[JBeilke]

Hi,

when I run the tutorial case "multiphase/settlingFoam/ras/tank3D" in seriell it crashes with following error message:

Time = 5158.4

Courant Number mean: 0.201469 max: 5395.01
diagonal:  Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
DILUPBiCG:  Solving for Ux, Initial residual = 0.978119, Final residual = 7.07031e-08, No Iterations 27
DILUPBiCG:  Solving for Uy, Initial residual = 0.75499, Final residual = 8.76912e-08, No Iterations 31
DILUPBiCG:  Solving for Uz, Initial residual = 0.961094, Final residual = 9.40318e-08, No Iterations 23
DILUPBiCG:  Solving for Alpha, Initial residual = 0.97067, Final residual = 4.07851e-08, No Iterations 52
Solid phase fraction = 0.000773407  Min(Alpha) = -0.000392519  Max(Alpha) = 0.0136277
DICPCG:  Solving for p_rgh, Initial residual = 0.950573, Final residual = 0.0087833, No Iterations 120
diagonal:  Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 8.19507e-14, global = 4.35535e-15, cumulative = 4.07452e-10
DICPCG:  Solving for p_rgh, Initial residual = 0.548982, Final residual = 7.58213e-08, No Iterations 146
diagonal:  Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 8.06092e-14, global = 4.61678e-15, cumulative = 4.07457e-10
#0  Foam::error::printStack(Foam::Ostream&) in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#1  Foam::sigFpe::sigHandler(int) in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#2  ?? in "/lib64/libc.so.6"
#3  Foam::DILUPreconditioner::calcReciprocalD(Foam::Field<double>&, Foam::lduMatrix const&) in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#4  Foam::DILUPreconditioner::DILUPreconditioner(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#5  Foam::lduMatrix::preconditioner::addasymMatrixConstructorToTable<Foam::DILUPreconditioner>::New(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#6  Foam::lduMatrix::preconditioner::New(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#7  Foam::PBiCG::solve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#8  Foam::fvMatrix<double>::solve(Foam::dictionary const&) in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/lib/libfiniteVolume.so"
#9  Foam::fvMatrix<double>::solve() in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/bin/settlingFoam"
#10  main in "/home/beilke/OpenFOAM/OpenFOAM-2.1.x/platforms/linux64GccDPOpt/bin/settlingFoam"
#11  __libc_start_main in "/lib64/libc.so.6"
#12  _start at /usr/src/packages/BUILD/glibc-2.10.1/csu/../sysdeps/x86_64/elf/start.S:116

I can make a restart from the last result and it runs fine to the end. There are also no problems in parallel.

It would be nice if anyone using the git version can try to reproduce this error.

I use Opensuse 11.2

[wyldckat]

Greetings Joern,

Only today did I finally manage to test this. I've done a clean build of the latest OpenFOAM 2.1.x on Ubuntu 11.10 x86_64 (gcc 4.6.1) and got pretty much the same error, but only several iterations before, namely on:

Time = 251.9

Courant Number mean: 0.00634015 max: 15.4453
diagonal:  Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
DILUPBiCG:  Solving for Ux, Initial residual = 0.979337, Final residual = 9.26653e-09, No Iterations 22
DILUPBiCG:  Solving for Uy, Initial residual = 0.998575, Final residual = 6.64669e-08, No Iterations 20
DILUPBiCG:  Solving for Uz, Initial residual = 0.995422, Final residual = 3.95081e-08, No Iterations 21
DILUPBiCG:  Solving for Alpha, Initial residual = 0.985569, Final residual = 1.34362e-08, No Iterations 21
Solid phase fraction = 8.19285e-05  Min(Alpha) = 3.10327e-96  Max(Alpha) = 0.00260623
DICPCG:  Solving for p_rgh, Initial residual = 0.978814, Final residual = 0.00737583, No Iterations 112
diagonal:  Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 8.06819e-14, global = 2.27841e-15, cumulative = 2.00363e-11
DICPCG:  Solving for p_rgh, Initial residual = 0.268882, Final residual = 8.87841e-08, No Iterations 136
diagonal:  Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 7.90055e-14, global = 1.86702e-15, cumulative = 2.00382e-11

Only now did I notice that it was the Courant number that blew up. Please report this on the official bug tracker, since this looks like a genuine bug: http://www.openfoam.com/mantisbt/my_view_page.php

By the way, you should also mention which gcc version you are using:

gcc --version

Best regards,
Bruno

[JBeilke]

Hi Bruno,

thanks for your test.

