Blake, The stan cluster at IU does not have this limitation either in local ext2/3 mounts or when they are nfs mounted. I need to look into this further. I will try to reproduce the problem. It may have to do with which clib versions get called (open vs open64, fread/fwrite vs fread64/fwrite64). I will get back to you today. -Richard Jones Blake Leverington wrote: > Thanks Richard, > > The issue I ran into is directly related to hddm files. I do not produce > any hbook or other. Why I suspected a 2GB limit was when I looked at > the log files from the jobs I had run I see at the very end "Filesize > limit exceeded". I look at the hddm file size and it's slightly larger > than 2GB. This happened for both bggen and hdgeant. I had thought, like > you said, that hddm was unlimited, so now I'm suspecting it's the NFS > filesystem I'm running on (stan@indiana). I guess I'll just have to > split up my jobs more. > > Cheers, > -Blake > > Richard Jones wrote: > >> Hello all, >> >> There are no file size limits in HDDM, apart from those issuing from >> the OS/filesystem on which the file is being written. I know there is >> no 2GB cutoff because I routinely exceed that limit. The design of >> HDDM has unlimited stream length because it is sequential access. >> Some people, myself included, like to write out ntuples as a >> side-effect of the simulation, either to monitor conditions in the >> simulation or do special simulation studies. If you are looking at >> the size of the hbook file produced in the simulation, I know that >> hbook has a hard 2GB filesize limit. Is that what you are running into? >> >> -Richard Jones >> >> >>
S/MIME Cryptographic Signature