FLUDAG CORE DUMP ISSUE

[David Breitenmoser]

Dear Community

Currently, I'm trying to setup a Fludag simulation. For that purpose, I'm running some simple test simulations. Now, I've encountered an error during the fludag simulation, which I was not able to resolve so far. 

Worflow:

1) I created and processed the same geometry as described in https://svalinn.github.io/DAGMC/usersguide/uw2.html in cubit. The corresponding cub5 and sat files are attached to this post [test.cub5,test.sat].

2) Exporting the geometry to h5m format using the following command:

export dagmc "test.h5m" faceting_tolerance 1e-3 make_watertight verbose  [test.h5m]

3) creating the material snippet using the following command:
mainfludag --dagmc 'test.h5m'. The corresponding mat.inp file is also attached to this post [mat.inp].

4) insert mat.inp in fluka input file test_ed.inp [test_ed.inp]

5) link the test.h5m with dagmc.h5m file: ln -s test.h5m dagmc.h5m

6) execute fludag:
version1: rfluka -N0 -M1 -e /path-to-DAGMC/mainfludag test_ed.inp

version2: rfluka -N0 -M1 -e /path-to-DAGMC/mainfludag -d test_ed.inp

In both cases, I've got an error. I've attached a screenshot from the command output [fludag_log.png]. In addition, I attach the fluka log files and output files (output was only created for version2) [version1.log,version2.log,version2.out].

Any input from your side would be highly appreciated

Best,

David

[Andrew Davis]

Hi David

It needed a few tweaks at my end, but seems to run fine for me.

1) FLUKA complained that your material numbers were not reals, so I put a '.' at the end so 1 -> 1.

2) You had two superfluous END statements in the GEOBEGIN/GEOEND section

Other than that it was fine, attached is the working file. Im using FLUKA-CERN at this end. So for me

ln -s test.h5m dagmc.h5m

/home/adavis/fluka-cern/fluka-4.0.1/bin/rfluka -e /home/adavis/opt/dagmc/bin/mainfludag test_ed.inp

Then I get a successful run. Let me know if that works for you.

Thanks

Andy

--
You received this message because you are subscribed to the Google Groups "DAGMC Users and Collaborators" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dagmc-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/a0dc7871-fb26-4576-ac06-22b3d9706773n%40googlegroups.com.

[David Breitenmoser]

Hi Andy

Thank you so much for your support. I really appreciate that. I've run the two versions of commands described in the previous code with your modified input file. Unfortunately, I still get the same error, i.e. "buffer overflow detected" in the command output as well as "Aborted (core dumped)" in the fluka log file for the version without "-d". If I run the command with "-d", I get the error "At line 257 of file comlat/flkcgi.f

Fortran runtime error: Bad value during integer read" in the fluka log file (see also log file of the previous post). Could it be an issue with my fortran version? Or do I miss maybe some fortran library on my WSL? For your note: I have gfortran 7.5.0. installed on my ubuntu system (see also attached screenshot).

Thanks for your help

Best,

David

[Andrew Davis]

Hi David

Can you confirm that you're using FLUKA-CERN or FLUKA-INFN? Does a normal FLUKA problem run?

Thanks

Andy

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/5d527a36-1789-4ace-b100-31086fb014ebn%40googlegroups.com.

[David Breitenmoser]

Hi Andy

Sure, I'm using FLUKA-Cern and have also installed the corresponding branch from your github repo ($ git clone https://github.com/makeclean/DAGMC.git; $ cd DAGMC; $ git checkout fluka-cern). I've run a simple test case with plain fluka and it runs through smoothly (see attached screenshot fluka_plain.PNG, I've used plain fluka already a lot on my system and I had no issues at all). In addition, I've run again the  fludag_unit_tests as suggested on the dagmc website (https://svalinn.github.io/DAGMC/install/fludag.html). You can find the command output of this test in the attached txt. file. In my opinion, the test looks good. So I have no clue what's actually the problem.

Thank you so much for your help. 

Best

David

[Andrew Davis]

Hi David

In the case of the case with buffer overflow, please cd in the FLUKA_XXXX directory that corresponds to it. You can then run gdb with the executable and the core dump (if one was produced) and then you'd be able to see a line number. Im happy to have a call with you to help debug, but Im drowning in stuff today, next week I can probably squeeze you in.

Thanks

Andy

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/3a83ea08-ac83-43c4-9fb8-2d92a7ff063bn%40googlegroups.com.

[David Breitenmoser]

Hi Andy

That would be great. I'm flexible in time (except Tuesday). Let me know, when you have time next week for a call.

Thank you so much for your support

Best,

David

[David Breitenmoser]

Hi Andy

Short update:  I've found a way how Fludag runs properly on my system. It's quite simple, if I add "-DCMAKE_BUILD_TYPE=Debug" to my cmake command during the building of fludag, i.e. 

