miniaod: cannot read RunIIFall17 MINIAOD with CMSSW_9_4_11_cand1 #25416
Comments
|
A new Issue was created by @cbernet Colin Bernet. @davidlange6, @Dr15Jones, @smuzaffar, @fabiocos, @kpedro88 can you please review it and eventually sign/assign? Thanks. cms-bot commands are listed here |
|
@cbernet is there a reason you are using the miniAODv1 rather than the miniAODv2? |
|
@kpedro88 no particular reason (actually this file was provided by somebody else, I had a v2 file there before and I had not noticed the change). I just tried with v2, and the crash still occurs, unfortunately. You can reproduce with this file: |
|
assign analysis |
cmsbuild
added
analysis-pending
pending-signatures
and removed
pending-assignment
labels
|
The following code does work #include "FWCore/FWLite/interface/FWLiteEnabler.h"
#include "DataFormats/HepMCCandidate/interface/GenParticle.h"
#include "TFile.h"
#include "TTree.h"
#include <vector>
#include <memory>
#include <cassert>
#include <iostream>
#include <cmath>
int main() {
FWLiteEnabler::enable();
//std::unique_ptr<TFile> f{ TFile::Open("root://cms-xrd-global.cern.ch//store/mc/RunIIFall17MiniAOD/DYJetsToLL_M-50_TuneCP5_13TeV-madgraphMLM-pythia8/MINIAODSIM/RECOSIMstep_94X_mc2017_realistic_v10-v1/00000/0293A280-B5F3-E711-8303-3417EBE33927.root") };
std::unique_ptr<TFile> f{ TFile::Open("0293A280-B5F3-E711-8303-3417EBE33927.root") };
auto events = dynamic_cast<TTree*>(f->Get("Events"));
assert(events != nullptr);
auto branch = events->GetBranch("recoGenParticles_prunedGenParticles__PAT.obj");
std::vector<reco::GenParticle> particles;
auto* pParticles = &particles;
int count = 0;
for(long e = 0; e< events->GetEntries(); ++e) {
if( (e % 100) == 0 ) {
std::cout <<"event "<<e<<std::endl;
}
events->GetEntry(e);
branch->SetAddress(pParticles);
branch->GetEntry(e);
for(auto const& p: particles) {
if ( std::abs(p.pdgId()) != 15 or not p.statusFlags().isPrompt() ) {
continue ;
}
//std::cout <<"found tau "<<std::endl;
for( int i=0; i< p.numberOfDaughters(); ++i) {
auto const d = p.daughter(i);
if( nullptr == d) {
std::cout <<"***missing daughter "<<i<<std::endl;
} else {
auto v = d->pdgId();
if( (count % 100) == 0 ) {
std::cout<< " daughter "<<i<<" "<<v<<std::endl;
}
}
}
++count;
}
}
return 0;
}where I compiled it in a CMSSW_9_4_11_cand1 workarea using |
|
For the C++ code, If I just tried to call That comes from the 'bare' FWLite handler, which is different from using |
|
The following also works #include "DataFormats/FWLite/interface/Event.h"
#include "DataFormats/FWLite/interface/Handle.h"
#include "DataFormats/HepMCCandidate/interface/GenParticle.h"
#include "TFile.h"
#include "TTree.h"
#include <vector>
#include <memory>
#include <cassert>
#include <iostream>
#include <cmath>
int main() {
//std::unique_ptr<TFile> f{ TFile::Open("root://cms-xrd-global.cern.ch//store/mc/RunIIFall17MiniAOD/DYJetsToLL_M-50_TuneCP5_13TeV-madgraphMLM-pythia8/MINIAODSIM/RECOSIMstep_94X_mc2017_realistic_v10-v1/00000/0293A280-B5F3-E711-8303-3417EBE33927.root") };
std::unique_ptr<TFile> f{ TFile::Open("0293A280-B5F3-E711-8303-3417EBE33927.root") };
fwlite::Event event(f.get());
int count = 0;
int e = 0;
for(event.toBegin(); ! event.atEnd(); ++event) {
if( (e % 100) == 0 ) {
std::cout <<"event "<<e<<std::endl;
}
++e;
fwlite::Handle<std::vector<reco::GenParticle>> particles;
particles.getByLabel(event, "prunedGenParticles");
for(auto const& p: *particles) {
if ( std::abs(p.pdgId()) != 15 or not p.statusFlags().isPrompt() ) {
continue ;
}
//std::cout <<"found tau "<<std::endl;
for( int i=0; i< p.numberOfDaughters(); ++i) {
