Write failure during repack on file

[zonexo]

Hi,

I'm using tec v112 to output binary files in parallel (48 - 192 cores) from my CFD code.

Recently, I kept getting this error:

Err: (TECEND112) Write failure during repack on file 1.

The result is that some files are corrupted and can't be used.

May I know why this problem occur? It happens on some of my clusters but not all, so I'm wondering if it could be the cluster's problem too.

Can anyone enlighten me?

Thanks.

[davetaflin]

Hi,

Tecio stores most of the data fed to it in a temporary file, then appends the contents of that temporary file to the .plt file when tecend is called. This error message is reported if one of two circumstances occurs: (1) Unable to read the temporary file, or (2) unable to write to the .plt file. I'm going to guess that (2) is the culprit, due to running out of disk space. Please make sure that there is enough disk space to hold a file twice the size of the final .plt file, or else use a different temporary directory that has more available space (the temp directory is specified in the call to tecini).

Dave

[zonexo]

Ok, but there's really lots of space left during write. So maybe there's some problems with the hdd. We will test.

Btw, I'm currently writing output sequentially in tecplot almost I'm running in parrallel MPI with each cpu having its own partition, and it corresponds to each tecplot file.

So I'm writing to file 1, then file 2 etc.

Will it work if I simply run it as a parallel job, ie writing to all the files at the same time?

Thanks.