[gmx-users] Incomplete frame,while running trjconv .....

[Mark Abraham]

pavan payghan wrote:
> HI
>     i have to create .xtc file to save my disk space,but while i was 
> running it using trjconv i got warning message 
>     "incomplete frame".Actually i was running my system on four nodes 
> with mdrun_mpi but it stopped due to 
>      unavailability of disk space so from this it is certain that my 
> system run well and ,when i tried to run trjconv i got such error    
>      message ,surprisingly this incomplete frame is in the middle 
> portion of the simulation.
>      if this frame was incomplete how come mdrun will continue the 
> simulation after that time frame?

A complete frame was not written to the trajectory file on disk, but it 
was fine in the program's memory, and so the run continued. The magic in 
modern buffered I/O means there is no warrant issued by the system that 
instructions to write to disk will be honoured or the program exit.

>      what else reason may be behind this and wht are the possible ways 
> to overcome this problem?

Get more disk. Plan your usage better. See 
http://www.gromacs.org/Documentation/How-tos/Reducing_Trajectory_Storage_Volume

>      another thing im using pressure coupling during production mdrun 
> has it been the reason ?

No relevance.

>      can u also explain why we generally cut off pressure coupling 
> during production mdrun?

We generally try to model a physical system, and either validate a 
methodology against experimental data, or re-use a validated protocol. 
In either case, an NVT or NPT or other ensemble might be indicated, but 
it depends on what you're doing.

Mark