Did you try a restart from the last result and a parallel run?

The reason I posted this problem here is that Henry was not able to reproduce this bug on his machine and I wanted to know, if it is a real bug. And it might be also helpful to get the results from different configurations.

[wyldckat]

Hi Joern,

Originally Posted by JBeilke

Did you try a restart from the last result and a parallel run?

No, I only tested with running Allrun. I haven't done more tests on the same machine, in order to diagnosed the reason. But I've been doing more tests on virtual machines... more below.

Originally Posted by JBeilke

The reason I posted this problem here is that Henry was not able to reproduce this bug on his machine and I wanted to know, if it is a real bug. And it might be also helpful to get the results from different configurations.

OK, after reading your post, I've done a few more tests (most done with the latest 2.1.x version) and a couple more are still ongoing (all gcc versions are the default ones on each Linux distribution):

• Ubuntu 11.10 x86_64 - gcc 4.6.1:
  • On real machine with AMD 1055T - crash at Time 251.9 s.
  • On virtual machine with i3 350M - identical crash at Time 251.9 s.
• Ubuntu 10.04 x86_64 - gcc 4.4.3:
  • On virtual machine with AMD 1055T - no crash - ClockTime = 30414 s.
• Ubuntu 11.04 x86_64 - gcc 4.5.2:
  • On virtual machine with AMD 1055T - identical crash at Time 251.9 s.
• openSUSE 12.1 x86_64 - gcc 4.6.2:
  • On virtual machine with AMD 1055T - no crash - ClockTime = 33869 s.
• Fedora 15 x86_64 - gcc 4.6.1:
  • On virtual machine with AMD 1055T - identical crash at Time 251.9 s.

I'll update this post as results continue to come in.

edit 21:02 2012-01-22: Fedora 15 crashed in the exact same way+place as the others.
edit 06:45 2012-01-23: openSUSE 12.1 completed with success.

Best regards,
Bruno

[JBeilke]

ScientificLinux 6.1 x86_64 - gcc 4.4.5:
- On virtual machine with i7-2600 - no crash - ClockTime = 14413 s.

[JBeilke]

CentOS 6.2 x86_64 - gcc 4.4.6:
- On virtual machine with i7-2600 - no crash

I just copied the OF installation from OpenSuse 11.2 (where it crashed) without recompilation. So the error is very likely not a compiler problem.

[wyldckat]

Hi Joern,

Originally Posted by JBeilke

I just copied the OF installation from OpenSuse 11.2 (where it crashed) without recompilation. So the error is very likely not a compiler problem.

Mmm... then it might be an issue related to GLIBC, MPFR, GMP or MPC... I'm not certain if one or more of the latter 3 are statically linked into the GNU G++ libraries.

Let me see... OK, I think DistroWatch.com has the right version list for GLIBC, the rest I will have to dig deeper:

• GLIBC:
  • Ubuntu 11.10: 2.13
  • Ubuntu 11.04: 2.13
  • Ubuntu 10.04: 2.11.1
  • openSUSE 12.1: 2.14.1
  • Fedora 15: 2.13.90
• MPFR:
  • Ubuntu 11.10: libmpfr4 3.0.1-5
  • Ubuntu 11.04: libmpfr4 3.0.0-7
  • Ubuntu 10.04: libmpfr1ldbl 2.4.2-3ubuntu1
  • openSUSE 12.1: 3.0.1-3.1.5
  • Fedora 15: mpfr-3.0.0-4.fc15
• GMP:
  • Ubuntu 11.10: libgmp10 2:5.0.1+dfsg-7ubuntu2
  • Ubuntu 11.04: libgmp3c2 2:4.3.2+dfsg-1ubuntu3
  • Ubuntu 10.04: libgmp3c2 2:4.3.2+dfsg-1ubuntu1
  • openSUSE 12.1: 5.0.2-8.1.5
  • Fedora 15: gmp-1:4.3.2-3.fc15
• MPC:
  • Ubuntu 11.10: libmpc2 0.9-3
  • Ubuntu 11.04: libmpc2 0.9-1
  • Ubuntu 10.04: not used by GCC.
  • openSUSE 12.1: 0.8.2-12.1.6
  • Fedora 15: libmpc-0.8.3-0.3.svn855.fc15

I'll fill out the rest of the list as soon as I can collect them. But the first assumption is that GLIBC 2.13 is veeeery suspicious...
edit: yep, GLIBC 2.13 looks very suspicious indeed!

Best regards,
Bruno