Re: Problems with hadd

From: Philippe Canal <pcanal_at_fnal.gov>
Date: Wed, 25 Jan 2012 11:20:11 -0600


Hi Duc,

> Can it happen that the trees are merged, but not
> the histograms?

No there is no (expected) way that this could happen so the problem is likely somewhere else.

Cheers,
Philippe.

On 11/22/11 8:43 AM, Duc Bao Ta wrote:
> Hi all,
>
> I have a question regarding hadd in 5.26.00 and merging files with
> trees and histograms. Can it happen that the trees are merged, but not
> the histograms?
>
> I saw the following problem:
> I run my script and create several root files (output.X.root) in which
> the number of total entries in the tree is counted in a special
> histogram. Then I merge those files (output.root) and then I slim the
> tree (output.slim.root). For some calculation (luminosity of the
> sample) I need to remember the total number of original entries and
> for saving some space I remove some entries from the tree
> I found in some cases that right after the merging the number of
> entries in the tree were smaller as the entries in the histogram.
>
> Unfortunately, I cannot reproduce it right away, because I ran on the
> grid over a large number of files. The original files and the merged
> file are deleted and I only have the final output.slim.root file. So
> far, I have only checked one job and I was lucky that I wrote out the
> number of entries in the log files. I have not seen any error message
> from hadd in the logs.
>
> Anyway, I was wondering if that could theoretically happen or if I
> should try to track down the problem somewhere else.
>
> Thanks
> Duc
>
Received on Wed Jan 25 2012 - 18:20:18 CET

This archive was generated by hypermail 2.2.0 : Wed Jan 25 2012 - 23:50:01 CET