potential cause for "Low Level I/O Error"
James T. Potemra
jimp at HAWAII.EDU
Wed Apr 23 13:57:11 EDT 2008
Arlindo:
Thanks for your reply.
A little more information for you. First, the file size is correct,
10124352000 bytes.
Second, I can read the first six time steps of this file, and then more
than half-way
through the seventh before I get the seg fault. What is curious to me
is that when I
read the data via a fortran program, it all looks okay. If I then
simply write it to a
new file, grads can read it without problem. Finally, I can
successfully read the data
using both grads-1.9b4 and grads-2.0.a1, but I get the seg fault only
when using
grads-1.9.0-rc1.
Jim
Arlindo da Silva wrote:
> On Tue, Apr 22, 2008 at 8:05 PM, James T. Potemra <jimp at hawaii.edu
> <mailto:jimp at hawaii.edu>> wrote:
>
> I get the following error when trying to access parts of a binary
> file:
>
> Low Level I/O Error: Seek error on data file
> Data file name =
> /disk/c2/ofes/Tropical_Pacific/hbudget/post_4601/source.dta
> 657 rc=-1 pos=536871600 pfi->fhdr =0
> Segmentation fault
>
> GrADS then exits upon encountering this problem. The file contains
> a single variable with (x,y,z,t) dimensions of 3600x420x54x31 and I
> get the error when trying to access beyond time=7, level=32, y=32.
> I wrote a script to read the data file, and there does not appear
> to be
> anything strange (e.g., NaN's). However, the data are extremely small
> (order 10^-19). Could this be the problem? Or, is there a meaning to
> the error that would help sort this out? Thanks,
>
>
> Are you sure this is a "big_endian" file? Having the wrong endianess
> may cause this kind of problems. Is the file size what you expect it
> to be? It should be 3600*420*54*31*4=10124352000 bytes; if it
> is (3600*420+8)*54*31*4=10124405568 bytes than your file is likely to
> be sequential (add "sequential" to OPTIONS). Anything else indicates a
> problem writing the file.
>
> Arlindo
>
> --
> Arlindo da Silva
> dasilva at alum.mit.edu <mailto:dasilva at alum.mit.edu>
More information about the gradsusr
mailing list