auto const d = p.daughter(i);
if( nullptr == d) {
std::cout <<"***missing daughter "<<i<<std::endl;
} else {
auto v = d->pdgId();
if( (count % 100) == 0 ) {
std::cout<< " daughter "<<i<<" "<<v<<std::endl;
}
}
}
++count;
}
}
return 0;
}compiled using Also it runs substantially faster than the 'bare' case I first made :). |
|
For the fwlite::Handle<std::vector<reco::Conversion>> conversions;
conversions.getByLabel(event, "reducedEgamma","reducedConversions");
assert(conversions.isValid());before getting the |
|
If I add The updated test program is #include "FWCore/FWLite/interface/FWLiteEnabler.h"
#include "DataFormats/HepMCCandidate/interface/GenParticle.h"
#include "DataFormats/EgammaCandidates/interface/Conversion.h"
#include "TFile.h"
#include "TTree.h"
#include <vector>
#include <memory>
#include <cassert>
#include <iostream>
#include <cmath>
int main() {
FWLiteEnabler::enable();
//std::unique_ptr<TFile> f{ TFile::Open("root://cms-xrd-global.cern.ch//store/mc/RunIIFall17MiniAOD/DYJetsToLL_M-50_TuneCP5_13TeV-madgraphMLM-pythia8/MINIAODSIM/RECOSIMstep_94X_mc2017_realistic_v10-v1/00000/0293A280-B5F3-E711-8303-3417EBE33927.root") };
std::unique_ptr<TFile> f{ TFile::Open("0293A280-B5F3-E711-8303-3417EBE33927.root") };
auto events = dynamic_cast<TTree*>(f->Get("Events"));
assert(events != nullptr);
auto branch = events->GetBranch("recoGenParticles_prunedGenParticles__PAT.obj");
std::vector<reco::GenParticle> particles;
auto* pParticles = &particles;
auto conversionBranch = events->GetBranch("recoConversions_reducedEgamma_reducedConversions_PAT.obj");
std::vector<reco::Conversion> conversions;
auto* pConversions = &conversions;
int count = 0;
for(long e = 0; e< events->GetEntries(); ++e) {
if( (e % 100) == 0 ) {
std::cout <<"event "<<e<<std::endl;
}
events->GetEntry(e);
conversionBranch->SetAddress(pConversions);
conversionBranch->GetEntry(e);
branch->SetAddress(pParticles);
branch->GetEntry(e);
for(auto const& p: particles) {
if ( std::abs(p.pdgId()) != 15 or not p.statusFlags().isPrompt() ) {
continue ;
}
//std::cout <<"found tau "<<std::endl;
for( int i=0; i< p.numberOfDaughters(); ++i) {
auto const d = p.daughter(i);
if( nullptr == d) {
std::cout <<"***missing daughter "<<i<<std::endl;
} else {
auto v = d->pdgId();
if( (count % 100) == 0 ) {
std::cout<< " daughter "<<i<<" "<<v<<std::endl;
}
}
}
++count;
}
}
return 0;
}compiled with |
|
Running valgrind on the failing 'bare' FWLite give which isn't as useful as one could have hoped. |
|
Thanks a lot for the very fast fix, @Dr15Jones and @pcanal , indeed it works. now trying it in the context of our analysis. |
|
So is this issue solved? It seems so |
|
Hi Fabio, yes, it works now. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hello, we encountered a very puzzling issue with the script below.
To test it, install a vanilla
CMSSW_9_4_11_cand1, initialize your environment, and just run the script.The problem is that when conversions are read, the gen particles get corrupted. In particular, all tau daughters cannot be accessed. After a segmentation fault, the script prints:
Are we doing anything wrong? This seems to point to a possible incompatibility between RunIIFall17 MINIAOD and CMSSW_9_4_11_cand1. If this incompatibility is known, what is the last release that can read RunIIFall17 MINIAOD?
Many thanks,
Colin, @lucastorterotot
ccing
@arizzi for mini AOD,
@guitargeek for conversions,
@Dr15Jones since the problem occurs in FWLite, maybe we are not initializing it properly
The text was updated successfully, but these errors were encountered: