Greetings,
Recently I’ve stumbled upon some issues while reading ROOT files.
An example of such an issue is here:
root -l /eos/nica/mpd/sim/data/exp/dst-BiBi-07.7GeV-mp09-20-500ev/BiBi/07.7GeV-mb/UrQMD/BiBi-07.0GeV-mp09-20-500ev-pwg3/urqmd-BiBi-07.7GeV-mb-eos0-500-6655.reco.root
And I get:
Attaching file urqmd-BiBi-07.7GeV-mb-eos0-500-6655.reco.root as _file0 …
Error in TFile::Init: urqmd-BiBi-07.7GeV-mb-eos0-500-6655.reco.root not a ROOT file
(TFile *) nullptr
The file size itself seems to be fine but ROOT cannot see any TObject in it.
Can anyone knows how this issue can be resolved?
It is one of the cases of broken files from all mpd-mass-productions I have tested.
Either file is missing, or is not recognized as a ROOT-file, or is a Zombie, or extracting the data while reading yields basket errors and crashes.
AFAIK no consistent check for all of those errors is performed after they are copied from LIT. (I may be wrong)
It is my personal opinion that it is necessary.
From this data set I can say that about 8-9% of the files are broken. From the first million events (2000 files) 910000 are available.
P.S.
we cannot read the submitted ticket it seems
“You are not allowed to view this ticket.”
I check the files for size match when copied.
This is not a problem of copying, when communicating with other colleagues, everything usually looks good at the beginning, and only after a while, damaged files appear.
From time to time, such problems happen on eos. But not so much - 8-9%. It happens that when the file servers are turned off, the files are not only damaged, but also cannot be restored, in this case the LIT administrators send a list of damaged files and I recalculate these files.
I don’t do regular monitoring of storages, there is a lot of other work and the main storage is located in LIT, where administrators monitor eos and report corrupted files.
Send me a list of damaged files, I will copy them from LIT storage.
I can’t see this ticket either.