweeper(pathResults + "/sweeps", combs) f = open(pathResults + '/results.txt', 'w') while len(sweeper.get_remaining()) > 0: # sort the parameters for i in interface: for dr in drive: for s in size: comb = sweeper.get_next(filtr=lambda r: filter( lambda x: x['drive'] == dr and x['size'] == s and x['if'] == i, r)) if not comb: continue # start of the workflow if comb['drive'] == 'amazon': p = providerS3.ProviderS3() elif comb['drive'] == 'dropbox': p = providerDB.ProviderDB() else: p = providerGD.ProviderGD() logger.info('Treating combination %s', pformat(comb)) comb_dir = pathResults + '/' + slugify(comb) if not os.path.isdir(comb_dir): os.mkdir(comb_dir) if not self.options.file: fname = self.create_file(comb['size']) else: fname = self.options.file timer = Timer() up_time = 0 dl_time = 0 start_date = datetime.datetime.now() if comb['if'] == 'sdk': if p.provider_name == "amazon": # AMAZON clientAmz = p.getConnexion() if self.OnlyDownload: p.bucketKey += fname else: p.bucketKey += '/' + fname if comb['transfert'] == "upload" or comb[ 'transfert'] == 'upDown': p.upload_file_sdk( clientAmz.get_bucket(p.bucketName), p.bucketKey, fname) up_time = timer.elapsed() if comb['transfert'] == "download" or comb[ 'transfert'] == 'upDown': p.download_file_sdk( clientAmz.get_bucket(p.bucketName), p.bucketKey, comb_dir + '/' + fname.split('/')[-1]) dl_time = timer.elapsed() - up_time if not self.OnlyDownload: p.delete_file_sdk( clientAmz.get_bucket(p.bucketName), p.bucketKey) elif p.provider_name == "dropbox": # DROPBOX client = p.getToken() if comb['transfert'] == "upload" or comb[ 'transfert'] == 'upDown': p.upload_file_sdk( client, fname, fname.split('/')[-1]) up_time = timer.elapsed() if comb['transfert'] == "download" or comb[ 'transfert'] == 'upDown': p.download_file_sdk( client, fname.split('/')[-1], comb_dir + '/' + fname.split('/')[-1]) dl_time = timer.elapsed() - up_time if not self.OnlyDownload: p.delete_file(client, fname.split('/')[-1]) elif p.provider_name == "googledrive": # GOOGLEDRIVE drive_service = p.getConnexion() new_file = None if comb['transfert'] == 'upload' or comb[ 'transfert'] == 'upDown': new_file = p.upload_file_sdk( drive_service, fname, fname.split('/')[-1], 'text/plain') up_time = timer.elapsed() if comb['transfert'] == 'download' or comb[ 'transfert'] == 'upDown': p.download_file_sdk( drive_service, new_file, comb_dir + '/' + fname.split('/')[-1]) dl_time = timer.elapsed() - up_time if not self.OnlyDownload: p.delete_file_sdk( drive_service, new_file['id']) sweeper.done(comb) elif comb['if'] == 'rest': logger.warning( 'REST interface not implemented') sweeper.skip(comb) if not self.OnlyDownload: # logger.info('delete de '+fname) if os.path.isfile(fname): os.remove(fname) # delete only if rest is implmented # os.remove(comb_dir + '/' + fname.split('/')[-1]) continue if comb['transfert'] == "upload" or comb[ 'transfert'] == "upDown": f.write("%s %s %s %s %s %s %s %f %i %s %f\n" % (self.localisation['ip'], self.localisation['lat'], self.localisation['lon'], self.localisation['city'], self.localisation['country'], comb['drive'], comb['if'], timer.start_date(), comb['size'], "upload", up_time)) mongo.collection.insert({ 'ip': self.localisation['ip'], 'latitude': self.localisation['lat'], 'longitude': self.localisation['lon'], 'city': self.localisation['city'], 'country': self.localisation['country'], 'drive': comb['drive'], 'interface': comb['if'], 'start_date': start_date, 'size': comb['size'], 'transfert': 'upload', 'time': up_time }) if