[gmx-users] Magic number error
chris.neale at utoronto.ca
Sat Jul 14 19:29:36 CEST 2007
>Program trjconv, VERSION 3.3.1
>Source code file: xtcio.c, line: 83
>Magic Number Error in XTC file (read 0, should be
If you are talking about trjconv -e while searching more than 2Gb into the file then it is a known issue
(see bug #126 http://www.gromacs.org/component/option,com_wrapper/Itemid,157/) and is fixed in CVS but not
in the standard downloadable versions yet (to 3.3.1).
If you are talking about a different case, please send lots more data (but do a search first please).
I have experienced such problems but that has something to do with our cluster and not gromacs.
For us, different clusters work fine for gromacs but it is only gromacs that has this problem on the problematic cluster.
Our problem is definately due to an overloaded NFS server and we appear to have solved the problem by writing
to /scratch (local node space) during the run. We are still looking into the issue but it may be a problem of
NFS delay and not NFS corruption and then the corruption occurs when we try to trjconv or gmxcheck a file
that is not yet completely written.
More information about the gromacs.org_gmx-users