TDataFrame segfault on error reading file [2]

ROOT
1

Connected to: TDataFrame segfault on error reading file

I’ve now upgraded to ROOT 6.14, and I still get segfaults.

ROOT Version: 6.14/01
Platform: LCG dev4
Compiler: GCC 7.3

2

Hi,

what is the code you are running? This is something unexpected and if reproducible to be fixed asap.

Cheers,
D

3

If you have access to ATLAS code:

The code is at https://gitlab.cern.ch/OxfordATLASExotics/hh4b-resolved-reconstruction.
The file it segfaults on is user.bstanisl:user.bstanisl.13910762._000304.MiniNTuple.root

4

Hi,

apologies, I missed the reply :frowning:
I think I don’t have access: can the file be shared with cernbox together with 20 lines of code reproducing the crash?

D

5

Huh, running on just that file seems to be working (it hasn’t finished yet).

Nevertheless, with a missing file (/ incorrect filename):

Error in <TExMap::Add>: key 140639610989984 is not unique
Error in <TFile::TFile>: file /data/atlas/atlasdata2/Exotics/HH4b/Ntuples/Rel21-Group-NTuples-2018-05/user.bstanisl.HH4B.2018-04-30T0908Z.data16-rel21_MiniNTuple.root/user.bstanisl:user.bstanisl.13910762._000304.MiniNTuple.root does not exist
Error in <TFile::TFile>: file /data/atlas/atlasdata2/Exotics/HH4b/Ntuples/Rel21-Group-NTuples-2018-05/user.bstanisl.HH4B.2018-04-30T0908Z.data16-rel21_MiniNTuple.root/user.bstanisl:user.bstanisl.13910762._000304.MiniNTuple.root does not exist
terminate called after throwing an instance of 'std::runtime_error'
  what():  Unknown columns: runNumber,lumiBlock

Though the TExMap error also appears when the file name is correct, and doesn’t cause a crash.

6

Hi,

I think I need that file in order to debug: are you sure it’s not corrupt? It looks like it.

Best,
D

7

Here: https://cernbox.cern.ch/index.php/s/rpCQG4Iq52ikDLX

But as I said, when I run on just that file, it doesn’t crash, though it does print the TExMap error.
The output above was when I got the filename wrong (by leaving the user.bstanisl: in at the beginning). Evidently, ROOT still crashes if the file does not exist / is empty.

8

Just FYI, here are some of the errors:

Error in <TBranchElement::GetBasket>: File: /data/atlas/atlasdata2/Exotics/HH4b/Ntuples/Rel21-Group-NTuples-2018-05/user.bstanisl.HH4B.2018-04-30T0908Z.data16-rel21_MiniNTuple.root/user.bstanisl.13910762._000084.MiniNTuple.root at byte:389515562, branch:resolvedJets_phi, entry:314988, badread=1, nerrors=5, basketnumber=397
Error in <TBranchElement::GetBasket>: File: /data/atlas/atlasdata2/Exotics/HH4b/Ntuples/Rel21-Group-NTuples-2018-05/user.bstanisl.HH4B.2018-04-30T0908Z.data16-rel21_MiniNTuple.root/user.bstanisl.13910762._000084.MiniNTuple.root at byte:389477808, branch:resolvedJets_E, entry:314988, badread=1, nerrors=6, basketnumber=397

 *** Break *** segmentation violation
Error in <TBranchElement::GetBasket>: File: /data/atlas/atlasdata2/Exotics/HH4b/Ntuples/Rel21-Group-NTuples-2018-05/user.bstanisl.HH4B.2018-04-30T0908Z.data16-rel21_MiniNTuple.root/user.bstanisl.13910762._000180.MiniNTuple.root at byte:123347244, branch:passedTriggers, entry:99930, badread=1, nerrors=7, basketnumber=1478

 *** Break *** segmentation violation
Error in <TBranchElement::GetBasket>: File: /data/atlas/atlasdata2/Exotics/HH4b/Ntuples/Rel21-Group-NTuples-2018-05/user.bstanisl.HH4B.2018-04-30T0908Z.data16-rel21_MiniNTuple.root/user.bstanisl.13910762._000609.MiniNTuple.root at byte:83520267, branch:passedTriggers, entry:67808, badread=1, nerrors=8, basketnumber=1019

 *** Break *** segmentation violation

I’m beginning to think it isn’t actually caused by corrupt files, but by some sort of issue on the cluster.