comb['transfert'] == "download" or comb[ 'transfert'] == "upDown": f.write("%s %s %s %s %s %s %s %f %i %s %f\n" % (self.localisation['ip'], self.localisation['lat'], self.localisation['lon'], self.localisation['city'], self.localisation['country'], comb['drive'], comb['if'], timer.start_date(), comb['size'], "download", dl_time)) mongo.collection.insert({ 'ip': self.localisation['ip'], 'latitude': self.localisation['lat'], 'longitude': self.localisation['lon'], 'city': self.localisation['city'], 'country': self.localisation['country'], 'drive': comb['drive'], 'interface': comb['if'], 'start_date': start_date, 'size': comb['size'], 'transfert': 'download', 'time': dl_time }) if not self.OnlyDownload: # logger.info('delete de '+fname) if os.path.isfile(fname): os.remove(fname) if os.path.isfile(comb_dir + '/' + fname): os.remove(comb_dir + '/' + fname.split('/')[-1]) f.close() # delete the Bench Folder os.rmdir(self.result_dir) logger.info("---------------------------------------") for t in check_Exp_database(self.options, self.localisation)['result']: logger.info(t)
def workflow(self, comb): """ Compute one application launch using a given parameter group """ comb_ok = False try: # Generate configuration file needed by MPI processes logger.info("Generating assembly file...") py = comb['cores'] / comb['px'] prepare = Process( 'cd %s && python %s %d %d %d %d %d %s app.lad' % (self.workingPath, self.genLadScript, comb['datasize'], comb['datasize'], comb['datasize'], comb['px'], py, comb['transposition'])) prepare.shell = True prepare.run() # Generate the MPI host file mfile = self.generate_machine_file() # Start L2C lad = "./app.lad" logger.info("Computing...") res = Process( "export OAR_JOB_KEY_FILE=~/.oar_key ; cd %s && l2c_loader -M,-machinefile,%s --mpi -c %d %s" % (self.workingPath, mfile, comb['cores'], lad)) res.shell = True res.stdout_handlers.append( os.path.join(self.result_dir, slugify(comb) + '.out')) res.stdout_handlers.append(sys.stdout) res.stderr_handlers.append( os.path.join(self.result_dir, slugify(comb) + '.err')) res.stderr_handlers.append(sys.stderr) res.run() if not res.ok: logger.error('Bad L2C termination') raise Exception('Bad L2C termination') if len( res.stderr ) > 0: # WARNING: when L2C cannot find the LAD file or something strange like this logger.warning('Not empty error output') # Clean configuration files logger.info("Removing assembly files...") res = Process('cd %s && rm -f app.lad*' % self.workingPath) res.shell = True res.run() comb_ok = True except Exception: pass finally: if comb_ok: self.sweeper.done(comb) logger.info(style.host(slugify(comb)) + ' has been done') else: self.sweeper.cancel(comb) logger.warning( style.host(slugify(comb)) + ' has been canceled') logger.info(style.step('%s Remaining'), len(self.sweeper.get_remaining()))
def workflow(self, comb, host, comb_dir): """ """ comb_ok = False thread_name = style.Thread(str(host).split('.')[0]) + ': ' logger.info(thread_name + 'Starting combination ' + slugify(comb)) try: self.export = "source ~/aevol_binary/intel/linux/mkl/bin/mklvars.sh intel64; " bucketname = self.working_dir + '/raevol_5_mut_lat/' + slugify( comb) + '/' if os.path.isdir(bucketname) and os.path.exists(bucketname + '/last_gener.txt'): logger.info(thread_name + "Resuming AEVOL from NFS backup") gen_file = open(bucketname + '/last_gener.txt', 'r') last_gen = gen_file.read() if int(last_gen) < 500000: logger.info(thread_name + "Resuming AEVOL Run from " + str(int(last_gen))) rem = Remote( self.export + 'cd ' + bucketname + '; /home/jorouzaudcornabas/aevol_binary/aevol/src/aevol_run -p 16' + ' -e 300000 -r ' + last_gen + ' >> aevol_run.log', [host]).run() if rem.ok: comb_ok = True else: comb_ok = True else: Remote('mkdir -p ' + bucketname, [host]).run() param_file = '/home/jorouzaudcornabas/aevol_binary/execo/mut_lat/param_tmpl.in' logger.info(thread_name + 'Generate config file ' + param_file) f_template = open(param_file) fd, outfile = mkstemp(dir='/tmp/', prefix=slugify(comb) + '_param') f = os.fdopen(fd, 'w') for line in f_template: line = line.replace('SEED_NUMBER', str(comb['seed'])) line = line.replace('FUZZY_VERSION', str(comb['fuzzy'])) if comb['move']: line = line.replace('FIRST_GAUSSIAN_MEDIAN', '0.25') line = line.replace('THIRD_GAUSSIAN_MEDIAN', '0.65') else: line = line.replace('FIRST_GAUSSIAN_MEDIAN', '0.2') line = line.replace('THIRD_GAUSSIAN_MEDIAN', '0.6') line = line.replace('GAUSSIAN_HEIGHT', str(comb['height'])) f.write(line) f_template.close() f.close() put_file = Put([host], [outfile], remote_location=bucketname).run() if not put_file.ok: exit() os.remove(outfile) Remote( 'cd ' + bucketname + '; cp ' + outfile.split('/')[-1] + ' param.in', [host]).run() logger.info(thread_name + "Launching AEVOL Create") Remote( self.export + 'cd ' + bucketname + '; /home/jorouzaudcornabas/aevol_diff_area/aevol/src/aevol_create > aevol_create.log', [host]).run() logger.info(thread_name + "Launching AEVOL Run") rem = Remote( self.export + 'cd ' + bucketname + '; /home/jorouzaudcornabas/aevol_diff_area/aevol/src/aevol_run -p 16 -n 500000 > aevol_run.log', [host]).run() if rem.ok: comb_ok = True logger.info(thread_name + 'Get results ' + comb_dir + "/" + slugify(comb)) #try: #os.mkdir(comb_dir + "/" + slugify(comb)) #except: #logger.warning(thread_name + #'%s already exists, removing existing files', comb_dir + "/" + slugify(comb)) #shutil.rmtree(comb_dir+ "/" + slugify(comb)) #try: #os.mkdir(comb_dir + "/" + slugify(comb)) #except: #logger.warning(thread_name + #'%s already exists, recreating directory', comb_dir + "/" + slugify(comb)) #get_results = Get([host], [bucketname+ "/aevol_create.log", bucketname+ "/aevol_run.log", bucketname+'/stats/'], #local_location=comb_dir + "/" + slugify(comb)).run() #for p in get_results.processes: #if not p.ok: #logger.error(thread_name + #': Unable to retrieve the files for combination %s', #slugify(comb)) #exit() finally: if comb_ok: self.sweeper.done(comb) # shutil.rmtree(bucketname) logger.info(thread_name + ': ' + slugify(comb) + \ ' has been done') else: self.sweeper.cancel(comb) logger.warning(thread_name + ': ' + slugify(comb) + \ ' has been canceled') logger.info(style.step('%s Remaining'), len(self.sweeper.get_remaining()))
def workflow(self, comb, host, comb_dir): """ """ comb_ok = False thread_name = style.Thread(str(host).split('.')[0]) + ': ' logger.info(thread_name + 'Starting combination ' + slugify(comb)) if 'parapluie' in str(host): nb_proc = 24 elif 'paranoia' in str(host): nb_proc = 20 elif 'parapide' in str(host): nb_proc = 8 else: nb_proc = 16 try: self.export = "source ~/aevol_binary/intel/linux/mkl/bin/mklvars.sh intel64; " bucketname = self.working_dir + '/raevol_5_mut_lat/' + slugify( comb) + '/' logger.info(thread_name + "Killing other RAevol") killa = Remote("killall -9 aevol_run", [host]) for killp in killa.processes: killp.ignore_error = True killa.run() if os.path.isdir(bucketname) and os.path.exists(bucketname + '/last_gener.txt'): logger.info(thread_name + "Resuming