[jmf]

Hello everybody

I can't get snappyHexMesh working in parallel with 2 processors, whereas the solvers do work fine.

I use 1.5-dev, compiled from sourceforge package. I have GCC4.4. I also tried GCC4.3.4 with the same results

The failure occurs during phase 2. Here is one of the error message :

Smoothing patch points ...
Smoothing iteration 0
Found 2331 non-mainfold point(s).

Scaling iteration 0
Moving mesh using diplacement scaling : min:1  max:1
Checking faces in error :
    non-orthogonality >  80 degrees                        : 279
    faces with face pyramid volume < 1e-13                 : 63
    faces with concavity >  90 degrees                     : 7
    faces with skewness >   4 (internal) or  20 (boundary) : 0
    faces with interpolation weights (0..1)  <  0.05       : 9
    faces with volume ratio of neighbour cells <  0.01     : 0
    faces with face twist <  0.05                          : 186
    faces on cells with determinant < 0.001                : 3
Added an additional 1 faces on coupled patches. (processorPolyPatch, cyclicPolyPatch)
Added an additional 52 faces on coupled patches. (processorPolyPatch, cyclicPolyPatch)
[linux0:17942] *** Process received signal ***
[linux0:17942] Signal: Segmentation fault (11)
[linux0:17942] Signal code:  (-6)
[linux0:17942] Failing at address: 0x3e800004616
[linux0:17942] [ 0] /lib64/libc.so.6 [0x7f00afacd560]
[linux0:17942] [ 1] /lib64/libc.so.6(gsignal+0x35) [0x7f00afacd4e5]
[linux0:17942] [ 2] /lib64/libc.so.6 [0x7f00afacd560]
[linux0:17942] [ 3] /lib64/libc.so.6 [0x7f00afb105eb]
[linux0:17942] [ 4] /lib64/libc.so.6(__libc_malloc+0x79) [0x7f00afb12a59]
[linux0:17942] [ 5] /usr/lib64/libstdc++.so.6(_Znwm+0x1d) [0x7f00b032774d]
[linux0:17942] [ 6] /usr/lib64/libstdc++.so.6(_Znam+0x9) [0x7f00b0327869]
[linux0:17942] [ 7] /home/nous/OpenFOAM/OpenFOAM-1.5-dev/lib/linux64GccDPOpt/libdynamicMesh.so(_ZN4Foam4ListIdEC1EiRKd+0x68) [0x7f00b12c83a8]
[linux0:17942] [ 8] /home/nous/OpenFOAM/OpenFOAM-1.5-dev/lib/linux64GccDPOpt/libdynamicMesh.so(_ZN4Foam9syncTools13syncPointListIdNS_8plusEqOpIdEEEEvRKNS_8polyMeshERNS_5UListIT_EERKT0_RKS8_b+0x1b7) [0x7f00b12cd5e7]
[linux0:17942] [ 9] /home/nous/OpenFOAM/OpenFOAM-1.5-dev/lib/linux64GccDPOpt/libdynamicMesh.so [0x7f00b12bfe76]
[linux0:17942] [10] /home/nous/OpenFOAM/OpenFOAM-1.5-dev/lib/linux64GccDPOpt/libdynamicMesh.so(_ZNK4Foam14motionSmoother9minSmoothERKNS_10PackedListILi1EEERKNS_4ListIiEERKNS_14GeometricFieldIdNS_15pointPatchFieldENS_9pointMeshEEERSC_+0x7f) [0x7f00b12c02ef]
[linux0:17942] [11] /home/nous/OpenFOAM/OpenFOAM-1.5-dev/lib/linux64GccDPOpt/libdynamicMesh.so(_ZN4Foam14motionSmoother9scaleMeshERNS_4ListIiEERKNS1_INS_4PairIiEEEEbi+0x1c12) [0x7f00b12c4372]
[linux0:17942] [12] /home/nous/OpenFOAM/OpenFOAM-1.5-dev/lib/linux64GccDPOpt/libautoMesh.so(_ZNK4Foam14autoSnapDriver14preSmoothPatchERKNS_14snapParametersEiRKNS_4ListINS_4PairIiEEEERNS_14motionSmootherE+0x167) [0x7f00b0f0c6f7]
[linux0:17942] [13] /home/nous/OpenFOAM/OpenFOAM-1.5-dev/lib/linux64GccDPOpt/libautoMesh.so(_ZN4Foam14autoSnapDriver6doSnapERKNS_10dictionaryES3_RKNS_14snapParametersE+0x463) [0x7f00b0f0fbb3]
[linux0:17942] [14] snappyHexMesh [0x405f04]
[linux0:17942] [15] /lib64/libc.so.6(__libc_start_main+0xfd) [0x7f00afab9a7d]
[linux0:17942] [16] snappyHexMesh [0x404169]
[linux0:17942] *** End of error message ***
mpirun noticed that job rank 0 with PID 17941 on node linux0 exited on signal 15 (Terminated).

Sometimes, the message is slightly different :

Added an additional 147 faces on coupled patches. (processorPolyPatch, cyclicPolyPatch)
*** glibc detected *** snappyHexMesh: corrupted double-linked list: 0x000000000234ab40 ***
[linux0:06075] *** Process received signal ***
[linux0:06075] Signal: Segmentation fault (11)
[linux0:06075] Signal code:  (-6)
[linux0:06075] Failing at address: 0x3e8000017bb
[linux0:06075] [ 0] /lib64/libc.so.6 [0x7f770ab3a560]

Thanks in advance for your help

Wishes

J-Michel

[jmf]

Hello everybody

I've solved it with a small modification in OpenFOAM-1.5-dev/etc :

commsType blocking; //instead of nonBlocking

I don't know the signification of it , but it just works

Thanks to the authors of all the topics in relation with MPI, where I picked the idea

Wishes

J-Michel