X-Git-Url: http://git.uio.no/git/?a=blobdiff_plain;f=ANALYSIS%2FAliAnalysisAlien.cxx;h=82f6d977ee81f4c04e59766f607701d25287b4ef;hb=ea7b474069195c842b2122d9aeb6e4715f4a341f;hp=a09a3129048542be3ba0297fe9cfd37b33ee7a80;hpb=652561d5ab372df6166d9cc9735ae91b260282f4;p=u%2Fmrichter%2FAliRoot.git diff --git a/ANALYSIS/AliAnalysisAlien.cxx b/ANALYSIS/AliAnalysisAlien.cxx index a09a3129048..82f6d977ee8 100644 --- a/ANALYSIS/AliAnalysisAlien.cxx +++ b/ANALYSIS/AliAnalysisAlien.cxx @@ -27,6 +27,8 @@ #include "TROOT.h" #include "TSystem.h" #include "TFile.h" +#include "TFileCollection.h" +#include "TChain.h" #include "TObjString.h" #include "TObjArray.h" #include "TGrid.h" @@ -56,6 +58,7 @@ AliAnalysisAlien::AliAnalysisAlien() fNtestFiles(0), fNrunsPerMaster(0), fMaxMergeFiles(0), + fMaxMergeStages(0), fNsubmitted(0), fProductionMode(0), fOutputToRunNo(0), @@ -63,6 +66,9 @@ AliAnalysisAlien::AliAnalysisAlien() fFastReadOption(0), fOverwriteMode(1), fNreplicas(2), + fNproofWorkers(0), + fNproofWorkersPerSlave(0), + fProofReset(0), fRunNumbers(), fExecutable(), fExecutableCommand(), @@ -70,6 +76,7 @@ AliAnalysisAlien::AliAnalysisAlien() fExecutableArgs(), fAnalysisMacro(), fAnalysisSource(), + fValidationScript(), fAdditionalRootLibs(), fAdditionalLibs(), fSplitMode(), @@ -87,6 +94,7 @@ AliAnalysisAlien::AliAnalysisAlien() fInputFormat(), fDatasetName(), fJDLName(), + fTerminateFiles(), fMergeExcludes(), fIncludePath(), fCloseSE(), @@ -94,6 +102,11 @@ AliAnalysisAlien::AliAnalysisAlien() fJobTag(), fOutputSingle(), fRunPrefix(), + fProofCluster(), + fProofDataSet(), + fFileForTestMode(), + fRootVersionForProof(), + fAliRootMode(), fInputFiles(0), fPackages(0) { @@ -114,6 +127,7 @@ AliAnalysisAlien::AliAnalysisAlien(const char *name) fNtestFiles(0), fNrunsPerMaster(0), fMaxMergeFiles(0), + fMaxMergeStages(0), fNsubmitted(0), fProductionMode(0), fOutputToRunNo(0), @@ -121,6 +135,9 @@ AliAnalysisAlien::AliAnalysisAlien(const char *name) fFastReadOption(0), fOverwriteMode(1), fNreplicas(2), + fNproofWorkers(0), + fNproofWorkersPerSlave(0), + fProofReset(0), fRunNumbers(), fExecutable(), fExecutableCommand(), @@ -128,6 +145,7 @@ AliAnalysisAlien::AliAnalysisAlien(const char *name) fExecutableArgs(), fAnalysisMacro(), fAnalysisSource(), + fValidationScript(), fAdditionalRootLibs(), fAdditionalLibs(), fSplitMode(), @@ -145,6 +163,7 @@ AliAnalysisAlien::AliAnalysisAlien(const char *name) fInputFormat(), fDatasetName(), fJDLName(), + fTerminateFiles(), fMergeExcludes(), fIncludePath(), fCloseSE(), @@ -152,6 +171,11 @@ AliAnalysisAlien::AliAnalysisAlien(const char *name) fJobTag(), fOutputSingle(), fRunPrefix(), + fProofCluster(), + fProofDataSet(), + fFileForTestMode(), + fRootVersionForProof(), + fAliRootMode(), fInputFiles(0), fPackages(0) { @@ -172,6 +196,7 @@ AliAnalysisAlien::AliAnalysisAlien(const AliAnalysisAlien& other) fNtestFiles(other.fNtestFiles), fNrunsPerMaster(other.fNrunsPerMaster), fMaxMergeFiles(other.fMaxMergeFiles), + fMaxMergeStages(other.fMaxMergeStages), fNsubmitted(other.fNsubmitted), fProductionMode(other.fProductionMode), fOutputToRunNo(other.fOutputToRunNo), @@ -179,6 +204,9 @@ AliAnalysisAlien::AliAnalysisAlien(const AliAnalysisAlien& other) fFastReadOption(other.fFastReadOption), fOverwriteMode(other.fOverwriteMode), fNreplicas(other.fNreplicas), + fNproofWorkers(other.fNproofWorkers), + fNproofWorkersPerSlave(other.fNproofWorkersPerSlave), + fProofReset(other.fProofReset), fRunNumbers(other.fRunNumbers), fExecutable(other.fExecutable), fExecutableCommand(other.fExecutableCommand), @@ -186,6 +214,7 @@ AliAnalysisAlien::AliAnalysisAlien(const AliAnalysisAlien& other) fExecutableArgs(other.fExecutableArgs), fAnalysisMacro(other.fAnalysisMacro), fAnalysisSource(other.fAnalysisSource), + fValidationScript(other.fValidationScript), fAdditionalRootLibs(other.fAdditionalRootLibs), fAdditionalLibs(other.fAdditionalLibs), fSplitMode(other.fSplitMode), @@ -203,6 +232,7 @@ AliAnalysisAlien::AliAnalysisAlien(const AliAnalysisAlien& other) fInputFormat(other.fInputFormat), fDatasetName(other.fDatasetName), fJDLName(other.fJDLName), + fTerminateFiles(other.fTerminateFiles), fMergeExcludes(other.fMergeExcludes), fIncludePath(other.fIncludePath), fCloseSE(other.fCloseSE), @@ -210,6 +240,11 @@ AliAnalysisAlien::AliAnalysisAlien(const AliAnalysisAlien& other) fJobTag(other.fJobTag), fOutputSingle(other.fOutputSingle), fRunPrefix(other.fRunPrefix), + fProofCluster(other.fProofCluster), + fProofDataSet(other.fProofDataSet), + fFileForTestMode(other.fFileForTestMode), + fRootVersionForProof(other.fRootVersionForProof), + fAliRootMode(other.fAliRootMode), fInputFiles(0), fPackages(0) { @@ -260,6 +295,7 @@ AliAnalysisAlien &AliAnalysisAlien::operator=(const AliAnalysisAlien& other) fNtestFiles = other.fNtestFiles; fNrunsPerMaster = other.fNrunsPerMaster; fMaxMergeFiles = other.fMaxMergeFiles; + fMaxMergeStages = other.fMaxMergeStages; fNsubmitted = other.fNsubmitted; fProductionMode = other.fProductionMode; fOutputToRunNo = other.fOutputToRunNo; @@ -267,6 +303,9 @@ AliAnalysisAlien &AliAnalysisAlien::operator=(const AliAnalysisAlien& other) fFastReadOption = other.fFastReadOption; fOverwriteMode = other.fOverwriteMode; fNreplicas = other.fNreplicas; + fNproofWorkers = other.fNproofWorkers; + fNproofWorkersPerSlave = other.fNproofWorkersPerSlave; + fProofReset = other.fProofReset; fRunNumbers = other.fRunNumbers; fExecutable = other.fExecutable; fExecutableCommand = other.fExecutableCommand; @@ -274,6 +313,7 @@ AliAnalysisAlien &AliAnalysisAlien::operator=(const AliAnalysisAlien& other) fExecutableArgs = other.fExecutableArgs; fAnalysisMacro = other.fAnalysisMacro; fAnalysisSource = other.fAnalysisSource; + fValidationScript = other.fValidationScript; fAdditionalRootLibs = other.fAdditionalRootLibs; fAdditionalLibs = other.fAdditionalLibs; fSplitMode = other.fSplitMode; @@ -291,6 +331,7 @@ AliAnalysisAlien &AliAnalysisAlien::operator=(const AliAnalysisAlien& other) fInputFormat = other.fInputFormat; fDatasetName = other.fDatasetName; fJDLName = other.fJDLName; + fTerminateFiles = other.fTerminateFiles; fMergeExcludes = other.fMergeExcludes; fIncludePath = other.fIncludePath; fCloseSE = other.fCloseSE; @@ -298,6 +339,11 @@ AliAnalysisAlien &AliAnalysisAlien::operator=(const AliAnalysisAlien& other) fJobTag = other.fJobTag; fOutputSingle = other.fOutputSingle; fRunPrefix = other.fRunPrefix; + fProofCluster = other.fProofCluster; + fProofDataSet = other.fProofDataSet; + fFileForTestMode = other.fFileForTestMode; + fRootVersionForProof = other.fRootVersionForProof; + fAliRootMode = other.fAliRootMode; if (other.fInputFiles) { fInputFiles = new TObjArray(); TIter next(other.fInputFiles); @@ -363,6 +409,7 @@ Bool_t AliAnalysisAlien::Connect() { // Try to connect to AliEn. User needs a valid token and /tmp/gclient_env_$UID sourced. if (gGrid && gGrid->IsConnected()) return kTRUE; + if (fProductionMode) return kTRUE; if (!gGrid) { Info("Connect", "Trying to connect to AliEn ..."); TGrid::Connect("alien://"); @@ -394,9 +441,9 @@ void AliAnalysisAlien::CdWork() gGrid->Cd(homedir); if (gGrid->Mkdir(workdir, "-p")) { gGrid->Cd(fGridWorkingDir); - Info("CreateJDL", "\n##### Created alien working directory %s", fGridWorkingDir.Data()); + Info("CdWork", "\n##### Created alien working directory %s", fGridWorkingDir.Data()); } else { - Warning("CreateJDL", "Working directory %s cannot be created.\n Using %s instead.", + Warning("CdWork", "Working directory %s cannot be created.\n Using %s instead.", workdir.Data(), homedir.Data()); fGridWorkingDir = ""; } @@ -406,6 +453,7 @@ void AliAnalysisAlien::CdWork() Bool_t AliAnalysisAlien::CheckFileCopy(const char *alienpath) { // Check if file copying is possible. + if (fProductionMode) return kTRUE; if (!Connect()) { Error("CheckFileCopy", "Not connected to AliEn. File copying cannot be tested."); return kFALSE; @@ -454,12 +502,15 @@ Bool_t AliAnalysisAlien::CheckFileCopy(const char *alienpath) Bool_t AliAnalysisAlien::CheckInputData() { // Check validity of input data. If necessary, create xml files. + if (fProductionMode) return kTRUE; if (!fInputFiles && !fRunNumbers.Length() && !fRunRange[0]) { if (!fGridDataDir.Length()) { Error("CkeckInputData", "AliEn path to base data directory must be set.\n = Use: SetGridDataDir()"); return kFALSE; } Info("CheckInputData", "Analysis will make a single xml for base data directory %s",fGridDataDir.Data()); + if (fDataPattern.Contains("tag") && TestBit(AliAnalysisGrid::kTest)) + TObject::SetBit(AliAnalysisGrid::kUseTags, kTRUE); // ADDED (fix problem in determining the tag usage in test mode) return kTRUE; } // Process declared files @@ -467,7 +518,7 @@ Bool_t AliAnalysisAlien::CheckInputData() Bool_t isXml = kFALSE; Bool_t useTags = kFALSE; Bool_t checked = kFALSE; - CdWork(); + if (!TestBit(AliAnalysisGrid::kTest)) CdWork(); TString file; TString workdir = gGrid->GetHomeDirectory(); workdir += fGridWorkingDir; @@ -555,7 +606,7 @@ Bool_t AliAnalysisAlien::CheckInputData() msg += " type: xml_collection;"; if (useTags) msg += " using_tags: Yes"; else msg += " using_tags: No"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); if (fNrunsPerMaster<2) { AddDataFile(Form("%s.xml", os->GetString().Data())); } else { @@ -583,7 +634,7 @@ Bool_t AliAnalysisAlien::CheckInputData() msg += " type: xml_collection;"; if (useTags) msg += " using_tags: Yes"; else msg += " using_tags: No"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); if (fNrunsPerMaster<2) { AddDataFile(Form("%s%d.xml",fRunPrefix.Data(),irun)); } else { @@ -609,14 +660,14 @@ Bool_t AliAnalysisAlien::CheckInputData() Bool_t AliAnalysisAlien::CreateDataset(const char *pattern) { // Create dataset for the grid data directory + run number. - if (TestBit(AliAnalysisGrid::kOffline)) return kTRUE; + if (fProductionMode || TestBit(AliAnalysisGrid::kOffline)) return kTRUE; if (!Connect()) { Error("CreateDataset", "Cannot create dataset with no grid connection"); return kFALSE; } // Cd workspace - CdWork(); + if (!TestBit(AliAnalysisGrid::kTest)) CdWork(); TString workdir = gGrid->GetHomeDirectory(); workdir += fGridWorkingDir; @@ -887,7 +938,7 @@ Bool_t AliAnalysisAlien::CreateJDL() Bool_t error = kFALSE; TObjArray *arr = 0; Bool_t copy = kTRUE; - if (TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; + if (fProductionMode || TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; Bool_t generate = kTRUE; if (TestBit(AliAnalysisGrid::kTest) || TestBit(AliAnalysisGrid::kSubmit)) generate = kFALSE; if (!Connect()) { @@ -895,8 +946,9 @@ Bool_t AliAnalysisAlien::CreateJDL() return kFALSE; } // Check validity of alien workspace - CdWork(); - TString workdir = gGrid->GetHomeDirectory(); + TString workdir; + if (!fProductionMode && !fGridWorkingDir.BeginsWith("/alice")) workdir = gGrid->GetHomeDirectory(); + if (!fProductionMode && !TestBit(AliAnalysisGrid::kTest)) CdWork(); workdir += fGridWorkingDir; if (generate) { TObjString *os; @@ -915,16 +967,18 @@ Bool_t AliAnalysisAlien::CreateJDL() Error("CreateJDL", "You must define AliEn output directory"); error = kTRUE; } else { - if (!fGridOutputDir.Contains("/")) fGridOutputDir = Form("%s/%s", workdir.Data(), fGridOutputDir.Data()); - if (!DirectoryExists(fGridOutputDir)) { - if (gGrid->Mkdir(fGridOutputDir,"-p")) { - Info("CreateJDL", "\n##### Created alien output directory %s", fGridOutputDir.Data()); - } else { - Error("CreateJDL", "Could not create alien output directory %s", fGridOutputDir.Data()); - // error = kTRUE; - } - } - gGrid->Cd(workdir); + if (!fProductionMode) { + if (!fGridOutputDir.Contains("/")) fGridOutputDir = Form("%s/%s", workdir.Data(), fGridOutputDir.Data()); + if (!DirectoryExists(fGridOutputDir)) { + if (gGrid->Mkdir(fGridOutputDir,"-p")) { + Info("CreateJDL", "\n##### Created alien output directory %s", fGridOutputDir.Data()); + } else { + Error("CreateJDL", "Could not create alien output directory %s", fGridOutputDir.Data()); + // error = kTRUE; + } + } + gGrid->Cd(workdir); + } } // Exit if any error up to now if (error) return kFALSE; @@ -941,7 +995,9 @@ Bool_t AliAnalysisAlien::CreateJDL() fMergingJDL->AddToInputSandbox(Form("LF:%s/%s", workdir.Data(),mergeExec.Data()), "List of input files to be uploaded to workers"); if (!fArguments.IsNull()) fGridJDL->SetArguments(fArguments, "Arguments for the executable command"); - fMergingJDL->SetArguments("$1 $2 $3"); + if (IsOneStageMerging()) fMergingJDL->SetArguments(fGridOutputDir); + else fMergingJDL->SetArguments("wn.xml $2 $3"); // xml, stage, laststage(0 or 1) + fGridJDL->SetValue("TTL", Form("\"%d\"",fTTL)); fGridJDL->SetDescription("TTL", Form("Time after which the job is killed (%d min.)", fTTL/60)); fMergingJDL->SetValue("TTL", Form("\"%d\"",fTTL)); @@ -954,11 +1010,17 @@ Bool_t AliAnalysisAlien::CreateJDL() if (fSplitMaxInputFileNumber > 0) { fGridJDL->SetValue("SplitMaxInputFileNumber", Form("\"%d\"", fSplitMaxInputFileNumber)); fGridJDL->SetDescription("SplitMaxInputFileNumber", "Maximum number of input files to be processed per subjob"); + } + if (!IsOneStageMerging()) { + fMergingJDL->SetValue("SplitMaxInputFileNumber", "\"$3\""); + fMergingJDL->SetDescription("SplitMaxInputFileNumber", "Maximum number of input files to be merged in one go"); } if (fSplitMode.Length()) { fGridJDL->SetValue("Split", Form("\"%s\"", fSplitMode.Data())); fGridJDL->SetDescription("Split", "We split per SE or file"); - } + } + fMergingJDL->SetValue("Split", "\"se\""); + fMergingJDL->SetDescription("Split", "We split per SE for merging in stages"); if (!fAliROOTVersion.IsNull()) { fGridJDL->AddToPackages("AliRoot", fAliROOTVersion,"VO_ALICE", "List of requested packages"); fMergingJDL->AddToPackages("AliRoot", fAliROOTVersion, "VO_ALICE", "List of requested packages"); @@ -986,6 +1048,8 @@ Bool_t AliAnalysisAlien::CreateJDL() } fGridJDL->SetInputDataListFormat(fInputFormat, "Format of input data"); fGridJDL->SetInputDataList("wn.xml", "Collection name to be processed on each worker node"); + fMergingJDL->SetInputDataListFormat(fInputFormat, "Format of input data"); + fMergingJDL->SetInputDataList("wn.xml", "Collection name to be processed on each worker node"); fGridJDL->AddToInputSandbox(Form("LF:%s/%s", workdir.Data(), fAnalysisMacro.Data()), "List of input files to be uploaded to workers"); TString analysisFile = fExecutable; analysisFile.ReplaceAll(".sh", ".root"); @@ -1026,16 +1090,29 @@ Bool_t AliAnalysisAlien::CreateJDL() first = kFALSE; } delete arr; - TString outputArchive = fOutputArchive; - if (!fMergeExcludes.IsNull()) { - arr = fMergeExcludes.Tokenize(" "); - TIter next1(arr); - while ((os=(TObjString*)next1())) { - outputArchive.ReplaceAll(Form("%s,",os->GetString().Data()),""); - outputArchive.ReplaceAll(os->GetString(),""); + // Output archive for the merging jdl + TString outputArchive; + if (TestBit(AliAnalysisGrid::kDefaultOutputs)) { + outputArchive = "log_archive.zip:std*,*.stat@disk=1 "; + // Add normal output files, extra files + terminate files + TString files = GetListOfFiles("outextter"); + // Do not register merge excludes + if (!fMergeExcludes.IsNull()) { + arr = fMergeExcludes.Tokenize(" "); + TIter next1(arr); + while ((os=(TObjString*)next1())) { + files.ReplaceAll(Form("%s,",os->GetString().Data()),""); + files.ReplaceAll(os->GetString(),""); + } + delete arr; } - delete arr; - } + files.ReplaceAll(".root", "*.root"); + outputArchive += Form("root_archive.zip:%s@disk=%d",files.Data(),fNreplicas); + } else { + TString files = fOutputArchive; + files.ReplaceAll(".root", "*.root"); // nreplicas etc should be already atttached by use + outputArchive = files; + } arr = outputArchive.Tokenize(" "); TIter next2(arr); comment = comment1; @@ -1043,10 +1120,8 @@ Bool_t AliAnalysisAlien::CreateJDL() while ((os=(TObjString*)next2())) { if (!first) comment = NULL; TString currentfile = os->GetString(); - currentfile.ReplaceAll(".root", "*.root"); - currentfile.ReplaceAll(".zip", "-Stage$2_$3.zip"); if (!currentfile.Contains("@") && fCloseSE.Length()) - fMergingJDL->AddToOutputArchive(Form("%s@%s",currentfile.Data(), fCloseSE.Data()), comment); + fMergingJDL->AddToOutputArchive(Form("%s@%s",currentfile.Data(), fCloseSE.Data()), comment); else fMergingJDL->AddToOutputArchive(currentfile, comment); first = kFALSE; @@ -1056,11 +1131,12 @@ Bool_t AliAnalysisAlien::CreateJDL() arr = fOutputFiles.Tokenize(","); TIter next(arr); Bool_t first = kTRUE; - const char *comment = "Files to be archived"; - const char *comment1 = comment; + const char *comment = "Files to be saved"; while ((os=(TObjString*)next())) { // Ignore ouputs in jdl that are also in outputarchive TString sout = os->GetString(); + sout.ReplaceAll("*", ""); + sout.ReplaceAll(".root", ""); if (sout.Index("@")>0) sout.Remove(sout.Index("@")); if (fOutputArchive.Contains(sout)) continue; if (!first) comment = NULL; @@ -1068,44 +1144,19 @@ Bool_t AliAnalysisAlien::CreateJDL() fGridJDL->AddToOutputSandbox(Form("%s@%s",os->GetString().Data(), fCloseSE.Data()), comment); else fGridJDL->AddToOutputSandbox(os->GetString(), comment); - first = kFALSE; + first = kFALSE; + if (fMergeExcludes.Contains(sout)) continue; + if (!os->GetString().Contains("@") && fCloseSE.Length()) + fMergingJDL->AddToOutputSandbox(Form("%s@%s",os->GetString().Data(), fCloseSE.Data()), comment); + else + fMergingJDL->AddToOutputSandbox(os->GetString(), comment); } delete arr; - if (fOutputFiles.Length()) { - TString outputFiles = fOutputFiles; - if (!fMergeExcludes.IsNull()) { - arr = fMergeExcludes.Tokenize(" "); - TIter next1(arr); - while ((os=(TObjString*)next1())) { - outputFiles.ReplaceAll(Form("%s,",os->GetString().Data()),""); - outputFiles.ReplaceAll(os->GetString(),""); - } - delete arr; - } - arr = outputFiles.Tokenize(" "); - TIter next2(arr); - comment = comment1; - first = kTRUE; - while ((os=(TObjString*)next2())) { - // Ignore ouputs in jdl that are also in outputarchive - TString sout = os->GetString(); - if (sout.Index("@")>0) sout.Remove(sout.Index("@")); - if (fOutputArchive.Contains(sout)) continue; - if (!first) comment = NULL; - if (!os->GetString().Contains("@") && fCloseSE.Length()) - fMergingJDL->AddToOutputSandbox(Form("%s@%s",os->GetString().Data(), fCloseSE.Data()), comment); - else - fMergingJDL->AddToOutputSandbox(os->GetString(), comment); - } - delete arr; - } fGridJDL->SetPrice((UInt_t)fPrice, "AliEn price for this job"); fMergingJDL->SetPrice((UInt_t)fPrice, "AliEn price for this job"); - TString validationScript = fExecutable; - validationScript.ReplaceAll(".sh", "_validation.sh"); + TString validationScript = fValidationScript; fGridJDL->SetValidationCommand(Form("%s/%s", workdir.Data(),validationScript.Data()), "Validation script to be run for each subjob"); - validationScript = fExecutable; - validationScript.ReplaceAll(".sh", "_mergevalidation.sh"); + validationScript.ReplaceAll(".sh", "_merge.sh"); fMergingJDL->SetValidationCommand(Form("%s/%s", workdir.Data(),validationScript.Data()), "Validation script to be run for each subjob"); if (fMasterResubmitThreshold) { fGridJDL->SetValue("MasterResubmitThreshold", Form("\"%d%%\"", fMasterResubmitThreshold)); @@ -1182,38 +1233,50 @@ Bool_t AliAnalysisAlien::WriteJDL(Bool_t copy) // all run numbers are considered in one go (jdl). For non-negative indices // they correspond to the indices in the array fInputFiles. if (!fInputFiles) return kFALSE; - TObjString *os; - TString workdir = gGrid->GetHomeDirectory(); + TObject *os; + TString workdir; + if (!fProductionMode && !fGridWorkingDir.BeginsWith("/alice")) workdir = gGrid->GetHomeDirectory(); workdir += fGridWorkingDir; + fMergingJDL->AddToInputDataCollection("LF:$1/Stage_$2.xml,nodownload", "Collection of files to be merged for stage $2"); + fMergingJDL->SetOutputDirectory("$1/Stage_$2/#alien_counter_03i#", "Output directory"); - if (!fRunNumbers.Length() && !fRunRange[0]) { - // One jdl with no parameters in case input data is specified by name. + if (fProductionMode) { TIter next(fInputFiles); - while ((os=(TObjString*)next())) - fGridJDL->AddToInputDataCollection(Form("LF:%s,nodownload", os->GetString().Data()), "Input xml collections"); - if (!fOutputSingle.IsNull()) - fGridJDL->SetOutputDirectory(Form("#alienfulldir#/../%s",fOutputSingle.Data()), "Output directory"); - else { - fGridJDL->SetOutputDirectory(Form("%s/#alien_counter_03i#", fGridOutputDir.Data()), "Output directory"); - fMergingJDL->SetOutputDirectory(fGridOutputDir); - } - } else { - // One jdl to be submitted with 2 input parameters: data collection name and output dir prefix - fGridJDL->AddToInputDataCollection(Form("LF:%s/$1,nodownload", workdir.Data()), "Input xml collections"); - if (!fOutputSingle.IsNull()) { - if (!fOutputToRunNo) fGridJDL->SetOutputDirectory(Form("#alienfulldir#/%s",fOutputSingle.Data()), "Output directory"); - else fGridJDL->SetOutputDirectory(Form("%s/$2",fGridOutputDir.Data()), "Output directory"); - } else { - fGridJDL->SetOutputDirectory(Form("%s/$2/#alien_counter_03i#", fGridOutputDir.Data()), "Output directory"); - fMergingJDL->SetOutputDirectory(Form("$1", fGridOutputDir.Data()), "Output directory"); - } + while ((os=next())) { + fGridJDL->AddToInputDataCollection(Form("LF:%s,nodownload", os->GetName()), "Input xml collections"); + } + fGridJDL->SetOutputDirectory(Form("%s/#alien_counter_04i#", fGridOutputDir.Data())); + } else { + if (!fRunNumbers.Length() && !fRunRange[0]) { + // One jdl with no parameters in case input data is specified by name. + TIter next(fInputFiles); + while ((os=next())) + fGridJDL->AddToInputDataCollection(Form("LF:%s,nodownload", os->GetName()), "Input xml collections"); + if (!fOutputSingle.IsNull()) + fGridJDL->SetOutputDirectory(Form("#alienfulldir#/../%s",fOutputSingle.Data()), "Output directory"); + else { + fGridJDL->SetOutputDirectory(Form("%s/#alien_counter_03i#", fGridOutputDir.Data()), "Output directory"); + fMergingJDL->SetOutputDirectory(fGridOutputDir); + } + } else { + // One jdl to be submitted with 2 input parameters: data collection name and output dir prefix + fGridJDL->AddToInputDataCollection(Form("LF:%s/$1,nodownload", workdir.Data()), "Input xml collections"); + if (!fOutputSingle.IsNull()) { + if (!fOutputToRunNo) fGridJDL->SetOutputDirectory(Form("#alienfulldir#/%s",fOutputSingle.Data()), "Output directory"); + else fGridJDL->SetOutputDirectory(Form("%s/$2",fGridOutputDir.Data()), "Output directory"); + } else { + fGridJDL->SetOutputDirectory(Form("%s/$2/#alien_counter_03i#", fGridOutputDir.Data()), "Output directory"); + } + } } - // Generate the JDL as a string TString sjdl = fGridJDL->Generate(); TString sjdl1 = fMergingJDL->Generate(); - Int_t index; + fMergingJDL->SetOutputDirectory("$1", "Output directory"); + fMergingJDL->AddToInputSandbox("LF:$1/$4"); + TString sjdl2 = fMergingJDL->Generate(); + Int_t index, index1; sjdl.ReplaceAll("\"LF:", "\n \"LF:"); sjdl.ReplaceAll("(member", "\n (member"); sjdl.ReplaceAll("\",\"VO_", "\",\n \"VO_"); @@ -1230,54 +1293,123 @@ Bool_t AliAnalysisAlien::WriteJDL(Bool_t copy) sjdl1.ReplaceAll("{\n \n", "{\n"); sjdl1.ReplaceAll("\n\n", "\n"); sjdl1.ReplaceAll("OutputDirectory", "OutputDir"); + sjdl2.ReplaceAll("\"LF:", "\n \"LF:"); + sjdl2.ReplaceAll("(member", "\n (member"); + sjdl2.ReplaceAll("\",\"VO_", "\",\n \"VO_"); + sjdl2.ReplaceAll("{", "{\n "); + sjdl2.ReplaceAll("};", "\n};"); + sjdl2.ReplaceAll("{\n \n", "{\n"); + sjdl2.ReplaceAll("\n\n", "\n"); + sjdl2.ReplaceAll("OutputDirectory", "OutputDir"); sjdl += "JDLVariables = \n{\n \"Packages\",\n \"OutputDir\"\n};\n"; sjdl.Prepend(Form("Jobtag = {\n \"comment:%s\"\n};\n", fJobTag.Data())); index = sjdl.Index("JDLVariables"); if (index >= 0) sjdl.Insert(index, "\n# JDL variables\n"); sjdl += "Workdirectorysize = {\"5000MB\"};"; + sjdl1 += "Workdirectorysize = {\"5000MB\"};"; sjdl1 += "JDLVariables = \n{\n \"Packages\",\n \"OutputDir\"\n};\n"; - sjdl1.Prepend(Form("Jobtag = {\n \"comment:%s_Merging\"\n};\n", fJobTag.Data())); - sjdl1.Prepend("# Generated