AEVOL from NFS backup") gen_file = open(bucketname + '/last_gener.txt', 'r') last_gen = gen_file.read() if int(last_gen) < 300000: logger.info(thread_name + "Resuming AEVOL Run from " + str(int(last_gen))) rem = Remote( self.export + 'cd ' + bucketname + '; /home/jorouzaudcornabas/aevol_binary/aevol/src/aevol_run -p ' + str(nb_proc) + ' -e 300000 -r ' + last_gen + ' >> aevol_run.log', [host], process_args={ 'default_stdout_handler': False, 'default_stderr_handler': False }).run() if rem.ok: comb_ok = True else: comb_ok = True else: Remote('mkdir -p ' + bucketname, [host]).run() param_file = '/home/jorouzaudcornabas/aevol_binary/aevol/execo/mut_lat/param_tmpl.in' logger.info(thread_name + 'Generate config file ' + param_file) f_template = open(param_file) fd, outfile = mkstemp(dir='/tmp/', prefix=slugify(comb) + '_param') f = os.fdopen(fd, 'w') for line in f_template: if 'CONFIGURE_ENVIRONMENT_VALUES' in line: if comb['env'] == 'const': line = line.replace('CONFIGURE_ENVIRONMENT_VALUES', 'NB_ENVIRONMENTS 1') f.write(line) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.8 0.05' + os.linesep) elif comb['env'] == 'lat_3': line = line.replace('CONFIGURE_ENVIRONMENT_VALUES', 'NB_ENVIRONMENTS 2') f.write(line) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.8 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 2 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 2 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 2 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 2 0.5 0.8 0.05' + os.linesep) elif comb['env'] == 'lat_all': line = line.replace('CONFIGURE_ENVIRONMENT_VALUES', 'NB_ENVIRONMENTS 16') f.write(line) #const f.write('ENV_ADD_GAUSSIAN 1 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 1 0.5 0.8 0.05' + os.linesep) # 1 f.write('ENV_ADD_GAUSSIAN 2 0.5 0.25 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 2 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 2 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 2 0.5 0.8 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 3 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 3 0.5 0.45 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 3 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 3 0.5 0.8 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 4 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 4 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 4 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 4 0.5 0.8 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 5 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 5 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 5 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 5 0.5 0.85 0.05' + os.linesep) # 2 f.write('ENV_ADD_GAUSSIAN 6 0.5 0.25 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 6 0.5 0.45 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 6 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 6 0.5 0.8 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 7 0.5 0.25 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 7 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 7 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 7 0.5 0.8 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 8 0.5 0.25 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 8 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 8 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 8 0.5 0.85 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 9 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 