Hi,
I’d like to re-establish our offscreen render capabilities in our cluster. Therefore, I’ve installed paraview 5.5.2 using the following options
configopts = '-DPARAVIEW_INSTALL_DEVELOPMENT_FILES=ON -DVTK_OPENGL_HAS_OSMESA=ON -DPARAVIEW_USE_MPI=ON ’
configopts += '-DOPENGL_INCLUDE_DIR=/usr/include -DOPENGL_gl_LIBRARY=/usr/lib64/libGL.%s ’ % SHLIB_EXT
configopts += '-DOSMESA_INCLUDE_DIR=/usr/include -DOSMESA_LIBRARY=/usr/lib64/libOSMesa.%s ’ % SHLIB_EXT
configopts += '-DOPENGL_glu_LIBRARY=/usr/lib64/libGLU.%s ’ % SHLIB_EXT
configopts += '-DVTK_USE_SYSTEM_HDF5=ON -DPARAVIEW_ENABLE_PYTHON=ON -DBUILD_SHARED_LIBS=ON ’
configopts += '-DBUILD_TESTING=OFF ’
configopts += '-DBUILD_TYPE=Debug ’
configopts += ’ -DPARAVIEW_QT_VERSION=5 -DQT_QMAKE_EXECUTABLE=/usr/bin/qmake-qt5 ’
on Centos 7.3 machine.
The compute nodes don’t have gpu and report glxinfo | grep OpenGL
OpenGL renderer string: Gallium 0.4 on llvmpipe (LLVM 3.8, 256bit)
OpenGL version string: 2.1 Mesa 11.2.2
OpenGL shading language version string: 1.30
…
I can start the pvserver (–force-offscreen-rendering) and connect to the client. I load a data set and as soon as I change representation from outline to any other option I receive a disconnect showing
GL version 2.1 with the gpu_shader4 extension is not supported by your graphics driver…
Does that mean that I can’t run paraview on those nodes? Or would a Mesa driver newer than 11.2.2 help?
Any hint is welcome.
Best regards,
Andreas