[gmx-users] Closing of .trr and .edr files after running out ofspace

[chris.neale at utoronto.ca]

Dear Sarah:

 From both of your posts, it appears as if you have corruption around  
this time (40 ns). Possibly your disk was temporarily full near that  
time.

You could try to excise that portion with -b and -e combinations:

trjconv -b 0 -e 40000 -o a.trr
trjconv -b 42000 -e 150000 -o b.trr

but I doubt that this will work.

If you need to convince yourself of corruption, you can always run  
some tool like g_dist (or g_energy for the .edr) that writes out a  
time-series and see where it stops.

Chris.

Thank you very much, that was a good idea!
I tried ~/gromacs-4.0.4/bin/trjconv -f dmpc.trr -e 150000 -o  
dmpc-150ns.trr to get the first 150 ns out of the simulation.
Trjconv writes this to me:

Will write trr: Trajectory in portable xdr format
trn version: GMX_trn_file (single precision)
  ->  frame   4210 time 42110.004        ->  frame   4000 time 40010.000
-------------------------------------------------------
Program trjconv, VERSION 4.0.4
Source code file: trnio.c, line: 66

File input/output error:
Can not determine precision of trn file
-------------------------------------------------------

When checking the new .trr file by ~/gromacs-4.0.4/bin/gmxcheck -f  
dmpc-150ns.trr the output is:

Checking file dmpc-150ns.trr
trn version: GMX_trn_file (single precision)
Reading frame       0 time   10.000
# Atoms  35513
Reading frame    4000 time 40010.000


Item        #frames Timestep (ps)
Step          4220    10
Time          4220    10
Lambda        4220    10
Coords        4220    10
Velocities    4220    10
Forces           0
Box           4220    10

So this means that the .trr only conatins 42.2 ns, right?

Best,
Sarah

-----Original Message-----
From: gmx-users-bounces at gromacs.org on behalf of chris.neale at utoronto.ca
Sent: Tue 09-02-2010 16:38
To: gmx-users at gromacs.org
Subject: [gmx-users] Closing of .trr and .edr files after running out ofspace

trjconv -e for the .trr
eneconv -e for the .edr

--original message --

Dear gmx-users,

I have been very unfortunate (and stupid): I was running a simulation
of 220 ns and due to limited space at our cluster-computer I was
writing the .trr file and .edr file directly to another disc. That was
un-clever since I did not write the .cpt file to that disc as well. As
I am not the only one using the storage disc it ran out of space very
shortly before the end of my simulation and I got this error message
in the .log file:
    Epot (kJ/mol)        Coul-SR          LJ-SR        Coul-14          LJ-14
        DMPC-DMPC    2.60809e+02    2.19979e+02    3.95442e+02    7.50289e+01
         DMPC-SOL    6.74362e+02    2.11485e+02    0.00000e+00    0.00000e+00
          SOL-SOL    9.65717e+02    6.15314e+02    0.00000e+00    0.00000e+00

           T-DMPC          T-SOL
      2.81532e+00    1.30102e+00


-------------------------------------------------------
Program mdrun_mpi, VERSION 4.0.4
Source code file: enxio.c, line: 212

File input/output error:
Cannot close energy file; it might be corrupt, or maybe you are out of quota?
-------------------------------------------------------



I guess it is unable to close due to a lack of space? When running
gmxcheck on the generated .trr file this message was printed:

Checking file dmpc.trr
trn version: GMX_trn_file (single precision)
Reading frame       0 time   10.000
# Atoms  35513
Reading frame    4000 time 40010.000
-------------------------------------------------------
Program gmxcheck, VERSION 4.0.4
Source code file: trnio.c, line: 66

File input/output error:
Can not determine precision of trn file
-------------------------------------------------------

This .trr file should be 220 ns and 18 G big, it is now 15 G.

Even though the .cpt file is only set to update every 15 min (default)
there is only 3 min between state_previous.cpt and state.cpt. I think
it is because it wrote a new .cpt just when finishing the run? But
then non of the .cpt files are "old" enough to allow rerun from the
point of the problem. Stupid of me.

Now, the simulations was supposed to be of 220 ns, but actually I only
need around 150 ns for my purpose. So my question is whether there is
anyy way to "close" the .trr file at the point it is at and then have
a functional .trr file?

Best regards,
Sarah

--