9 0.5 0.45 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 9 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 9 0.5 0.8 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 10 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 10 0.5 0.45 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 10 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 10 0.5 0.85 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 11 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 11 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 11 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 11 0.5 0.85 0.05' + os.linesep) # 3 f.write('ENV_ADD_GAUSSIAN 12 0.5 0.25 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 12 0.5 0.45 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 12 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 12 0.5 0.8 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 13 0.5 0.25 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 13 0.5 0.45 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 13 0.5 0.6 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 13 0.5 0.85 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 14 0.5 0.25 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 14 0.5 0.4 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 14 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 14 0.5 0.85 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 15 0.5 0.2 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 15 0.5 0.45 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 15 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 15 0.5 0.85 0.05' + os.linesep) # 4 f.write('ENV_ADD_GAUSSIAN 16 0.5 0.25 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 16 0.5 0.45 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 16 0.5 0.65 0.05' + os.linesep) f.write('ENV_ADD_GAUSSIAN 16 0.5 0.85 0.05' + os.linesep) elif 'CONFIGURE_SIGNAL_VALUES' in line: if comb['env'] == 'const': line = line.replace('CONFIGURE_SIGNAL_VALUES', '') f.write(line) elif comb['env'] == 'lat_3': line = line.replace( 'CONFIGURE_SIGNAL_VALUES', 'CREATE_SIGNAL h0 h0 h0 w0 m0 m1 m0 h1 h0 m0 h0 m1 h1 w0 h1 h0 m1 h1 m0 w0 w0 m0 w0 h0 h1 m1 w0 m0 m1 m0 w0 h1 h0 m0 h0 m1 h1 w0 h0 w0 m0 m1 m0 w0 h1 h0 w0 w0 h1' ) f.write(line) f.write('ENV_ADD_SIGNAL 2 1' + os.linesep) elif comb['env'] == 'lat_all': line = line.replace( 'CONFIGURE_SIGNAL_VALUES', 'CREATE_SIGNAL h0 w0 h1 m1 w0 h1 m0 h0 h1 w0 h0 m1 h1 h1 m1 m0 h0 w0 h1 m1 w0 h1 m0 h0 h1 w0 h0 m1 h1 h1 m1 m0 h1 m0 m1' ) f.write(line) f.write( 'CREATE_SIGNAL m0 h0 m1 h1 m1 w0 m0 m1 m0 h0 m1 h1 w0 h0 h0 h1 m1 m0 h1 w0 h1 h0 m1 h1 m0 w0 w0 m0 m1 w0 w0 h1 h0 w0 h1 h0 h0 m0 h0 w0 h0 m1 m0 w0 h1 w0 w0 h1 m0' + os.linesep) f.write( 'CREATE_SIGNAL h0 h0 h0 w0 m0 m1 m0 h1 h0 m0 h0 m1 h1 w0 h1 h0 m1 h1 m0 w0 w0 m0 w0 h0 h1 m1 w0 m0 m1 m0 w0 h1 h0 m0 h0 m1 h1 w0 h0 w0 m0 m1 m0 w0 h1 h0 w0 w0 h1' + os.linesep) f.write( 'CREATE_SIGNAL h1 h1 m0 w0 w0 h1 m1 h1 h1 m1 m0 w0 m1 m0 m0 w0 m0 h0 m0 h0 w0 h0 m0 h0 h1 m1 h0 h1 w0 h0 h1 m1 h1 m1 m0' + os.linesep) f.write('ENV_ADD_SIGNAL 2 1' + os.linesep) f.write('ENV_ADD_SIGNAL 3 2' + os.linesep) f.write('ENV_ADD_SIGNAL 4 3' + os.linesep) f.write('ENV_ADD_SIGNAL 5 4' + os.linesep) f.write('ENV_ADD_SIGNAL 6 1' + os.linesep) f.write('ENV_ADD_SIGNAL 6 2' + os.linesep) f.write('ENV_ADD_SIGNAL 7 1' + os.linesep) f.write('ENV_ADD_SIGNAL 7 3' + os.linesep) f.write('ENV_ADD_SIGNAL 8 1' + os.linesep) f.write('ENV_ADD_SIGNAL 8 