New open() for DCDs in v.2.6 ?

From: Nicholas M Glykos (
Date: Thu Jan 29 2009 - 11:34:37 CST

Dear Developers,

Using NAMD_2.6_amd64 on stateless Beowulf cluster based on Caos NSA with
the xfs filesystem, I'm getting some unexpected results when trying to
read a growing DCD file (from an active job) :

 - The DCD header always reports a small and constant number of frames
   (3), although the DCD file grows normally.

 - When using carma to quickly view the growing trajectory, the DCD
   doesn't look normal with portions of the structure disappearing &
   re-appearing (but note that carma does show all frames that are really
   present in the DCD because it ignores the header's 'number of frames').

The interesting thing is that when I kill the job and the DCD file is
properly closed, the DCD file looks fine on all accounts (correct number
of frames, correct structures, etc).

Because these tests are part of checking & setting-up this cluster, I
wonder whether these findings are due to a difference between how v.2.5
and v.2.6 of NAMD handle the DCD file, or whether they are operating
system dependant (ie. I've messed it up while configuring the cluster, for
example with xfs or nfs).

Thank you for your time,

             Dr Nicholas M. Glykos, Department of Molecular 
         Biology and Genetics, Democritus University of Thrace,
   University Campus, 68100 Alexandroupolis, Greece, Fax +302551030613
      Tel ++302551030620 (77620),

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:52:18 CST