Error in multiple file HDF5 output

[Peter Alt-Epping]

Hi all,

I am running a large simulation on an implicit unstructured grid (hex-mesh). I produce snapshot output as multiple HDF5 files. I have noticed that after the second or third snapshot output, the cell ID array in the output file (/Domain/Cells) becomes corrupted, that is, entries are either missing or are zeros. The data output is complete and looks normal - but I can only visualize the data if I replace the corrupted cell array with the correct array from the first snapshot output. 

The problem exists only when I output multiple HDF5 files, there is no problem with a single HDF5 file.

I am using PFLOTRAN6 and hdf5-1.14.3 compiled with an Intel compiler (the "classic" version).

Any help would be greatly appreciated.

Best,

Peter

 

Exerts from xmf files showing time varying "Dimensions" in multi-file output mode:

-- 
Peter Alt-Epping, PD PhD
RWI Group, Institute of Geological Sciences
University of Bern
Baltzerstrasse 3
3012 Bern
Switzerland

Tel.: 0041 31 684 8494

[Hammond, Glenn E]

Peter,

 

I am having difficulty reproducing the error.

 

1. Please send me the output from

 

> git log -1

 

executed in $PFLOTRAN_DIR.

 

2. Please send me the OUTPUT block in your input file.

 

Thanks,

 

Glenn

 

From: pflotra...@googlegroups.com <pflotra...@googlegroups.com> on behalf of Peter Alt-Epping <alt-e...@geo.unibe.ch>
Date: Tuesday, June 3, 2025 at 2:10 AM
To: 'Hammond, Glenn E' via pflotran-users <pflotra...@googlegroups.com>
Subject: [pflotran-users: 8410] Error in multiple file HDF5 output

Check twice before you click! This email originated from outside PNNL.

 

--
You received this message because you are subscribed to the Google Groups "pflotran-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pflotran-user...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pflotran-users/1f093579-e15e-498f-a703-53ee1cd00bf5%40geo.unibe.ch.

[Peter Alt-Epping]

Hi Glenn,

Thanks for looking into this. Please find the >git -log -1 output and the OUTPUT Block below.

There is no problem with smaller simulations - I suspect it could be a memory issue.

Best,

Peter

git log -1:

commit fa3ffa2533fb6a1fc50c992c34f4e1f0f2eae431 (HEAD -> master, origin/master, origin/HEAD)
Merge: 1e8b6c863 3381cfffc
Author: Rosie Leone <rle...@sandia.gov>
Date:   Wed May 28 19:56:22 2025 +0000

    Merged in david/fix-cc-destroy (pull request #1198)

    Initialize polynomial so it can be destroyed.

    Approved-by: Glenn Hammond

OUTPUT Block

 

OUTPUT
 TIMES y 1e-10 1000. 10000. 25000. 50000. 100000. 150000. 200000. 250000. 300000.
   FORMAT HDF5 MULTIPLE_FILES
   FLUXES
   MASS_BALANCE
   PRINT_PRIMAL_GRID
  VARIABLES
    TEMPERATURE
    LIQUID_PRESSURE
    LIQUID_DENSITY
    POROSITY
    PERMEABILITY
    VOLUME
    MATERIAL_ID
 /
   PERIODIC_OBSERVATION TIMESTEP 10
 END

To view this discussion visit https://groups.google.com/d/msgid/pflotran-users/PH0PR09MB7836EBCFF93C4069BC53C9039A6CA%40PH0PR09MB7836.namprd09.prod.outlook.com.