4' + os.linesep) f.write('ENV_ADD_SIGNAL 9 2' + os.linesep) f.write('ENV_ADD_SIGNAL 9 3' + os.linesep) f.write('ENV_ADD_SIGNAL 10 2' + os.linesep) f.write('ENV_ADD_SIGNAL 10 4' + os.linesep) f.write('ENV_ADD_SIGNAL 11 3' + os.linesep) f.write('ENV_ADD_SIGNAL 11 4' + os.linesep) f.write('ENV_ADD_SIGNAL 12 1' + os.linesep) f.write('ENV_ADD_SIGNAL 12 2' + os.linesep) f.write('ENV_ADD_SIGNAL 12 3' + os.linesep) f.write('ENV_ADD_SIGNAL 13 1' + os.linesep) f.write('ENV_ADD_SIGNAL 13 2' + os.linesep) f.write('ENV_ADD_SIGNAL 13 4' + os.linesep) f.write('ENV_ADD_SIGNAL 14 1' + os.linesep) f.write('ENV_ADD_SIGNAL 14 3' + os.linesep) f.write('ENV_ADD_SIGNAL 14 4' + os.linesep) f.write('ENV_ADD_SIGNAL 15 2' + os.linesep) f.write('ENV_ADD_SIGNAL 15 3' + os.linesep) f.write('ENV_ADD_SIGNAL 15 4' + os.linesep) f.write('ENV_ADD_SIGNAL 16 1' + os.linesep) f.write('ENV_ADD_SIGNAL 16 2' + os.linesep) f.write('ENV_ADD_SIGNAL 16 3' + os.linesep) f.write('ENV_ADD_SIGNAL 16 4' + os.linesep) else: line = line.replace('SEED_NUMBER', str(comb['seed'])) line = line.replace('MUTATION_RATE_VALUE', comb['mutation']) line = line.replace('SELECTION_PRESSURE', str(comb['selection'])) f.write(line) f_template.close() f.close() put_file = Put([host], [outfile], remote_location=bucketname).run() if not put_file.ok: exit() os.remove(outfile) Remote( 'cd ' + bucketname + '; cp ' + outfile.split('/')[-1] + ' param.in; cp /home/jorouzaudcornabas/aevol_binary/aevol/execo/mut_lat/binding_matrix.rae .', [host]).run() logger.info(thread_name + "Launching AEVOL Create") Remote( self.export + 'cd ' + bucketname + '; /home/jorouzaudcornabas/aevol_binary/aevol/src/aevol_create > aevol_create.log', [host], process_args={ 'default_stdout_handler': False, 'default_stderr_handler': False }).run() logger.info(thread_name + "Launching AEVOL Run") rem = Remote( self.export + 'cd ' + bucketname + '; /home/jorouzaudcornabas/aevol_binary/aevol/src/aevol_run -p ' + str(nb_proc) + ' -n 300000 > aevol_run.log', [host], process_args={ 'default_stdout_handler': False, 'default_stderr_handler': False }).run() if rem.ok: comb_ok = True logger.info(thread_name + 'Get results ' + comb_dir + "/" + slugify(comb)) #try: #os.mkdir(comb_dir + "/" + slugify(comb)) #except: #logger.warning(thread_name + #'%s already exists, removing existing files', comb_dir + "/" + slugify(comb)) #shutil.rmtree(comb_dir+ "/" + slugify(comb)) #try: #os.mkdir(comb_dir + "/" + slugify(comb)) #except: #logger.warning(thread_name + #'%s already exists, recreating directory', comb_dir + "/" + slugify(comb)) #get_results = Get([host], [bucketname+ "/aevol_create.log", bucketname+ "/aevol_run.log", bucketname+'/stats/'], #local_location=comb_dir + "/" + slugify(comb)).run() #for p in get_results.processes: #if not p.ok: #logger.error(thread_name + #': Unable to retrieve the files for combination %s', #slugify(comb)) #exit() finally: if comb_ok: self.sweeper.done(comb) # shutil.rmtree(bucketname) logger.info(thread_name + ': ' + slugify(comb) + \ ' has been done') else: self.sweeper.cancel(comb) logger.warning(thread_name + ': ' + slugify(comb) + \ ' has been canceled') logger.info(style.step('%s Remaining'), len(self.sweeper.get_remaining()))
