[grath]

Hello,

I am working with Converge to perform a Large Eddy Simulation, but my job crashes during post-processing writing.

I have checked the converge.log file in the main folder and in output_original, but there is nothing unusual, as if the job stopped without an error.

However, in the file JOBNAME.eJOBID, I see an MPI error, and in JOBID.output, I notice that the crash always occurs during post-processing writing "Writting Post path/post.h5". I have tried several solvers, but the issue persists.

Is this error common with a very fine mesh or MPI libraries?

Specifications:
- I Run on high performance cluster.
- Converge V4.1.0.
- Fine Mesh, Order of Magnitude 1e7 cells.

Thanks.

[ksrivast]

Greetings,

Is the case setup crashing when writing out the very first post.h5 at the start of the simulation? Are there any abort_trace_rank*.out or hdf5_rank*.err files created in the outputs_original or an error_log folder?

I'd recommend checking whether you are running out of memory when writing out the post file. Also check if your HPC is preventing you from writing out large files. 10million cell count post files can take up a lot of memory.

Sincerely,

[grath]

Kislaya,
Thanks for your reply.

I do not have this type of file in my log folders.

This error does not always appear at the same time step, but I spotted another memory error about opened files...
So, I will investigate a bit more before drawing conclusions.

Have a good day.

Sincerely.