$ cmake ../src -DMOAB_DIR=/home/bredavid/dagmc_bld/MOAB \ -DBUILD_FLUKA=ON \ -DFLUKA_DIR=$FLUPRO \               -DCMAKE_INSTALL_PREFIX=$INSTALL_PATH \  -DCMAKE_BUILD_TYPE=Debug

, the resulting mainfludag executable is working perfectly. I've added also my own executable usreou.f to the CMakeList.txt in src/fluka/app as described by you in a previous post. The resulting scoring files look fine. So that's very good news for me. There is however a but.

I can't explain, why it's working with the additional CMAKE variable and not working without it. Do you have any idea? I'm afraid that my solution is not stable in the sense that it might not work for other cases.

Have a nice weekend

Best,

David

[Andrew Davis]

Hi David

Could be a whole range of things, I'll try and add more unit tests and see if that picks anything up

Thanks

Andy

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/093ad296-c18a-4383-9139-be437e22660dn%40googlegroups.com.

[David Breitenmoser]

Hi Andy

Perfect, please let me know, if you need something or if you want to meet in a video call.

Best,

David

[Nikos Nikolopoulos]

Dear Experts,

first of all I should say that I am using FLUKA-CERN and that I have successfully installed fludag from the cern branch by makeclean.

Having said that, I followed a similar workflow with the one described by David, in particular:

1) I created the example.sat file as described  in https://svalinn.github.io/DAGMC/usersguide/uw2.html under the "Worked example" section and executed the command:  export dagmc "example.h5m" faceting_tolerance 1e-3 make_watertight verbose in the cubit command line, which produced the goemetry.h5m [example.sat].

2) Moving on I produced the mat.inp file after exectuing the command: ./mainfludag --dagmc /mnt/c/users/nnikolop/downloads/example.h5m from the associated directory (in my case: /home/nikos/dagmc_bld/DAGMC/bin) [example.h5m].

3) I created a FLUKA input file named example.inp to include mat.inp [example.inp].

4) After that I linked the example.h5m file with dagmc.h5m using the command:  ln -s example.h5m dagmc.h5m [dagmc.h5m].

5) Finally I execute the mainfludag like so: /usr/local/fluka/bin/rfluka -e /home/nikos/dagmc_bld/DAGMC/bin/mainfludag example.inp.

I get an error of similar nature to the one described by David (screenshot attached), I am also attaching a zip folder containing the aforementioned filenames enclosed in brackets. Note that I tested FLUKA and it works fine with another .inp file I prepared. 

Apologies for the long message, any ideas on how to overcome this would be gratefully appreciated! 

Best regards,

Nikos. 

[Andrew Davis]

Hi Nikos

Ok, this runs at my end, but you're not using appropriate FLUKA names i.e. Lead -> LEAD. Could you please attach the fluka log and err files for your failed run?

Thanks

Andy

--

You received this message because you are subscribed to the Google Groups "DAGMC Users and Collaborators" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dagmc-users...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/95959a5c-1b2e-44d6-87c2-ff256dda9823n%40googlegroups.com.

[Nikos Nikolopoulos]

Dear Andy,

thank you for your repIy, I modified the .inp and capitalized the "LEAD" material. I am attaching below a screenshot of the new error message I get along with the zip folder produced by FLUKA after the run (it seems that it doesn't even contain .log and .err files which is strange).

Best regards,

Nikos.

[Nikos Nikolopoulos]

Dear Andy,

I am reattaching the file produced by FLUKA, after repeating the same process it looks like I get a similar error with one described by David. In essence "buffer overflow detected" in the command output as well as "Aborted (core dumped)" in the FLUKA log file. I guess the same fix for David's problem should apply in my case as well.

Kind regards,

Nikos.

[Andrew Davis]

Hi Nikos

I think we're going to have to have a call about this, I can't reproduce your error locally, so I think it must be something to do with your setup. Given that you're using WSL, can you do the following

wsl --export Ubuntu-18.04 ubuntu.tar

This will save your ubuntu instance to a file called ubuntu.tar, feel free to tar.gz it too. Then upload it some where for me to download.

Thanks

Andy

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/2d8e2d10-5c30-40e3-9ec8-e44a286e13b8n%40googlegroups.com.

[Nikos Nikolopoulos]

Hello Andy,

thankfully David's recipe to repeat the DAGMC installation with the extra -DCMAKE_BUILD_TYPE=Debug line worked for me as well. FLUKA now runs properly and the along with the FluDAG executable which is great!

I am now curious if I could visualize the converted CAD geometry (perhaps a modification of the .h5m file) using flair, but I should create a different topic for this issue.

Thank you for your support,

Nikos.