def prepare_node(conf_file, force_deploy, tags): g5k = G5kEngine(conf_file, force_deploy) g5k.start(args=[]) STATE['config'].update(g5k.load()) g5k.get_job() deployed, undeployed = g5k.deploy() if len(undeployed) > 0: sys.exit(31) roles = g5k.build_roles() # Get an IP for # kolla (haproxy) # docker registry # influx db # grafana vip_addresses = g5k.get_free_ip(5) # Get the NIC devices of the reserved cluster # XXX: this only works if all nodes are on the same cluster, # or if nodes from different clusters have the same devices interfaces = g5k.get_cluster_nics(STATE['config']['resources'].keys()[0]) network_interface = str(interfaces[0]) external_interface = None if len(interfaces) > 1: external_interface = str(interfaces[1]) else: external_interface = 'veth0' logger.warning("%s has only one NIC. The same interface " "will be used for network_interface and " "neutron_external_interface." % STATE['config']['resources'].keys()[0]) g5k.exec_command_on_nodes( g5k.deployed_nodes, 'apt-get update && apt-get -y --force-yes install apt-transport-https', 'Installing apt-transport-https...') # Install python on the nodes g5k.exec_command_on_nodes(g5k.deployed_nodes, 'apt-get -y install python', 'Installing Python on all the nodes...') # Generates files for ansible/kolla inventory_path = os.path.join(g5k.result_dir, 'multinode') base_inventory = STATE['config']['inventory'] generate_inventory(roles, base_inventory, inventory_path) # Symlink current directory link = os.path.abspath(SYMLINK_NAME) try: os.remove(link) except OSError: pass os.symlink(g5k.result_dir, link) logger.info("Symlinked %s to %s" % (g5k.result_dir, link)) STATE['config'].update({ 'vip': str(vip_addresses[0]), 'registry_vip': str(vip_addresses[1]), 'influx_vip': str(vip_addresses[2]), 'grafana_vip': str(vip_addresses[3]), 'network_interface': network_interface }) # Run the Ansible playbooks playbook_path = os.path.join(SCRIPT_PATH, 'ansible', 'prepare-node.yml') passwords_path = os.path.join(TEMPLATE_DIR, "passwords.yml") inventory_path = os.path.join(SYMLINK_NAME, 'multinode') config = STATE['config'].copy() with open(passwords_path) as passwords_file: config.update(yaml.load(passwords_file)) kolla_vars = { 'kolla_internal_vip_address': str(vip_addresses[0]), 'network_interface': network_interface, 'neutron_external_interface': external_interface, 'enable_veth': external_interface == 'veth0', 'neutron_external_address': str(vip_addresses[4]) } config.update(kolla_vars) run_ansible([playbook_path], inventory_path, config, tags) # Generating Ansible globals.yml, passwords.yml generate_kolla_files(g5k.config["kolla"], kolla_vars, g5k.result_dir) # Fills the state and save it in the `current` directory # TODO: Manage STATE at __main__ level STATE['config_file'] = conf_file STATE['nodes'] = roles STATE['user'] = g5k.user
def prepare_node(conf_file, force_deploy, tags): g5k = G5kEngine(conf_file, force_deploy) g5k.start(args=[]) STATE['config'].update(g5k.load()) g5k.get_job() deployed, undeployed = g5k.deploy() if len(undeployed) > 0: sys.exit(31) roles = g5k.build_roles() # Get an IP for # kolla (haproxy) # docker registry # influx db # grafana vip_addresses = g5k.get_free_ip(5) # Get the NIC devices of the reserved cluster # XXX: this only works if all nodes are on the same cluster, # or if nodes from different clusters have the same devices interfaces = g5k.get_cluster_nics(STATE['config']['resources'].keys()[0]) network_interface = str(interfaces[0]) external_interface = None if len(interfaces) > 1: external_interface = str(interfaces[1]) else: external_interface = 'veth0' logger.warning("%s has only one NIC. The same interface " "will be used for network_interface and " "neutron_external_interface." % STATE['config']['resources'].keys()[0]) g5k.exec_command_on_nodes( g5k.deployed_nodes, 'apt-get update && apt-get -y --force-yes install apt-transport-https', 'Installing apt-transport-https...') # Install python on the nodes g5k.exec_command_on_nodes( g5k.deployed_nodes, 'apt-get -y install python', 'Installing Python on all the nodes...') # Generates files for ansible/kolla inventory_path = os.path.join(g5k.result_dir, 'multinode') base_inventory = STATE['config']['inventory'] generate_inventory(roles, base_inventory, inventory_path) # Symlink current directory link = os.path.abspath(SYMLINK_NAME) try: os.remove(link) except OSError: pass os.symlink(g5k.result_dir, link) logger.info("Symlinked %s to %s" % (g5k.result_dir, link)) STATE['config'].update({ 'vip': str(vip_addresses[0]), 'registry_vip': str(vip_addresses[1]), 'influx_vip': str(vip_addresses[2]), 'grafana_vip': str(vip_addresses[3]), 'network_interface': network_interface }) # Run the Ansible playbooks playbook_path = os.path.join(SCRIPT_PATH, 'ansible', 'prepare-node.yml') passwords_path = os.path.join(TEMPLATE_DIR, "passwords.yml") inventory_path = os.path.join(SYMLINK_NAME, 'multinode') config = STATE['config'].copy() with open(passwords_path) as passwords_file: config.update(yaml.load(passwords_file)) kolla_vars = { 'kolla_internal_vip_address' : str(vip_addresses[0]), 'network_interface' : network_interface, 'neutron_external_interface' : external_interface, 'enable_veth' : external_interface == 'veth0', 'neutron_external_address' : str(vip_addresses[4]) } config.update(kolla_vars) run_ansible([playbook_path], inventory_path, config, tags) # Generating Ansible globals.yml, passwords.yml generate_kolla_files(g5k.config["kolla"], kolla_vars, g5k.result_dir) # Fills the state and save it in the `current` directory # TODO: Manage STATE at __main__ level STATE['config_file'] = conf_file STATE['nodes'] = roles STATE['user'] = g5k.user
def run(self): sweeper = self.create_paramsweeper() while True: comb = sweeper.get_next() if not comb: break comb_dir = self.result_dir + '/' + slugify(comb) if not os.path.isdir(comb_dir): os.mkdir(comb_dir) comb_file = comb_dir + '/trace' g5k_configuration['kadeploy3'] = comb['version'] logger.info('Treating combination %s', pformat(comb)) get_version = SshProcess(comb['version'] + ' -v', comb['site'], connection_params=default_frontend_connection_params).run() logger.info(get_version.stdout) resources = "" if comb['kavlan']: resources += "{type='kavlan'}/vlan=1+" resources += "nodes=" + str(comb['n_nodes']) sub = OarSubmission(resources=resources, job_type='deploy', walltime="0:30:00", name='Kadeploy_Tests') logger.info('Performing submission of %s on site %s', resources, comb['site']) jobs = oarsub([(sub, comb['site'])]) if jobs[0][0]: try: logger.info('Waiting for job to start') wait_oar_job_start(jobs[0][0], jobs[0][1]) hosts = get_oar_job_nodes(jobs[0][0], jobs[0][1]) logger.info('Deployment of %s', ' '.join([host.address for host in hosts])) kavlan = get_oar_job_kavlan(jobs[0][0], jobs[0][1]) if kavlan: logger.info('In kavlan %s', kavlan) deployment = Deployment(hosts, env_name=comb['env'], vlan=kavlan) deployed, undeployed = deploy(deployment, stdout_handlers=[comb_file], stderr_handlers=[comb_file]) finally: logger.info('Destroying job %s on %s', str(jobs[0][0]), jobs[0][1]) oardel([(jobs[0][0], jobs[0][1])]) else: deployed = [] if len(undeployed) == 0: logger.info('%s is OK', slugify(comb)) elif len(deployed) == 0: logger.error('%s is KO', slugify(comb)) else: logger.warning('%s encountered problems with some hosts', slugify(comb)) sweeper.done(comb)