merging jdl\n# $1 = full alien path to output directory to be merged\n# $2 = merging stage\n# $3 = merged chunk\n"); + index = fJobTag.Index(":"); + if (index < 0) index = fJobTag.Length(); + TString jobTag = fJobTag; + sjdl1.Prepend(Form("Jobtag = {\n \"comment:%s_Merging\"\n};\n", jobTag.Data())); + sjdl1.Prepend("# Generated merging jdl \ + \n# $1 = full alien path to output directory to be merged \ + \n# $2 = merging stage \ + \n# $3 = maximum number of files to merge (must be >= 10000 for the last stage) \ + \n# $4 = xml made via: find *Stage/*root_archive.zip\n"); + sjdl2.Prepend(Form("Jobtag = {\n \"comment:%s_FinalMerging\"\n};\n", jobTag.Data())); + sjdl2.Prepend("# Generated merging jdl \ + \n# $1 = full alien path to output directory to be merged \ + \n# $2 = merging stage \ + \n# $3 = maximum number of files to merge (must be >= 10000 for the last stage) \ + \n# $4 = xml made via: find *Stage/*root_archive.zip\n"); index = sjdl1.Index("JDLVariables"); if (index >= 0) sjdl1.Insert(index, "\n# JDL variables\n"); + index = sjdl2.Index("JDLVariables"); + if (index >= 0) sjdl2.Insert(index, "\n# JDL variables\n"); sjdl1 += "Workdirectorysize = {\"5000MB\"};"; + sjdl2 += "Workdirectorysize = {\"5000MB\"};"; + index = sjdl2.Index("Split ="); + if (index>=0) { + index1 = sjdl2.Index("\n", index); + sjdl2.Remove(index, index1-index+1); + } + index = sjdl2.Index("SplitMaxInputFileNumber"); + if (index>=0) { + index1 = sjdl2.Index("\n", index); + sjdl2.Remove(index, index1-index+1); + } + index = sjdl2.Index("InputDataCollection"); + if (index>=0) { + index1 = sjdl2.Index(";", index); + sjdl2.Remove(index, index1-index+1); + } + index = sjdl2.Index("InputDataListFormat"); + if (index>=0) { + index1 = sjdl2.Index("\n", index); + sjdl2.Remove(index, index1-index+1); + } + index = sjdl2.Index("InputDataList"); + if (index>=0) { + index1 = sjdl2.Index("\n", index); + sjdl2.Remove(index, index1-index+1); + } + sjdl2.ReplaceAll("wn.xml", "$4"); // Write jdl to file ofstream out; out.open(fJDLName.Data(), ios::out); if (out.bad()) { - Error("CreateJDL", "Bad file name: %s", fJDLName.Data()); + Error("WriteJDL", "Bad file name: %s", fJDLName.Data()); return kFALSE; } out << sjdl << endl; + out.close(); TString mergeJDLName = fExecutable; mergeJDLName.ReplaceAll(".sh", "_merge.jdl"); if (fMergeViaJDL) { ofstream out1; out1.open(mergeJDLName.Data(), ios::out); - if (out.bad()) { - Error("CreateJDL", "Bad file name: %s", mergeJDLName.Data()); + if (out1.bad()) { + Error("WriteJDL", "Bad file name: %s", mergeJDLName.Data()); return kFALSE; } out1 << sjdl1 << endl; + out1.close(); + ofstream out2; + TString finalJDL = mergeJDLName; + finalJDL.ReplaceAll(".jdl", "_final.jdl"); + out2.open(finalJDL.Data(), ios::out); + if (out2.bad()) { + Error("WriteJDL", "Bad file name: %s", finalJDL.Data()); + return kFALSE; + } + out2 << sjdl2 << endl; + out2.close(); } // Copy jdl to grid workspace if (!copy) { - Info("CreateJDL", "\n##### You may want to review jdl:%s and analysis macro:%s before running in mode", fJDLName.Data(), fAnalysisMacro.Data()); + Info("WriteJDL", "\n##### You may want to review jdl:%s and analysis macro:%s before running in mode", fJDLName.Data(), fAnalysisMacro.Data()); } else { TString locjdl = Form("%s/%s", fGridOutputDir.Data(),fJDLName.Data()); TString locjdl1 = Form("%s/%s", fGridOutputDir.Data(),mergeJDLName.Data()); + TString finalJDL = mergeJDLName; + finalJDL.ReplaceAll(".jdl", "_final.jdl"); + TString locjdl2 = Form("%s/%s", fGridOutputDir.Data(),finalJDL.Data()); if (fProductionMode) { locjdl = Form("%s/%s", workdir.Data(),fJDLName.Data()); locjdl1 = Form("%s/%s", workdir.Data(),mergeJDLName.Data()); + locjdl2 = Form("%s/%s", workdir.Data(),finalJDL.Data()); } if (FileExists(locjdl)) gGrid->Rm(locjdl); if (FileExists(locjdl1)) gGrid->Rm(locjdl1); - Info("CreateJDL", "\n##### Copying JDL file <%s> to your AliEn output directory", fJDLName.Data()); + if (FileExists(locjdl2)) gGrid->Rm(locjdl2); + Info("WriteJDL", "\n##### Copying JDL file <%s> to your AliEn output directory", fJDLName.Data()); TFile::Cp(Form("file:%s",fJDLName.Data()), Form("alien://%s", locjdl.Data())); if (fMergeViaJDL) { - Info("CreateJDL", "\n##### Copying merging JDL file <%s> to your AliEn output directory", mergeJDLName.Data()); + Info("WriteJDL", "\n##### Copying merging JDL files <%s> to your AliEn output directory", mergeJDLName.Data()); TFile::Cp(Form("file:%s",mergeJDLName.Data()), Form("alien://%s", locjdl1.Data())); + TFile::Cp(Form("file:%s",finalJDL.Data()), Form("alien://%s", locjdl2.Data())); } } return kTRUE; @@ -1288,7 +1420,9 @@ Bool_t AliAnalysisAlien::FileExists(const char *lfn) { // Returns true if file exists. if (!gGrid) return kFALSE; - TGridResult *res = gGrid->Ls(lfn); + TString slfn = lfn; + slfn.ReplaceAll("alien://",""); + TGridResult *res = gGrid->Ls(slfn); if (!res) return kFALSE; TMap *map = dynamic_cast(res->At(0)); if (!map) { @@ -1355,20 +1489,20 @@ void AliAnalysisAlien::CheckDataType(const char *lfn, Bool_t &isCollection, Bool TGridResult *res = gGrid->Command(Form("listFilesFromCollection -z -v %s",lfn), kFALSE); if (!res) { msg += " using_tags: No (unknown)"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); return; } const char* typeStr = res->GetKey(0, "origLFN"); if (!typeStr || !strlen(typeStr)) { msg += " using_tags: No (unknown)"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); return; } TString file = typeStr; useTags = file.Contains(".tag"); if (useTags) msg += " using_tags: Yes"; else msg += " using_tags: No"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); return; } TString slfn(lfn); @@ -1380,13 +1514,13 @@ void AliAnalysisAlien::CheckDataType(const char *lfn, Bool_t &isCollection, Bool TGridCollection *coll = (TGridCollection*)gROOT->ProcessLine(Form("TAlienCollection::Open(\"alien://%s\",1);",lfn)); if (!coll) { msg += " using_tags: No (unknown)"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); return; } TMap *map = coll->Next(); if (!map) { msg += " using_tags: No (unknown)"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); return; } map = (TMap*)map->GetValue(""); @@ -1396,7 +1530,7 @@ void AliAnalysisAlien::CheckDataType(const char *lfn, Bool_t &isCollection, Bool delete coll; if (useTags) msg += " using_tags: Yes"; else msg += " using_tags: No"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); return; } useTags = slfn.Contains(".tag"); @@ -1404,7 +1538,7 @@ void AliAnalysisAlien::CheckDataType(const char *lfn, Bool_t &isCollection, Bool else msg += " type: unknown file;"; if (useTags) msg += " using_tags: Yes"; else msg += " using_tags: No"; - Info("CheckDataType", msg.Data()); + Info("CheckDataType", "%s", msg.Data()); } //______________________________________________________________________________ @@ -1428,6 +1562,50 @@ void AliAnalysisAlien::EnablePackage(const char *package) fPackages->Add(new TObjString(pkg)); } +//______________________________________________________________________________ +TChain *AliAnalysisAlien::GetChainForTestMode(const char *treeName) const +{ +// Make a tree from files having the location specified in fFileForTestMode. +// Inspired from JF's CreateESDChain. + if (fFileForTestMode.IsNull()) { + Error("GetChainForTestMode", "For proof test mode please use SetFileForTestMode() pointing to a file that contains data file locations."); + return NULL; + } + if (gSystem->AccessPathName(fFileForTestMode)) { + Error("GetChainForTestMode", "File not found: %s", fFileForTestMode.Data()); + return NULL; + } + // Open the file + ifstream in; + in.open(fFileForTestMode); + Int_t count = 0; + // Read the input list of files and add them to the chain + TString line; + TChain *chain = new TChain(treeName); + while (in.good()) + { + in >> line; + if (line.IsNull()) continue; + if (count++ == fNtestFiles) break; + TString esdFile(line); + TFile *file = TFile::Open(esdFile); + if (file) { + if (!file->IsZombie()) chain->Add(esdFile); + file->Close(); + } else { + Error("GetChainforTestMode", "Skipping un-openable file: %s", esdFile.Data()); + } + } + in.close(); + if (!chain->GetListOfFiles()->GetEntries()) { + Error("GetChainForTestMode", "No file from %s could be opened", fFileForTestMode.Data()); + delete chain; + return NULL; + } +// chain->ls(); + return chain; +} + //______________________________________________________________________________ const char *AliAnalysisAlien::GetJobStatus(Int_t jobidstart, Int_t lastid, Int_t &nrunning, Int_t &nwaiting, Int_t &nerror, Int_t &ndone) { @@ -1498,6 +1676,45 @@ void AliAnalysisAlien::Print(Option_t *) const { // Print current plugin settings. printf("### AliEn analysis plugin current settings ###\n"); + AliAnalysisManager *mgr = AliAnalysisManager::GetAnalysisManager(); + if (mgr && mgr->IsProofMode()) { + TString proofType = "= PLUGIN IN PROOF MODE ON CLUSTER:_________________"; + if (TestBit(AliAnalysisGrid::kTest)) + proofType = "= PLUGIN IN PROOF LITE MODE ON CLUSTER:____________"; + printf("%s %s\n", proofType.Data(), fProofCluster.Data()); + if (!fProofDataSet.IsNull()) + printf("= Requested data set:___________________________ %s\n", fProofDataSet.Data()); + if (fProofReset==1) + printf("= Soft reset signal will be send to master______ CHANGE BEHAVIOR AFTER COMPLETION\n"); + if (fProofReset>1) + printf("= Hard reset signal will be send to master______ CHANGE BEHAVIOR AFTER COMPLETION\n"); + if (!fRootVersionForProof.IsNull()) + printf("= ROOT version requested________________________ %s\n", fRootVersionForProof.Data()); + else + printf("= ROOT version requested________________________ default\n"); + printf("= AliRoot version requested_____________________ %s\n", fAliROOTVersion.Data()); + if (!fAliRootMode.IsNull()) + printf("= Requested AliRoot mode________________________ %s\n", fAliRootMode.Data()); + if (fNproofWorkers) + printf("= Number of PROOF workers limited to____________ %d\n", fNproofWorkers); + if (fNproofWorkersPerSlave) + printf("= Maximum number of workers per slave___________ %d\n", fNproofWorkersPerSlave); + if (TestSpecialBit(kClearPackages)) + printf("= ClearPackages requested...\n"); + if (fIncludePath.Data()) + printf("= Include path for runtime task compilation: ___ %s\n", fIncludePath.Data()); + printf("= Additional libs to be loaded or souces to be compiled runtime: <%s>\n",fAdditionalLibs.Data()); + if (fPackages && fPackages->GetEntries()) { + TIter next(fPackages); + TObject *obj; + TString list; + while ((obj=next())) list += obj->GetName(); + printf("= Par files to be used: ________________________ %s\n", list.Data()); + } + if (TestSpecialBit(kProofConnectGrid)) + printf("= Requested PROOF connection to grid\n"); + return; + } printf("= OverwriteMode:________________________________ %d\n", fOverwriteMode); if (fOverwriteMode) { printf("***** NOTE: Overwrite mode will overwrite the input generated datasets and partial results from previous analysis. \ @@ -1532,6 +1749,7 @@ void AliAnalysisAlien::Print(Option_t *) const printf("= List of output files to be registered: _______ %s\n", fOutputFiles.Data()); printf("= List of outputs going to be archived: ________ %s\n", fOutputArchive.Data()); printf("= List of outputs that should not be merged: ___ %s\n", fMergeExcludes.Data()); + printf("= List of outputs produced during Terminate: ___ %s\n", fTerminateFiles.Data()); printf("=====================================================================\n"); printf("= Job price: ___________________________________ %d\n", fPrice); printf("= Time to live (TTL): __________________________ %d\n", fTTL); @@ -1563,7 +1781,7 @@ void AliAnalysisAlien::Print(Option_t *) const printf("= Force job outputs to storage element: ________ %s\n", fCloseSE.Data()); if (fFriendChainName.Length()) printf("= Open friend chain file on worker: ____________ %s\n", fFriendChainName.Data()); - if (fPackages) { + if (fPackages && fPackages->GetEntries()) { TIter next(fPackages); TObject *obj; TString list; @@ -1608,7 +1826,7 @@ void AliAnalysisAlien::SetDefaults() fDataPattern = "*AliESDs.root"; // Can be like: *AliESDs.root, */pass1/*AliESDs.root, ... fFriendChainName = ""; fGridOutputDir = "output"; - fOutputArchive = "log_archive.zip:std*@disk=1 root_archive.zip:*.root@disk=2"; + fOutputArchive = "log_archive.zip:std*,*.stat@disk=1 root_archive.zip:*.root@disk=2"; fOutputFiles = ""; // Like "AliAODs.root histos.root" fInputFormat = "xml-single"; fJDLName = "analysis.jdl"; @@ -1622,117 +1840,69 @@ void AliAnalysisAlien::SetDefaults() } //______________________________________________________________________________ -Bool_t AliAnalysisAlien::CheckMergedFiles(const char *filename, const char *aliendir, Int_t nperchunk, Bool_t submit, const char *jdl) +Bool_t AliAnalysisAlien::CheckMergedFiles(const char *filename, const char *aliendir, Int_t nperchunk, const char *jdl) { -// Static method that checks the status of merging. This can submit merging jobs that did not produced the expected -// output. If is false (checking) returns true only when the final merged file was found. If submit is true returns -// true if the jobs were successfully submitted. - Int_t countOrig = 0; - Int_t countStage = 0; - Int_t stage = 0; - Int_t i; - Bool_t doneFinal = kFALSE; - TBits chunksDone; - TString saliendir(aliendir); - TString sfilename, stmp; - saliendir.ReplaceAll("//","/"); - saliendir = saliendir.Strip(TString::kTrailing, '/'); - if (!gGrid) { - ::Error("GetNregisteredFiles", "You need to be connected to AliEn."); +// Checks current merge stage, makes xml for the next stage, counts number of files, submits next stage. + // First check if the result is already in the output directory. + if (FileExists(Form("%s/%s",aliendir,filename))) { + printf("Final merged results found. Not merging again.\n"); return kFALSE; } - sfilename = filename; - sfilename.ReplaceAll(".root", "*.root"); - printf("Checking directory <%s> for merged files <%s> ...\n", aliendir, sfilename.Data()); - TString command = Form("find %s/ *%s", saliendir.Data(), sfilename.Data()); - TGridResult *res = gGrid->Command(command); - if (!res) { - ::Error("GetNregisteredFiles","Error: No result for the find command\n"); + // Now check the last stage done. + Int_t stage = 0; + while (1) { + if (!FileExists(Form("%s/Stage_%d.xml",aliendir, stage+1))) break; + stage++; + } + // Next stage of merging + stage++; + TString pattern = "*root_archive.zip"; + if (stage>1) pattern = Form("Stage_%d/*root_archive.zip", stage-1); + TGridResult *res = gGrid->Command(Form("find -x Stage_%d %s %s", stage, aliendir, pattern.Data())); + if (res) delete res; + // Write standard output to file + gROOT->ProcessLine(Form("gGrid->Stdout(); > %s", Form("Stage_%d.xml",stage))); + // Count the number of files inside + ifstream ifile; + ifile.open(Form("Stage_%d.xml",stage)); + if (!ifile.good()) { + ::Error("CheckMergedFiles", "Could not redirect result of the find command to file %s", Form("Stage_%d.xml",stage)); return kFALSE; - } - TIter nextmap(res); - TMap *map = 0; - while ((map=(TMap*)nextmap())) { - TString turl = map->GetValue("turl")->GetName(); - if (!turl.Length()) { - // Nothing found - delete res; - return kFALSE; - } - turl.ReplaceAll("alien://", ""); - turl.ReplaceAll(saliendir, ""); - sfilename = gSystem->BaseName(turl); - turl = turl.Strip(TString::kLeading, '/'); - // Now check to what the file corresponds to: - // original output - aliendir/%03d/filename - // merged file (which stage) - aliendir/filename-Stage%02d_%04d - // final merged file - aliendir/filename - if (sfilename == turl) { - if (sfilename == filename) { - doneFinal = kTRUE; - } else { - // check stage - Int_t index = sfilename.Index("Stage"); - if (index<0) continue; - stmp = sfilename(index+5,2); - Int_t istage = atoi(stmp); - stmp = sfilename(index+8,4); - Int_t ijob = atoi(stmp); - if (istagestage) { - countStage = 0; - chunksDone.ResetAllBits(); - stage = istage; - } - countStage++; - chunksDone.SetBitNumber(ijob); - } - } else { - countOrig++; - } - if (doneFinal) { - delete res; - printf("=> Removing files from previous stages...\n"); - gGrid->Rm(Form("%s/*Stage*.root", aliendir)); - return kTRUE; - } + } + TString line; + Int_t nfiles = 0; + while (!ifile.eof()) { + ifile >> line; + if (line.Contains("/event")) nfiles++; } - delete res; - // Compute number of jobs that were submitted for the current stage - Int_t ntotstage = countOrig; - for (i=1; i<=stage; i++) { - if (ntotstage%nperchunk) ntotstage = (ntotstage/nperchunk)+1; - else ntotstage = (ntotstage/nperchunk); + ifile.close(); + if (!nfiles) { + ::Error("CheckMergedFiles", "Cannot start Stage_%d merging since Stage_%d did not produced yet output", stage, stage-1); + return kFALSE; + } else { + printf("=== Stage_%d produced %d files\n", stage-1, nfiles); } - // Now compare with the number of set bits in the chunksDone array - Int_t nmissing = (stage>0)?(ntotstage - countStage):0; - // Print the info - printf("*** Found %d original files\n", countOrig); - if (stage==0) printf("*** No merging completed so far.\n"); - else printf("*** Found %d out of %d files merged for stage %d\n", countStage, ntotstage, stage); - if (nmissing) printf("*** Number of merged files missing for this stage: %d -> check merging job completion\n", nmissing); - if (!submit) return doneFinal; - // Sumbit merging jobs for all missing chunks for the current stage. - TString query = Form("submit %s %s", jdl, aliendir); - Int_t ichunk = -1; - if (nmissing) { - for (i=0; i Copying collection %s in the output directory %s\n", Form("Stage_%d.xml",stage), aliendir); + TFile::Cp(Form("Stage_%d.xml",stage), Form("alien://%s/Stage_%d.xml",aliendir,stage)); + // Check if this is the last stage to be done. + Bool_t laststage = (nfiles=fMaxMergeStages) laststage = kTRUE; + if (laststage) { + printf("### Submiting final merging stage %d\n", stage); + TString finalJDL = jdl; + finalJDL.ReplaceAll(".jdl", "_final.jdl"); + TString query = Form("submit %s %s %d 10000 Stage_%d.xml", finalJDL.Data(), aliendir, stage, stage); + Int_t jobId = SubmitSingleJob(query); + if (!jobId) return kFALSE; + } else { + printf("### Submiting merging stage %d\n", stage); + TString query = Form("submit %s %s %d %d wn.xml", jdl, aliendir, stage, nperchunk); + Int_t jobId = SubmitSingleJob(query); + if (!jobId) return kFALSE; } - // Submit next stage of merging - if (stage==0) countStage = countOrig; - Int_t nchunks = (countStage/nperchunk); - if (countStage%nperchunk) nchunks += 1; - for (i=0; i -// stage=n : does a find command for files named Stage_*. If their number is bigger than -// nmaxmerge, merge just the chunk ichunk, otherwise write the merged output to the file -// named . +// Merge given output files from basedir. Basedir can be an alien output directory +// but also an xml file with root_archive.zip locations. The file merger will merge nmaxmerge +// files in a group (ignored for xml input). Merging can be done in stages: +// stage=0 : will merge all existing files in a single stage, supporting resume if run locally +// stage=1 : works with an xml of all root_archive.zip in the output directory +// stage>1 : works with an xml of all root_archive.zip in the Stage_ directory TString outputFile = output; TString command; TString outputChunk; TString previousChunk = ""; + TObjArray *listoffiles = new TObjArray(); +// listoffiles->SetOwner(); Int_t countChunk = 0; Int_t countZero = nmaxmerge; Bool_t merged = kTRUE; Int_t index = outputFile.Index("@"); if (index > 0) outputFile.Remove(index); TString inputFile = outputFile; - if (stage>1) inputFile.ReplaceAll(".root", Form("-Stage%02d_*.root", stage-1)); - command = Form("find %s/ *%s", basedir, inputFile.Data()); - printf("command: %s\n", command.Data()); - TGridResult *res = gGrid->Command(command); - if (!res) { + TString sbasedir = basedir; + if (sbasedir.Contains(".xml")) { + // Merge files pointed by the xml - ignore nmaxmerge and set ichunk to 0 + nmaxmerge = 9999999; + TGridCollection *coll = (TGridCollection*)gROOT->ProcessLine(Form("TAlienCollection::Open(\"%s\");", basedir)); + if (!coll) { + ::Error("MergeOutput", "Input XML collection empty."); + return kFALSE; + } + // Iterate grid collection + while (coll->Next()) { + TString fname = gSystem->DirName(coll->GetTURL()); + fname += "/"; + fname += inputFile; + listoffiles->Add(new TNamed(fname.Data(),"")); + } + } else { + command = Form("find %s/ *%s", basedir, inputFile.Data()); + printf("command: %s\n", command.Data()); + TGridResult *res = gGrid->Command(command); + if (!res) { + ::Error("MergeOutput","No result for the find command\n"); + delete listoffiles; + return kFALSE; + } + TIter nextmap(res); + TMap *map = 0; + while ((map=(TMap*)nextmap())) { + TObjString *objs = dynamic_cast(map->GetValue("turl")); + if (!objs || !objs->GetString().Length()) { + // Nothing found - skip this output + delete res; + delete listoffiles; + return kFALSE; + } + listoffiles->Add(new TNamed(objs->GetName(),"")); + } + delete res; + } + if (!listoffiles->GetEntries()) { ::Error("MergeOutput","No result for the find command\n"); + delete listoffiles; return kFALSE; } TFileMerger *fm = 0; - TIter nextmap(res); - TMap *map = 0; + TIter next0(listoffiles); + TObjArray *listoffilestmp = new TObjArray(); + listoffilestmp->SetOwner(); + TObject *nextfile; + TString snextfile; + // Keep only the files at upper level + Int_t countChar = 0; + while ((nextfile=next0())) { + snextfile = nextfile->GetName(); + Int_t crtCount = snextfile.CountChar('/'); + if (nextfile == listoffiles->First()) countChar = crtCount; + if (crtCount < countChar) countChar = crtCount; + } + next0.Reset(); + while ((nextfile=next0())) { + snextfile = nextfile->GetName(); + Int_t crtCount = snextfile.CountChar('/'); + if (crtCount > countChar) { + delete nextfile; + continue; + } + listoffilestmp->Add(nextfile); + } + delete listoffiles; + listoffiles = listoffilestmp; // Now contains 'good' files + listoffiles->Print(); + TIter next(listoffiles); // Check if there is a merge operation to resume. Works only for stage 0 or 1. outputChunk = outputFile; outputChunk.ReplaceAll(".root", "_*.root"); @@ -1799,11 +2029,14 @@ Bool_t AliAnalysisAlien::MergeOutput(const char *output, const char *basedir, In if (!gSystem->Exec(Form("ls %s 2>/dev/null", outputChunk.Data()))) { while (1) { // Skip as many input files as in a chunk - for (Int_t counter=0; counter, nentries=%d", outputFile.Data(), res->GetSize()); - delete res; - return kFALSE; + for (Int_t counter=0; counterGetName(); } outputChunk = outputFile; outputChunk.ReplaceAll(".root", Form("_%04d.root", countChunk)); @@ -1817,8 +2050,9 @@ Bool_t AliAnalysisAlien::MergeOutput(const char *output, const char *basedir, In } countZero = nmaxmerge; - while ((map=(TMap*)nextmap())) { - // Loop 'find' results and get next LFN + while ((nextfile=next())) { + snextfile = nextfile->GetName(); + // Loop 'find' results and get next LFN if (countZero == nmaxmerge) { // First file in chunk - create file merger and add previous chunk if any. fm = new TFileMerger(kFALSE); @@ -1828,92 +2062,57 @@ Bool_t AliAnalysisAlien::MergeOutput(const char *output, const char *basedir, In outputChunk.ReplaceAll(".root", Form("_%04d.root", countChunk)); } // If last file found, put merged results in the output file - if (map == res->Last()) outputChunk = outputFile; - TObjString *objs = dynamic_cast(map->GetValue("turl")); - if (!objs || !objs->GetString().Length()) { - // Nothing found - skip this output - delete res; - delete fm; - return kFALSE; - } + if (nextfile == listoffiles->Last()) outputChunk = outputFile; // Add file to be merged and decrement chunk counter. - fm->AddFile(objs->GetString()); + fm->AddFile(snextfile); countZero--; - if (countZero==0 || map == res->Last()) { + if (countZero==0 || nextfile == listoffiles->Last()) { if (!fm->GetMergeList() || !fm->GetMergeList()->GetSize()) { // Nothing found - skip this output ::Warning("MergeOutput", "No <%s> files found.", inputFile.Data()); - delete res; - delete fm; - return kFALSE; + merged = kFALSE; + break; } fm->OutputFile(outputChunk); // Merge the outputs, then go to next chunk if (!fm->Merge()) { ::Error("MergeOutput", "Could not merge all <%s> files", outputFile.Data()); - delete res; - delete fm; - return kFALSE; + merged = kFALSE; + break; } else { ::Info("MergeOutputs", "\n##### Merged %d output files to <%s>", fm->GetMergeList()->GetSize(), outputChunk.Data()); gSystem->Unlink(previousChunk); } - if (map == res->Last()) { - delete res; - delete fm; - break; - } + if (nextfile == listoffiles->Last()) break; countChunk++; countZero = nmaxmerge; previousChunk = outputChunk; } } + delete listoffiles; + delete fm; return merged; } // Merging stage different than 0. // Move to the begining of the requested chunk. - outputChunk = outputFile; - if (nmaxmerge < res->GetSize()) { - if (ichunk*nmaxmerge >= res->GetSize()) { - ::Error("MergeOutput", "Cannot merge merge chunk %d grouping %d files from %d total.", ichunk, nmaxmerge, res->GetSize()); - delete res; - return kFALSE; - } - for (Int_t counter=0; counterSetFastMethod(kTRUE); - while ((map=(TMap*)nextmap())) { - // Loop 'find' results and get next LFN - TObjString *objs = dynamic_cast(map->GetValue("turl")); - if (!objs || !objs->GetString().Length()) { - // Nothing found - skip this output - delete res; - delete fm; - return kFALSE; - } - // Add file to be merged and decrement chunk counter. - fm->AddFile(objs->GetString()); - countZero--; - if (countZero==0) break; - } - delete res; + while ((nextfile=next())) fm->AddFile(nextfile->GetName()); + delete listoffiles; if (!fm->GetMergeList() || !fm->GetMergeList()->GetSize()) { // Nothing found - skip this output ::Warning("MergeOutput", "No <%s> files found.", inputFile.Data()); delete fm; return kFALSE; } - fm->OutputFile(outputChunk); + fm->OutputFile(outputFile); // Merge the outputs if (!fm->Merge()) { ::Error("MergeOutput", "Could not merge all <%s> files", outputFile.Data()); delete fm; return kFALSE; } else { - ::Info("MergeOutput", "\n##### Merged %d output files to <%s>", fm->GetMergeList()->GetSize(), outputChunk.Data()); + ::Info("MergeOutput", "\n##### Merged %d output files to <%s>", fm->GetMergeList()->GetSize(), outputFile.Data()); } delete fm; return kTRUE; @@ -1960,10 +2159,10 @@ Bool_t AliAnalysisAlien::MergeOutputs() if (fFastReadOption) { Warning("MergeOutputs", "You requested FastRead option. Using xrootd flags to reduce timeouts. This may skip some files that could be accessed ! \ \n+++ NOTE: To disable this option, use: plugin->SetFastReadOption(kFALSE)"); - gEnv->SetValue("XNet.ConnectTimeout",10); - gEnv->SetValue("XNet.RequestTimeout",10); + gEnv->SetValue("XNet.ConnectTimeout",50); + gEnv->SetValue("XNet.RequestTimeout",50); gEnv->SetValue("XNet.MaxRedirectCount",2); - gEnv->SetValue("XNet.ReconnectTimeout",10); + gEnv->SetValue("XNet.ReconnectTimeout",50); gEnv->SetValue("XNet.FirstConnectMaxCnt",1); } // Make sure we change the temporary directory @@ -2049,7 +2248,7 @@ void AliAnalysisAlien::SetOutputFiles(const char *list) fOutputFiles += sout; } delete arr; -} +} //______________________________________________________________________________ void AliAnalysisAlien::SetOutputArchive(const char *list) @@ -2075,38 +2274,248 @@ void AliAnalysisAlien::SetPreferedSE(const char */*se*/) Bool_t AliAnalysisAlien::StartAnalysis(Long64_t /*nentries*/, Long64_t /*firstEntry*/) { // Start remote grid analysis. - - // Check if output files have to be taken from the analysis manager - if (TestBit(AliAnalysisGrid::kDefaultOutputs)) { - AliAnalysisManager *mgr = AliAnalysisManager::GetAnalysisManager(); - if (!mgr || !mgr->IsInitialized()) { - Error("StartAnalysis", "You need an initialized analysis manager for this"); + AliAnalysisManager *mgr = AliAnalysisManager::GetAnalysisManager(); + Bool_t testMode = TestBit(AliAnalysisGrid::kTest); + if (!mgr || !mgr->IsInitialized()) { + Error("StartAnalysis", "You need an initialized analysis manager for this"); + return kFALSE; + } + // Are we in PROOF mode ? + if (mgr->IsProofMode()) { + Info("StartAnalysis", "##### Starting PROOF analysis on cluster <%s> via the plugin #####", fProofCluster.Data()); + if (fProofCluster.IsNull()) { + Error("StartAnalysis", "You need to specify the proof cluster name via SetProofCluster"); + return kFALSE; + } + if (fProofDataSet.IsNull() && !testMode) { + Error("StartAnalysis", "You need to specify a dataset using SetProofDataSet()"); return kFALSE; + } + // Set the needed environment + gEnv->SetValue("XSec.GSI.DelegProxy","2"); + // Do we need to reset PROOF ? The success of the Reset operation cannot be checked + if (fProofReset && !testMode) { + if (fProofReset==1) { + Info("StartAnalysis", "Sending soft reset signal to proof cluster %s", fProofCluster.Data()); + gROOT->ProcessLine(Form("TProof::Reset(\"%s\", kFALSE);", fProofCluster.Data())); + } else { + Info("StartAnalysis", "Sending hard reset signal to proof cluster %s", fProofCluster.Data()); + gROOT->ProcessLine(Form("TProof::Reset(\"%s\", kTRUE);", fProofCluster.Data())); + } + Info("StartAnalysis", "Stopping the analysis. Please use SetProofReset(0) to resume."); + return kFALSE; + } + // Do we need to change the ROOT version ? The success of this cannot be checked. + if (!fRootVersionForProof.IsNull() && !testMode) { + gROOT->ProcessLine(Form("TProof::Mgr(\"%s\")->SetROOTVersion(\"%s\");", + fProofCluster.Data(), fRootVersionForProof.Data())); + } + // Connect to PROOF and check the status + Long_t proof = 0; + TString sworkers; + if (fNproofWorkersPerSlave) sworkers = Form("workers=%dx", fNproofWorkersPerSlave); + else if (fNproofWorkers) sworkers = Form("workers=%d", fNproofWorkers); + if (!testMode) { + if (!sworkers.IsNull()) + proof = gROOT->ProcessLine(Form("TProof::Open(\"%s\", \"%s\");", fProofCluster.Data(), sworkers.Data())); + else + proof = gROOT->ProcessLine(Form("TProof::Open(\"%s\");", fProofCluster.Data())); + } else { + proof = gROOT->ProcessLine("TProof::Open(\"\");"); + if (!proof) { + Error("StartAnalysis", "Could not start PROOF in test mode"); + return kFALSE; + } + } + if (!proof) { + Error("StartAnalysis", "Could not connect to PROOF cluster <%s>", fProofCluster.Data()); + return kFALSE; + } + if (fNproofWorkersPerSlave*fNproofWorkers > 0) + gROOT->ProcessLine(Form("gProof->SetParallel(%d);", fNproofWorkers)); + // Is dataset existing ? + if (!testMode) { + TString dataset = fProofDataSet; + Int_t index = dataset.Index("#"); + if (index>=0) dataset.Remove(index); +// if (!gROOT->ProcessLine(Form("gProof->ExistsDataSet(\"%s\");",fProofDataSet.Data()))) { +// Error("StartAnalysis", "Dataset %s not existing", fProofDataSet.Data()); +// return kFALSE; +// } +// Info("StartAnalysis", "Dataset %s found", dataset.Data()); + } + // Is ClearPackages() needed ? + if (TestSpecialBit(kClearPackages)) { + Info("StartAnalysis", "ClearPackages signal sent to PROOF. Use SetClearPackages(kFALSE) to reset this."); + gROOT->ProcessLine("gProof->ClearPackages();"); + } + // Is a given aliroot mode requested ? + TList optionsList; + TString parLibs; + if (!fAliRootMode.IsNull()) { + TString alirootMode = fAliRootMode; + if (alirootMode == "default") alirootMode = ""; + Info("StartAnalysis", "You are requesting AliRoot mode: %s", fAliRootMode.Data()); + optionsList.SetOwner(); + optionsList.Add(new TNamed("ALIROOT_MODE", alirootMode.Data())); + // Check the additional libs to be loaded + TString extraLibs; + Bool_t parMode = kFALSE; + if (!alirootMode.IsNull()) extraLibs = "ANALYSIS:ANALYSISalice"; + // Parse the extra libs for .so + if (fAdditionalLibs.Length()) { + TObjArray *list = fAdditionalLibs.Tokenize(" "); + TIter next(list); + TObjString *str; + while((str=(TObjString*)next())) { + if (str->GetString().Contains(".so")) { + if (parMode) { + Warning("StartAnalysis", "Plugin does not support loading libs after par files in PROOF mode. Library %s and following will not load on workers", str->GetName()); + break; + } + TString stmp = str->GetName(); + if (stmp.BeginsWith("lib")) stmp.Remove(0,3); + stmp.ReplaceAll(".so",""); + if (!extraLibs.IsNull()) extraLibs += ":"; + extraLibs += stmp; + continue; + } + if (str->GetString().Contains(".par")) { + // The first par file found in the list will not allow any further .so + parMode = kTRUE; + if (!parLibs.IsNull()) parLibs += ":"; + parLibs += str->GetName(); + continue; + } + } + if (list) delete list; + } + if (!extraLibs.IsNull()) optionsList.Add(new TNamed("ALIROOT_EXTRA_LIBS",extraLibs.Data())); + // Check extra includes + if (!fIncludePath.IsNull()) { + TString includePath = fIncludePath; + includePath.ReplaceAll(" ",":"); + includePath.Strip(TString::kTrailing, ':'); + Info("StartAnalysis", "Adding extra includes: %s",includePath.Data()); + optionsList.Add(new TNamed("ALIROOT_EXTRA_INCLUDES",includePath.Data())); + } + // Check if connection to grid is requested + if (TestSpecialBit(kProofConnectGrid)) + optionsList.Add(new TNamed("ALIROOT_ENABLE_ALIEN", "1")); + // Enable AliRoot par + if (testMode) { + // Enable proof lite package + TString alirootLite = gSystem->ExpandPathName("$ALICE_ROOT/ANALYSIS/macros/AliRootProofLite.par"); + for (Int_t i=0; iGetName(), obj->GetTitle()); + } + if (!gROOT->ProcessLine(Form("gProof->UploadPackage(\"%s\");",alirootLite.Data())) + && !gROOT->ProcessLine(Form("gProof->EnablePackage(\"%s\", (TList*)0x%lx);",alirootLite.Data(),(ULong_t)&optionsList))) { + Info("StartAnalysis", "AliRootProofLite enabled"); + } else { + Error("StartAnalysis", "There was an error trying to enable package AliRootProofLite.par"); + return kFALSE; + } + } else { + if (gROOT->ProcessLine(Form("gProof->EnablePackage(\"VO_ALICE@AliRoot::%s\", (TList*)0x%lx);", + fAliROOTVersion.Data(), (ULong_t)&optionsList))) { + Error("StartAnalysis", "There was an error trying to enable package VO_ALICE@AliRoot::%s", fAliROOTVersion.Data()); + return kFALSE; + } + } + // Enable first par files from fAdditionalLibs + if (!parLibs.IsNull()) { + TObjArray *list = parLibs.Tokenize(":"); + TIter next(list); + TObjString *package; + while((package=(TObjString*)next())) { + TString spkg = package->GetName(); + spkg.ReplaceAll(".par", ""); + gSystem->Exec(TString::Format("rm -rf %s", spkg.Data())); + if (!gROOT->ProcessLine(Form("gProof->UploadPackage(\"%s\");", package->GetName()))) { + TString enablePackage = (testMode)?Form("gProof->EnablePackage(\"%s\",kFALSE);", package->GetName()):Form("gProof->EnablePackage(\"%s\",kTRUE);", package->GetName()); + if (gROOT->ProcessLine(enablePackage)) { + Error("StartAnalysis", "There was an error trying to enable package %s", package->GetName()); + return kFALSE; + } + } else { + Error("StartAnalysis", "There was an error trying to upload package %s", package->GetName()); + return kFALSE; + } + } + if (list) delete list; + } + } else { + if (fAdditionalLibs.Contains(".so") && !testMode) { + Error("StartAnalysis", "You request additional libs to be loaded but did not enabled any AliRoot mode. Please refer to: \ + \n http://aaf.cern.ch/node/83 and use a parameter for SetAliRootMode()"); + return kFALSE; + } } - fOutputFiles = ""; - TIter next(mgr->GetOutputs()); - AliAnalysisDataContainer *output; - while ((output=(AliAnalysisDataContainer*)next())) { - const char *filename = output->GetFileName(); - if (!(strcmp(filename, "default"))) { - if (!mgr->GetOutputEventHandler()) continue; - filename = mgr->GetOutputEventHandler()->GetOutputFileName(); + // Enable par files if requested + if (fPackages && fPackages->GetEntries()) { + TIter next(fPackages); + TObject *package; + while ((package=next())) { + // Skip packages already enabled + if (parLibs.Contains(package->GetName())) continue; + TString spkg = package->GetName(); + spkg.ReplaceAll(".par", ""); + gSystem->Exec(TString::Format("rm -rf %s", spkg.Data())); + if (gROOT->ProcessLine(Form("gProof->UploadPackage(\"%s\");", package->GetName()))) { + if (gROOT->ProcessLine(Form("gProof->EnablePackage(\"%s\",kTRUE);", package->GetName()))) { + Error("StartAnalysis", "There was an error trying to enable package %s", package->GetName()); + return kFALSE; + } + } else { + Error("StartAnalysis", "There was an error trying to upload package %s", package->GetName()); + return kFALSE; + } } - if (fOutputFiles.Contains(filename)) continue; - if (fOutputFiles.Length()) fOutputFiles += ","; - fOutputFiles += filename; } + // Do we need to load analysis source files ? + // NOTE: don't load on client since this is anyway done by the user to attach his task. + if (fAnalysisSource.Length()) { + TObjArray *list = fAnalysisSource.Tokenize(" "); + TIter next(list); + TObjString *str; + while((str=(TObjString*)next())) { + gROOT->ProcessLine(Form("gProof->Load(\"%s+g\", kTRUE);", str->GetName())); + } + if (list) delete list; + } + if (testMode) { + // Register dataset to proof lite. + if (fFileForTestMode.IsNull()) { + Error("GetChainForTestMode", "For proof test mode please use SetFileForTestMode() pointing to a file that contains data file locations."); + return kFALSE; + } + if (gSystem->AccessPathName(fFileForTestMode)) { + Error("GetChainForTestMode", "File not found: %s", fFileForTestMode.Data()); + return kFALSE; + } + TFileCollection *coll = new TFileCollection(); + coll->AddFromFile(fFileForTestMode); + gROOT->ProcessLine(Form("gProof->RegisterDataSet(\"test_collection\", (TFileCollection*)0x%lx, \"OV\");", (ULong_t)coll)); + gROOT->ProcessLine("gProof->ShowDataSets()"); + } + return kTRUE; + } + + // Check if output files have to be taken from the analysis manager + if (TestBit(AliAnalysisGrid::kDefaultOutputs)) { + // Add output files and AOD files + fOutputFiles = GetListOfFiles("outaod"); // Add extra files registered to the analysis manager - if (mgr->GetExtraFiles().Length()) { - if (fOutputFiles.Length()) fOutputFiles += ","; - TString extra = mgr->GetExtraFiles(); - extra.ReplaceAll(" ", ","); - // Protection in case extra files do not exist (will it work?) + TString extra = GetListOfFiles("ext"); + if (!extra.IsNull()) { extra.ReplaceAll(".root", "*.root"); + if (!fOutputFiles.IsNull()) fOutputFiles += ","; fOutputFiles += extra; } // Compose the output archive. - fOutputArchive = "log_archive.zip:std*@disk=1 "; + fOutputArchive = "log_archive.zip:std*,*.stat@disk=1 "; fOutputArchive += Form("root_archive.zip:%s@disk=%d",fOutputFiles.Data(),fNreplicas); } // if (!fCloseSE.Length()) fCloseSE = gSystem->Getenv("alien_CLOSE_SE"); @@ -2133,7 +2542,7 @@ Bool_t AliAnalysisAlien::StartAnalysis(Long64_t /*nentries*/, Long64_t /*firstEn Error("StartAnalysis", "Cannot start grid analysis without grid connection"); return kFALSE; } - if (IsCheckCopy()) CheckFileCopy(gGrid->GetHomeDirectory()); + if (IsCheckCopy() && gGrid) CheckFileCopy(gGrid->GetHomeDirectory()); if (!CheckInputData()) { Error("StartAnalysis", "There was an error in preprocessing your requested input data"); return kFALSE; @@ -2158,7 +2567,7 @@ Bool_t AliAnalysisAlien::StartAnalysis(Long64_t /*nentries*/, Long64_t /*firstEn } if (!CreateJDL()) return kFALSE; if (TestBit(AliAnalysisGrid::kOffline)) return kFALSE; - if (TestBit(AliAnalysisGrid::kTest)) { + if (testMode) { // Locally testing the analysis Info("StartAnalysis", "\n_______________________________________________________________________ \ \n Running analysis script in a daughter shell as on a worker node \ @@ -2175,9 +2584,7 @@ Bool_t AliAnalysisAlien::StartAnalysis(Long64_t /*nentries*/, Long64_t /*firstEn } delete list; gSystem->Exec(Form("bash %s 2>stderr", fExecutable.Data())); - TString validationScript = fExecutable; - validationScript.ReplaceAll(".sh", "_validation.sh"); - gSystem->Exec(Form("bash %s",validationScript.Data())); + gSystem->Exec(Form("bash %s",fValidationScript.Data())); // gSystem->Exec("cat stdout"); return kFALSE; } @@ -2228,6 +2635,104 @@ Bool_t AliAnalysisAlien::StartAnalysis(Long64_t /*nentries*/, Long64_t /*firstEn return kTRUE; } +//______________________________________________________________________________ +const char *AliAnalysisAlien::GetListOfFiles(const char *type) +{ +// Get a comma-separated list of output files of the requested type. +// Type can be (case unsensitive): +// aod - list of aod files (std, extensions and filters) +// out - list of output files connected to containers (but not aod's or extras) +// ext - list of extra files registered to the manager +// ter - list of files produced in terminate + static TString files; + files = ""; + TString stype = type; + stype.ToLower(); + TString aodfiles, extra; + AliAnalysisManager *mgr = AliAnalysisManager::GetAnalysisManager(); + if (!mgr) { + ::Error("GetListOfFiles", "Cannot call this without analysis manager"); + return files.Data(); + } + if (mgr->GetOutputEventHandler()) { + aodfiles = mgr->GetOutputEventHandler()->GetOutputFileName(); + TString extraaod = mgr->GetOutputEventHandler()->GetExtraOutputs(); + if (!extraaod.IsNull()) { + aodfiles += ","; + aodfiles += extraaod; + } + } + if (stype.Contains("aod")) { + files = aodfiles; + if (stype == "aod") return files.Data(); + } + // Add output files that are not in the list of AOD files + TString outputfiles = ""; + TIter next(mgr->GetOutputs()); + AliAnalysisDataContainer *output; + const char *filename = 0; + while ((output=(AliAnalysisDataContainer*)next())) { + filename = output->GetFileName(); + if (!(strcmp(filename, "default"))) continue; + if (outputfiles.Contains(filename)) continue; + if (aodfiles.Contains(filename)) continue; + if (!outputfiles.IsNull()) outputfiles += ","; + outputfiles += filename; + } + if (stype.Contains("out")) { + if (!files.IsNull()) files += ","; + files += outputfiles; + if (stype == "out") return files.Data(); + } + // Add extra files registered to the analysis manager + TString sextra; + extra = mgr->GetExtraFiles(); + if (!extra.IsNull()) { + extra.Strip(); + extra.ReplaceAll(" ", ","); + TObjArray *fextra = extra.Tokenize(","); + TIter nextx(fextra); + TObject *obj; + while ((obj=nextx())) { + if (aodfiles.Contains(obj->GetName())) continue; + if (outputfiles.Contains(obj->GetName())) continue; + if (sextra.Contains(obj->GetName())) continue; + if (!sextra.IsNull()) sextra += ","; + sextra += obj->GetName(); + } + delete fextra; + if (stype.Contains("ext")) { + if (!files.IsNull()) files += ","; + files += sextra; + } + } + if (stype == "ext") return files.Data(); + TString termfiles; + if (!fTerminateFiles.IsNull()) { + fTerminateFiles.Strip(); + fTerminateFiles.ReplaceAll(" ",","); + TObjArray *fextra = fTerminateFiles.Tokenize(","); + TIter nextx(fextra); + TObject *obj; + while ((obj=nextx())) { + if (aodfiles.Contains(obj->GetName())) continue; + if (outputfiles.Contains(obj->GetName())) continue; + if (termfiles.Contains(obj->GetName())) continue; + if (sextra.Contains(obj->GetName())) continue; + if (!termfiles.IsNull()) termfiles += ","; + termfiles += obj->GetName(); + } + delete fextra; + } + if (stype.Contains("ter")) { + if (!files.IsNull() && !termfiles.IsNull()) { + files += ","; + files += termfiles; + } + } + return files.Data(); +} + //______________________________________________________________________________ Bool_t AliAnalysisAlien::Submit() { @@ -2278,7 +2783,7 @@ Bool_t AliAnalysisAlien::SubmitMerging() if (!fMergeExcludes.Contains(outputFile)) break; } delete list; - Bool_t done = CheckMergedFiles(outputFile, runOutDir, fMaxMergeFiles, kTRUE, mergeJDLName); + Bool_t done = CheckMergedFiles(outputFile, runOutDir, fMaxMergeFiles, mergeJDLName); if (!done) return kFALSE; } if (!ntosubmit) return kTRUE; @@ -2292,7 +2797,10 @@ Bool_t AliAnalysisAlien::SubmitMerging() //______________________________________________________________________________ Bool_t AliAnalysisAlien::SubmitNext() { -// Submit next bunch of master jobs if the queue is free. +// Submit next bunch of master jobs if the queue is free. The first master job is +// submitted right away, while the next will not be unless the previous was split. +// The plugin will not submit new master jobs if there are more that 500 jobs in +// waiting phase. static Bool_t iscalled = kFALSE; static Int_t firstmaster = 0; static Int_t lastmaster = 0; @@ -2303,21 +2811,28 @@ Bool_t AliAnalysisAlien::SubmitNext() Int_t ntosubmit = 0; TGridResult *res; TString jobID = ""; - if (!fNsubmitted) ntosubmit = 1; - else { + Int_t nmasterjobs = fInputFiles->GetEntries(); + if (!fNsubmitted) { + ntosubmit = 1; + if (!IsUseSubmitPolicy()) { + if (nmasterjobs>5) + Info("SubmitNext","### Warning submit policy not used ! Submitting too many jobs at a time may be prohibitted. \ + \n### You can use SetUseSubmitPolicy() to enable if you have problems."); + ntosubmit = nmasterjobs; + } + } else { TString status = GetJobStatus(firstmaster, lastmaster, nrunning, nwaiting, nerror, ndone); printf("=== master %d: %s\n", lastmaster, status.Data()); // If last master not split, just return if (status != "SPLIT") {iscalled = kFALSE; return kTRUE;} // No more than 100 waiting jobs - if (nwaiting>100) {iscalled = kFALSE; return kTRUE;} + if (nwaiting>500) {iscalled = kFALSE; return kTRUE;} npermaster = (nrunning+nwaiting+nerror+ndone)/fNsubmitted; - if (npermaster) ntosubmit = (100-nwaiting)/npermaster; + if (npermaster) ntosubmit = (500-nwaiting)/npermaster; if (!ntosubmit) ntosubmit = 1; printf("=== WAITING(%d) RUNNING(%d) DONE(%d) OTHER(%d) NperMaster=%d => to submit %d jobs\n", nwaiting, nrunning, ndone, nerror, npermaster, ntosubmit); } - Int_t nmasterjobs = fInputFiles->GetEntries(); for (Int_t i=0; i=nmasterjobs) {iscalled = kFALSE; return kTRUE;} @@ -2395,12 +2910,12 @@ void AliAnalysisAlien::WriteAnalysisFile() Info("WriteAnalysisFile", "\n##### Analysis manager: %s wrote to file <%s>\n", mgr->GetName(),analysisFile.Data()); } Bool_t copy = kTRUE; - if (TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; + if (fProductionMode || TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; if (copy) { CdWork(); TString workdir = gGrid->GetHomeDirectory(); workdir += fGridWorkingDir; - Info("CreateJDL", "\n##### Copying file <%s> containing your initialized analysis manager to your alien workspace", analysisFile.Data()); + Info("WriteAnalysisFile", "\n##### Copying file <%s> containing your initialized analysis manager to your alien workspace", analysisFile.Data()); if (FileExists(analysisFile)) gGrid->Rm(analysisFile); TFile::Cp(Form("file:%s",analysisFile.Data()), Form("alien://%s/%s", workdir.Data(),analysisFile.Data())); } @@ -2445,12 +2960,20 @@ void AliAnalysisAlien::WriteAnalysisMacro() out << "// Automatically generated analysis steering macro executed in grid subjobs" << endl << endl; out << " TStopwatch timer;" << endl; out << " timer.Start();" << endl << endl; - out << "// load base root libraries" << endl; - out << " gSystem->Load(\"libTree\");" << endl; - out << " gSystem->Load(\"libGeom\");" << endl; - out << " gSystem->Load(\"libVMC\");" << endl; - out << " gSystem->Load(\"libPhysics\");" << endl << endl; - out << " gSystem->Load(\"libMinuit\");" << endl << endl; + // Change temp directory to current one + out << "// Set temporary merging directory to current one" << endl; + out << " gSystem->Setenv(\"TMPDIR\", gSystem->pwd());" << endl << endl; + // Reset existing include path + out << "// Reset existing include path and add current directory first in the search" << endl; + out << " gSystem->SetIncludePath(\"-I.