Same file in several workers

~Dear proof experts, (this is not to report a problem but to verify a feature)

Just by chance I was checking missing files in my job analysis using proof and I found that exactly the same file is opened in different workers (please see log output below). At the beginning I was quite alarmed about that because I am doing a cross section measurement. Also that does not happened very often, only for the last run I was analysis and only for a few files in that run. However, and triggered in part but this, after counting the number of events for that particular run I checked the event if the file is sent an opened in different workers at the end the total number of events for the corresponding run is correct.

So my question is, does proof not only splits the files among the workers but also the events in a particular files can be split among several workers?

Thanks a lot in advance for the answer,

best regards


Log file:


Packets are not file-wide, so packets (i.e. sub-sets of files) of the same file can be assigned to different workers. Smaller the number of files, more probable this is, because PROOF tries to keep assigning packets from a given file to the worker which has the file already open.
How many files are you processing?

G. Ganis

Thanks a lot for the answer. That explains everything. I am reading 29052 files. One further question, is there a way to know how many files I am analyzing? With so many files I would like to check I am not missing any. So far I was increasing a counter after each call to Notify() but that does not work with that scenario.


You have a list of missing files at the end:

root [] gProof->ShowMissingFiles()
root [] TFileCollection *fc = gProof->GetMissingFiles()

which, compared with the initial number, gives you the number of files analyzed.

G. Ganis