No .e export file generated

[manpan-1]

I am not getting a .e file with the results after an analysis is successfully completed. I am getting all the "DICe_solution_XX.txt" files in the results folder. Looking at the stdout of dice, no errors or warnings occur. Though it gives:

[...]

[DICe_DEBUG]: create_exodus_mesh(): creating an exodus mesh
[DICe_DEBUG]: Output file name: DICe_solution.e
[DICe_DEBUG]: Creating the element and node field maps for the mesh
[DICe_DEBUG]:   ------------------ Analysis Model Definition (processsor 0) --------------------------
[DICe_DEBUG]:   Title:             autogenerated Triangle mesh, or subset points
[DICe_DEBUG]:   Output file:       DICe_solution.e
[DICe_DEBUG]:   Spatial dimension: 2
[DICe_DEBUG]:   Nodes:             1438
[DICe_DEBUG]:   Elements:          1438
[DICe_DEBUG]:   Dirichlet nodes:   0
[DICe_DEBUG]:   Neumann nodes:     0
[DICe_DEBUG]:   Node sets:         0
[DICe_DEBUG]:   --------------------------------------------------------------------------------------
[...]

this file seems like it's never created...

I have just built DICe from the latest commit on openSUSE Leap 15.1. I attach some relevant files of an analysis. 2d_plate_with_hole_no_export.zip

[dicengine]

A couple things to check: you don't have write_exodus_output set to false in the params.xml or input.xml files you do have DICE_ENABLE_GLOBAL set to ON in your do-cmake script when you built DICe

Exodus files only exist in memory until they get closed at which point they are written to disk. As you point out, it looks like the exodus mesh is getting created which is good, but it will only get written to file if DICE_ENABLE_GLOBAL is on.

[dicengine]

From looking at your output from running DICe on this problem it appears it is failing silently after the post processors are run, before the output step happens. Can you run the problem with valgrind or some other tool to see if a memory error has happened? I don't think the analysis is completing successfully for this problem.

[manpan-1]

So, here is the out and err with valgrind --leak-check=full dice.log

I surely see that something goes wrong (by the summary at the end) but I don't really have the experience to figure out anything more!

[dicengine]

The results from valgrind aren’t showing any major issues. The leak summary at the end is somewhat of a false positive. You will see similar results with any code you run valgrind on with leak-check=full. For some reason the code is exiting without throwing an error.

My hunch is that something is going wrong with your Trilinos build. Which version of Trilinos are you building and can you share your configuration cmake file for trilinos and DICe?

On Nov 8, 2019, at 3:22 PM, Panagiotis Manoleas <[email protected]mailto:[email protected]> wrote:

So, here is the out and err with valgrind --leak-check=full dice.loghttps://github.com/dicengine/dice/files/3826492/dice.log

I surely see that something goes wrong (by the summary at the end) but I don't really have the experience to figure out anything more!

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/dicengine/dice/issues/147?email_source=notifications&email_token=ADT7TOW6GZ7RNRTDEURSD53QSXRBRA5CNFSM4JIIHEU2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEDTQ2OI#issuecomment-552013113, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADT7TOQK7MX5JUWRMRAR5PTQSXRBRANCNFSM4JIIHEUQ.