\");" << endl; + if (!fExecutableCommand.Contains("aliroot")) { + out << "// load base root libraries" << endl; + out << " gSystem->Load(\"libTree\");" << endl; + out << " gSystem->Load(\"libGeom\");" << endl; + out << " gSystem->Load(\"libVMC\");" << endl; + out << " gSystem->Load(\"libPhysics\");" << endl << endl; + out << " gSystem->Load(\"libMinuit\");" << endl << endl; + } if (fAdditionalRootLibs.Length()) { // in principle libtree /lib geom libvmc etc. can go into this list, too out << "// Add aditional libraries" << endl; @@ -2463,15 +2986,14 @@ void AliAnalysisAlien::WriteAnalysisMacro() } if (list) delete list; } - out << "// include path" << endl; - if (fIncludePath.Length()) out << " gSystem->AddIncludePath(\"" << fIncludePath.Data() << "\");" << endl; - out << " gSystem->AddIncludePath(\"-I$ALICE_ROOT/include\");" << endl << endl; out << "// Load analysis framework libraries" << endl; TString setupPar = "AliAnalysisAlien::SetupPar"; if (!fPackages) { - out << " gSystem->Load(\"libSTEERBase\");" << endl; - out << " gSystem->Load(\"libESD\");" << endl; - out << " gSystem->Load(\"libAOD\");" << endl; + if (!fExecutableCommand.Contains("aliroot")) { + out << " gSystem->Load(\"libSTEERBase\");" << endl; + out << " gSystem->Load(\"libESD\");" << endl; + out << " gSystem->Load(\"libAOD\");" << endl; + } out << " gSystem->Load(\"libANALYSIS\");" << endl; out << " gSystem->Load(\"libANALYSISalice\");" << endl; out << " gSystem->Load(\"libCORRFW\");" << endl << endl; @@ -2526,6 +3048,21 @@ void AliAnalysisAlien::WriteAnalysisMacro() out << " if (!" << setupPar << "(\"" << obj->GetName() << "\")) return;" << endl; } } + out << "// include path" << endl; + // Get the include path from the interpreter and remove entries pointing to AliRoot + out << " TString intPath = gInterpreter->GetIncludePath();" << endl; + out << " TObjArray *listpaths = intPath.Tokenize(\" \");" << endl; + out << " TIter nextpath(listpaths);" << endl; + out << " TObjString *pname;" << endl; + out << " while ((pname=(TObjString*)nextpath())) {" << endl; + out << " TString current = pname->GetName();" << endl; + out << " if (current.Contains(\"AliRoot\") || current.Contains(\"ALICE_ROOT\")) continue;" << endl; + out << " gSystem->AddIncludePath(current);" << endl; + out << " }" << endl; + out << " if (listpaths) delete listpaths;" << endl; + if (fIncludePath.Length()) out << " gSystem->AddIncludePath(\"" << fIncludePath.Data() << "\");" << endl; + out << " gROOT->ProcessLine(\".include $ALICE_ROOT/include\");" << endl; + out << " printf(\"Include path: %s\\n\", gSystem->GetIncludePath());" << endl << endl; if (fAdditionalLibs.Length()) { out << "// Add aditional AliRoot libraries" << endl; TObjArray *list = fAdditionalLibs.Tokenize(" "); @@ -2551,27 +3088,21 @@ void AliAnalysisAlien::WriteAnalysisMacro() if (list) delete list; } out << endl; +// out << " printf(\"Currently load libraries:\\n\");" << endl; +// out << " printf(\"%s\\n\", gSystem->GetLibraries());" << endl; if (fFastReadOption) { Warning("WriteAnalysisMacro", "!!! You requested FastRead option. Using xrootd flags to reduce timeouts in the grid jobs. This may skip some files that could be accessed !!! \ \n+++ NOTE: To disable this option, use: plugin->SetFastReadOption(kFALSE)"); out << "// fast xrootd reading enabled" << endl; out << " printf(\"!!! You requested FastRead option. Using xrootd flags to reduce timeouts. Note that this may skip some files that could be accessed !!!\");" << endl; - out << " gEnv->SetValue(\"XNet.ConnectTimeout\",10);" << endl; - out << " gEnv->SetValue(\"XNet.RequestTimeout\",10);" << endl; + out << " gEnv->SetValue(\"XNet.ConnectTimeout\",50);" << endl; + out << " gEnv->SetValue(\"XNet.RequestTimeout\",50);" << endl; out << " gEnv->SetValue(\"XNet.MaxRedirectCount\",2);" << endl; - out << " gEnv->SetValue(\"XNet.ReconnectTimeout\",10);" << endl; + out << " gEnv->SetValue(\"XNet.ReconnectTimeout\",50);" << endl; out << " gEnv->SetValue(\"XNet.FirstConnectMaxCnt\",1);" << endl << endl; } - // Change temp directory to current one - out << "// Set temporary merging directory to current one" << endl; - out << " gSystem->Setenv(\"TMPDIR\", gSystem->pwd());" << endl << endl; out << "// connect to AliEn and make the chain" << endl; out << " if (!TGrid::Connect(\"alien://\")) return;" << endl; - if (IsUsingTags()) { - out << " TChain *chain = CreateChainFromTags(\"wn.xml\", anatype);" << endl << endl; - } else { - out << " TChain *chain = CreateChain(\"wn.xml\", anatype);" << endl << endl; - } out << "// read the analysis manager from file" << endl; TString analysisFile = fExecutable; analysisFile.ReplaceAll(".sh", ".root"); @@ -2593,9 +3124,17 @@ void AliAnalysisAlien::WriteAnalysisMacro() if (AliAnalysisManager::GetAnalysisManager()->GetDebugLevel()>3) { out << " gEnv->SetValue(\"XNet.Debug\", \"1\");" << endl; } else { - out << " AliLog::SetGlobalLogLevel(AliLog::kError);" << endl; + if (TestBit(AliAnalysisGrid::kTest)) + out << " AliLog::SetGlobalLogLevel(AliLog::kWarning);" << endl; + else + out << " AliLog::SetGlobalLogLevel(AliLog::kError);" << endl; } } + if (IsUsingTags()) { + out << " TChain *chain = CreateChainFromTags(\"wn.xml\", anatype);" << endl << endl; + } else { + out << " TChain *chain = CreateChain(\"wn.xml\", anatype);" << endl << endl; + } out << " mgr->StartAnalysis(\"localfile\", chain);" << endl; out << " timer.Stop();" << endl; out << " timer.Print();" << endl; @@ -2638,7 +3177,7 @@ void AliAnalysisAlien::WriteAnalysisMacro() msg += " AliLHCTagCuts *lhcCuts,\n"; msg += " AliDetectorTagCuts *detCuts,\n"; msg += " AliEventTagCuts *evCuts)"; - Info("WriteAnalysisMacro", msg.Data()); + Info("WriteAnalysisMacro", "%s", msg.Data()); } } if (!IsUsingTags() || fFriendChainName!="") { @@ -2646,6 +3185,8 @@ void AliAnalysisAlien::WriteAnalysisMacro() out << "TChain* CreateChain(const char *xmlfile, const char *type=\"ESD\")" << endl; out << "{" << endl; out << "// Create a chain using url's from xml file" << endl; + out << " TString filename;" << endl; + out << " Int_t run = 0;" << endl; out << " TString treename = type;" << endl; out << " treename.ToLower();" << endl; out << " treename += \"Tree\";" << endl; @@ -2657,13 +3198,23 @@ void AliAnalysisAlien::WriteAnalysisMacro() out << " ::Error(\"CreateChain\", \"Cannot create an AliEn collection from %s\", xmlfile);" << endl; out << " return NULL;" << endl; out << " }" << endl; + out << " AliAnalysisManager *mgr = AliAnalysisManager::GetAnalysisManager();" << endl; out << " TChain *chain = new TChain(treename);" << endl; if(fFriendChainName!="") { out << " TChain *chainFriend = new TChain(treename);" << endl; } out << " coll->Reset();" << endl; out << " while (coll->Next()) {" << endl; - out << " chain->Add(coll->GetTURL(\"\"));" << endl; + out << " filename = coll->GetTURL("");" << endl; + out << " if (mgr) {" << endl; + out << " Int_t nrun = AliAnalysisManager::GetRunFromAlienPath(filename);" << endl; + out << " if (nrun && nrun != run) {" << endl; + out << " printf(\"### Run number detected from chain: %d\\n\", nrun);" << endl; + out << " mgr->SetRunFromPath(nrun);" << endl; + out << " run = nrun;" << endl; + out << " }" << endl; + out << " }" << endl; + out << " chain->Add(filename);" << endl; if(fFriendChainName!="") { out << " TString fileFriend=coll->GetTURL(\"\");" << endl; out << " fileFriend.ReplaceAll(\"AliAOD.root\",\""<Exec(Form(\"tar xvzf %s.par\", pkgdir.Data()));" << endl; + out << " gSystem->Exec(TString::Format(\"tar xvzf %s.par\", pkgdir.Data()));" << endl; out << " TString cdir = gSystem->WorkingDirectory();" << endl; out << " gSystem->ChangeDirectory(pkgdir);" << endl; out << " // Check for BUILD.sh and execute" << endl; @@ -2724,7 +3275,7 @@ void AliAnalysisAlien::WriteAnalysisMacro() Info("WriteAnalysisMacro", "\n##### Analysis macro to run on worker nodes <%s> written",fAnalysisMacro.Data()); } Bool_t copy = kTRUE; - if (TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; + if (fProductionMode || TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; if (copy) { CdWork(); TString workdir = gGrid->GetHomeDirectory(); @@ -2768,11 +3319,14 @@ void AliAnalysisAlien::WriteMergingMacro() TString func = mergingMacro; TString comment; func.ReplaceAll(".C", ""); - out << "void " << func.Data() << "(const char *dir, Int_t stage=0, Int_t ichunk=0)" << endl; + out << "void " << func.Data() << "(const char *dir, Int_t stage=0, Int_t laststage=0)" << endl; out << "{" << endl; out << "// Automatically generated merging macro executed in grid subjobs" << endl << endl; out << " TStopwatch timer;" << endl; out << " timer.Start();" << endl << endl; + // Reset existing include path + out << "// Reset existing include path and add current directory first in the search" << endl; + out << " gSystem->SetIncludePath(\"-I.\");" << endl; if (!fExecutableCommand.Contains("aliroot")) { out << "// load base root libraries" << endl; out << " gSystem->Load(\"libTree\");" << endl; @@ -2793,9 +3347,6 @@ void AliAnalysisAlien::WriteMergingMacro() } if (list) delete list; } - out << "// include path" << endl; - if (fIncludePath.Length()) out << " gSystem->AddIncludePath(\"" << fIncludePath.Data() << "\");" << endl; - out << " gSystem->AddIncludePath(\"-I$ALICE_ROOT/include\");" << endl << endl; out << "// Load analysis framework libraries" << endl; if (!fPackages) { if (!fExecutableCommand.Contains("aliroot")) { @@ -2858,6 +3409,21 @@ void AliAnalysisAlien::WriteMergingMacro() out << " if (!" << setupPar << "(\"" << obj->GetName() << "\")) return;" << endl; } } + out << "// include path" << endl; + // Get the include path from the interpreter and remove entries pointing to AliRoot + out << " TString intPath = gInterpreter->GetIncludePath();" << endl; + out << " TObjArray *listpaths = intPath.Tokenize(\" \");" << endl; + out << " TIter nextpath(listpaths);" << endl; + out << " TObjString *pname;" << endl; + out << " while ((pname=(TObjString*)nextpath())) {" << endl; + out << " TString current = pname->GetName();" << endl; + out << " if (current.Contains(\"AliRoot\") || current.Contains(\"ALICE_ROOT\")) continue;" << endl; + out << " gSystem->AddIncludePath(current);" << endl; + out << " }" << endl; + out << " if (listpaths) delete listpaths;" << endl; + if (fIncludePath.Length()) out << " gSystem->AddIncludePath(\"" << fIncludePath.Data() << "\");" << endl; + out << " gROOT->ProcessLine(\".include $ALICE_ROOT/include\");" << endl; + out << " printf(\"Include path: %s\\n\", gSystem->GetIncludePath());" << endl << endl; if (fAdditionalLibs.Length()) { out << "// Add aditional AliRoot libraries" << endl; TObjArray *list = fAdditionalLibs.Tokenize(" "); @@ -2886,10 +3452,10 @@ void AliAnalysisAlien::WriteMergingMacro() Warning("WriteMergingMacro", "!!! You requested FastRead option. Using xrootd flags to reduce timeouts in the grid merging jobs. Note that this may skip some files that could be accessed !!!"); out << "// fast xrootd reading enabled" << endl; out << " printf(\"!!! You requested FastRead option. Using xrootd flags to reduce timeouts. Note that this may skip some files that could be accessed !!!\");" << endl; - out << " gEnv->SetValue(\"XNet.ConnectTimeout\",10);" << endl; - out << " gEnv->SetValue(\"XNet.RequestTimeout\",10);" << endl; + out << " gEnv->SetValue(\"XNet.ConnectTimeout\",50);" << endl; + out << " gEnv->SetValue(\"XNet.RequestTimeout\",50);" << endl; out << " gEnv->SetValue(\"XNet.MaxRedirectCount\",2);" << endl; - out << " gEnv->SetValue(\"XNet.ReconnectTimeout\",10);" << endl; + out << " gEnv->SetValue(\"XNet.ReconnectTimeout\",50);" << endl; out << " gEnv->SetValue(\"XNet.FirstConnectMaxCnt\",1);" << endl << endl; } // Change temp directory to current one @@ -2897,11 +3463,9 @@ void AliAnalysisAlien::WriteMergingMacro() out << " gSystem->Setenv(\"TMPDIR\", gSystem->pwd());" << endl << endl; out << "// Connect to AliEn" << endl; out << " if (!TGrid::Connect(\"alien://\")) return;" << endl; - out << " Bool_t laststage = kFALSE;" << endl; out << " TString outputDir = dir;" << endl; - out << " TString outputFiles = \"" << fOutputFiles << "\";" << endl; + out << " TString outputFiles = \"" << GetListOfFiles("out") << "\";" << endl; out << " TString mergeExcludes = \"" << fMergeExcludes << "\";" << endl; - out << " mergeExcludes += \"" << AliAnalysisManager::GetAnalysisManager()->GetExtraFiles() << "\";" << endl; out << " TObjArray *list = outputFiles.Tokenize(\",\");" << endl; out << " TIter *iter = new TIter(list);" << endl; out << " TObjString *str;" << endl; @@ -2918,13 +3482,11 @@ void AliAnalysisAlien::WriteMergingMacro() out << " continue;" << endl; out << " }" << endl; out << " if (mergeExcludes.Contains(outputFile.Data())) continue;" << endl; - out << " merged = AliAnalysisAlien::MergeOutput(outputFile, outputDir, " << fMaxMergeFiles << ", stage, ichunk);" << endl; + out << " merged = AliAnalysisAlien::MergeOutput(outputFile, outputDir, " << fMaxMergeFiles << ", stage);" << endl; out << " if (!merged) {" << endl; out << " printf(\"ERROR: Cannot merge %s\\n\", outputFile.Data());" << endl; out << " return;" << endl; out << " }" << endl; - out << " // Check if this was the last stage. If yes, run terminate for the tasks." << endl; - out << " if (!gSystem->AccessPathName(outputFile)) laststage = kTRUE;" << endl; out << " }" << endl; out << " // all outputs merged, validate" << endl; out << " ofstream out;" << endl; @@ -2933,7 +3495,7 @@ void AliAnalysisAlien::WriteMergingMacro() out << " // read the analysis manager from file" << endl; TString analysisFile = fExecutable; analysisFile.ReplaceAll(".sh", ".root"); - out << " if (!laststage) return;" << endl; + out << " if (laststage<10000) return;" << endl; out << " TFile *file = TFile::Open(\"" << analysisFile << "\");" << endl; out << " if (!file) return;" << endl; out << " TIter nextkey(file->GetListOfKeys());" << endl; @@ -2947,16 +3509,21 @@ void AliAnalysisAlien::WriteMergingMacro() out << " ::Error(\"" << func.Data() << "\", \"No analysis manager found in file" << analysisFile <<"\");" << endl; out << " return;" << endl; out << " }" << endl << endl; + out << " mgr->SetRunFromPath(mgr->GetRunFromAlienPath(dir));" << endl; out << " mgr->SetSkipTerminate(kFALSE);" << endl; out << " mgr->PrintStatus();" << endl; if (AliAnalysisManager::GetAnalysisManager()) { if (AliAnalysisManager::GetAnalysisManager()->GetDebugLevel()>3) { out << " gEnv->SetValue(\"XNet.Debug\", \"1\");" << endl; } else { - out << " AliLog::SetGlobalLogLevel(AliLog::kError);" << endl; + if (TestBit(AliAnalysisGrid::kTest)) + out << " AliLog::SetGlobalLogLevel(AliLog::kWarning);" << endl; + else + out << " AliLog::SetGlobalLogLevel(AliLog::kError);" << endl; } } - out << " mgr->StartAnalysis(\"gridterminate\");" << endl; + out << " TTree *tree = NULL;" << endl; + out << " mgr->StartAnalysis(\"gridterminate\", tree);" << endl; out << "}" << endl << endl; if (hasANALYSISalice) { out <<"//________________________________________________________________________________" << endl; @@ -2964,7 +3531,7 @@ void AliAnalysisAlien::WriteMergingMacro() out << "// Compile the package and set it up." << endl; out << " TString pkgdir = package;" << endl; out << " pkgdir.ReplaceAll(\".par\",\"\");" << endl; - out << " gSystem->Exec(Form(\"tar xvzf %s.par\", pkgdir.Data()));" << endl; + out << " gSystem->Exec(TString::Format(\"tar xvzf %s.par\", pkgdir.Data()));" << endl; out << " TString cdir = gSystem->WorkingDirectory();" << endl; out << " gSystem->ChangeDirectory(pkgdir);" << endl; out << " // Check for BUILD.sh and execute" << endl; @@ -3000,7 +3567,7 @@ void AliAnalysisAlien::WriteMergingMacro() } } Bool_t copy = kTRUE; - if (TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; + if (fProductionMode || TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; if (copy) { CdWork(); TString workdir = gGrid->GetHomeDirectory(); @@ -3019,7 +3586,7 @@ Bool_t AliAnalysisAlien::SetupPar(const char *package) // LD_LIBRARY_PATH TString pkgdir = package; pkgdir.ReplaceAll(".par",""); - gSystem->Exec(Form("tar xvzf %s.par", pkgdir.Data())); + gSystem->Exec(TString::Format("tar xzf %s.par", pkgdir.Data())); TString cdir = gSystem->WorkingDirectory(); gSystem->ChangeDirectory(pkgdir); // Check for BUILD.sh and execute @@ -3066,6 +3633,8 @@ void AliAnalysisAlien::WriteExecutable() return; } out << "#!/bin/bash" << endl; + // Make sure we can properly compile par files + out << "export LD_LIBRARY_PATH=.:$LD_LIBRARY_PATH" << endl; out << "echo \"=========================================\"" << endl; out << "echo \"############## PATH : ##############\"" << endl; out << "echo $PATH" << endl; @@ -3084,8 +3653,6 @@ void AliAnalysisAlien::WriteExecutable() out << "echo \"############## memory : ##############\"" << endl; out << "free -m" << endl; out << "echo \"=========================================\"" << endl << endl; - // Make sure we can properly compile par files - if (TObject::TestBit(AliAnalysisGrid::kUsePars)) out << "export LD_LIBRARY_PATH=.:$LD_LIBRARY_PATH" << endl; out << fExecutableCommand << " "; out << fAnalysisMacro.Data() << " " << fExecutableArgs.Data() << endl << endl; out << "echo \"======== " << fAnalysisMacro.Data() << " finished with exit code: $? ========\"" << endl; @@ -3093,7 +3660,7 @@ void AliAnalysisAlien::WriteExecutable() out << "free -m" << endl; } Bool_t copy = kTRUE; - if (TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; + if (fProductionMode || TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; if (copy) { CdWork(); TString workdir = gGrid->GetHomeDirectory(); @@ -3102,7 +3669,7 @@ void AliAnalysisAlien::WriteExecutable() workdir += fGridWorkingDir; TString executable = Form("%s/bin/%s", gGrid->GetHomeDirectory(), fExecutable.Data()); if (FileExists(executable)) gGrid->Rm(executable); - Info("CreateJDL", "\n##### Copying executable file <%s> to your AliEn bin directory", fExecutable.Data()); + Info("WriteExecutable", "\n##### Copying executable file <%s> to your AliEn bin directory", fExecutable.Data()); TFile::Cp(Form("file:%s",fExecutable.Data()), Form("alien://%s", executable.Data())); } } @@ -3122,6 +3689,8 @@ void AliAnalysisAlien::WriteMergeExecutable() return; } out << "#!/bin/bash" << endl; + // Make sure we can properly compile par files + out << "export LD_LIBRARY_PATH=.:$LD_LIBRARY_PATH" << endl; out << "echo \"=========================================\"" << endl; out << "echo \"############## PATH : ##############\"" << endl; out << "echo $PATH" << endl; @@ -3140,18 +3709,19 @@ void AliAnalysisAlien::WriteMergeExecutable() out << "echo \"############## memory : ##############\"" << endl; out << "free -m" << endl; out << "echo \"=========================================\"" << endl << endl; - // Make sure we can properly compile par files - if (TObject::TestBit(AliAnalysisGrid::kUsePars)) out << "export LD_LIBRARY_PATH=.:$LD_LIBRARY_PATH" << endl; TString mergeMacro = fExecutable; mergeMacro.ReplaceAll(".sh", "_merge.C"); - out << "export ARG=\"" << mergeMacro << "(\\\"$1\\\",$2,$3)\"" << endl; + if (IsOneStageMerging()) + out << "export ARG=\"" << mergeMacro << "(\\\"$1\\\")\"" << endl; + else + out << "export ARG=\"" << mergeMacro << "(\\\"$1\\\",$2,$3)\"" << endl; out << fExecutableCommand << " " << "$ARG" << endl; out << "echo \"======== " << mergeMacro.Data() << " finished with exit code: $? ========\"" << endl; out << "echo \"############## memory after: ##############\"" << endl; out << "free -m" << endl; } Bool_t copy = kTRUE; - if (TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; + if (fProductionMode || TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; if (copy) { CdWork(); TString workdir = gGrid->GetHomeDirectory(); @@ -3160,7 +3730,7 @@ void AliAnalysisAlien::WriteMergeExecutable() workdir += fGridWorkingDir; TString executable = Form("%s/bin/%s", gGrid->GetHomeDirectory(), mergeExec.Data()); if (FileExists(executable)) gGrid->Rm(executable); - Info("CreateJDL", "\n##### Copying executable file <%s> to your AliEn bin directory", mergeExec.Data()); + Info("WriteMergeExecutable", "\n##### Copying executable file <%s> to your AliEn bin directory", mergeExec.Data()); TFile::Cp(Form("file:%s",mergeExec.Data()), Form("alien://%s", executable.Data())); } } @@ -3178,7 +3748,9 @@ void AliAnalysisAlien::WriteProductionFile(const char *filename) const Error("WriteProductionFile", "Bad file name: %s", filename); return; } - TString workdir = gGrid->GetHomeDirectory(); + TString workdir; + if (!fProductionMode && !fGridWorkingDir.BeginsWith("/alice")) + workdir = gGrid->GetHomeDirectory(); workdir += fGridWorkingDir; Int_t njobspermaster = 1000*fNrunsPerMaster/fSplitMaxInputFileNumber; TString locjdl = Form("%s/%s", workdir.Data(),fJDLName.Data()); @@ -3192,9 +3764,11 @@ void AliAnalysisAlien::WriteProductionFile(const char *filename) const else out << Form("%s", fInputFiles->At(i)->GetName()) << " " << Form("%03d", i) << endl; } - Info("WriteProductionFile", "\n##### Copying production file <%s> to your work directory", filename); - if (FileExists(filename)) gGrid->Rm(filename); - TFile::Cp(Form("file:%s",filename), Form("alien://%s/%s", workdir.Data(),filename)); + if (gGrid) { + Info("WriteProductionFile", "\n##### Copying production file <%s> to your work directory", filename); + if (FileExists(filename)) gGrid->Rm(filename); + TFile::Cp(Form("file:%s",filename), Form("alien://%s/%s", workdir.Data(),filename)); + } } //______________________________________________________________________________ @@ -3203,13 +3777,20 @@ void AliAnalysisAlien::WriteValidationScript(Bool_t merge) // Generate the alien validation script. // Generate the validation script TObjString *os; - TString validationScript = fExecutable; - if (merge) validationScript.ReplaceAll(".sh", "_mergevalidation.sh"); - else validationScript.ReplaceAll(".sh", "_validation.sh"); + if (fValidationScript.IsNull()) { + fValidationScript = fExecutable; + fValidationScript.ReplaceAll(".sh", "_validation.sh"); + } + TString validationScript = fValidationScript; + if (merge) validationScript.ReplaceAll(".sh", "_merge.sh"); if (!Connect()) { Error("WriteValidationScript", "Alien connection required"); return; } + if (!fTerminateFiles.IsNull()) { + fTerminateFiles.Strip(); + fTerminateFiles.ReplaceAll(" ",","); + } TString outStream = ""; if (!TestBit(AliAnalysisGrid::kTest)) outStream = " >> stdout"; if (!TestBit(AliAnalysisGrid::kSubmit)) { @@ -3281,18 +3862,20 @@ void AliAnalysisAlien::WriteValidationScript(Bool_t merge) // Part dedicated to the specific analyses running into the train - TObjArray *arr = fOutputFiles.Tokenize(","); + TString outputFiles = fOutputFiles; + if (merge && !fTerminateFiles.IsNull()) { + outputFiles += ","; + outputFiles += fTerminateFiles; + } + TObjArray *arr = outputFiles.Tokenize(","); TIter next1(arr); TString outputFile; - AliAnalysisManager *mgr = AliAnalysisManager::GetAnalysisManager(); - TString extra = mgr->GetExtraFiles(); - while ((os=(TObjString*)next1())) { - if (merge) break; + while (!merge && (os=(TObjString*)next1())) { + // No need to validate outputs produced by merging since the merging macro does this outputFile = os->GetString(); Int_t index = outputFile.Index("@"); if (index > 0) outputFile.Remove(index); - if (merge && fMergeExcludes.Contains(outputFile)) continue; - if (extra.Contains(outputFile)) continue; + if (fTerminateFiles.Contains(outputFile)) continue; if (outputFile.Contains("*")) continue; out << "if ! [ -f " << outputFile.Data() << " ] ; then" << endl; out << " error=1" << endl; @@ -3322,12 +3905,12 @@ void AliAnalysisAlien::WriteValidationScript(Bool_t merge) out << "exit $error" << endl; } Bool_t copy = kTRUE; - if (TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; + if (fProductionMode || TestBit(AliAnalysisGrid::kOffline) || TestBit(AliAnalysisGrid::kTest)) copy = kFALSE; if (copy) { CdWork(); TString workdir = gGrid->GetHomeDirectory(); workdir += fGridWorkingDir; - Info("CreateJDL", "\n##### Copying validation script <%s> to your AliEn working space", validationScript.Data()); + Info("WriteValidationScript", "\n##### Copying validation script <%s> to your AliEn working space", validationScript.Data()); if (FileExists(validationScript)) gGrid->Rm(validationScript); TFile::Cp(Form("file:%s",validationScript.Data()), Form("alien://%s/%s", workdir.Data(),validationScript.Data())); }