I’ve tried to install mfix-24.3.1 on the cluster using conda but I can not seem to get to gui to work. After activating mfix and running the medic command to launch the gui I receive the following error.
libGL error: MESA-LOADER: failed to open swrast: /usr/lib64/dri/swrast_dri.so: cannot open shared object file: No such file or directory (search paths /usr/lib64/dri, suffix _dri)
libGL error: failed to load driver: swrast
The issue is that when I try and install Mesa-libgl using conda install I run into this problem.
The credentials you have provided for this URL are invalid.
You will need to modify your conda configuration to proceed.
Use conda config --show to view your configuration’s current state.
Further configuration help can be found at https://conda.io/docs/config.html.
I have tried modifying the ~/.condarc file to solve this issue. I saw recommendations to set the ssl_verify to true, but that did not work either. Any suggestions?
Thanks for your response. I have tried launching with vglrun command but I run into the same error message. The mfix gui window launches but it gets stuck on the loading Mfix page. The true interface never loads and I get the libGL error.
Hmm interesting thought both glxgears work both in the base environment and the mfix environment. Vglrun glxgears runs in the base environment but not in the mfix(-24.3.1) environment
Very interesting. I hope you can get some help from the local sysadmins, I can’t really diagnose this from here. But I can offer some tips.
I don’t think there’s a glxgears or vglrun in the MFiX distribution (which glxgears shows /usr/bin/glxgears with mfix-24.3.1 activated).
One thing you could try is, in both enviroments, do ldd glxgears to see if they are pulling in different libraries (since conda does not set LD_LIBRARY_PATH this should not be the case).
Or try strace vglrun glxgears in both envs and see if you can spot the difference.
I just wanted to provide an update for anyone in the future. In this instance the best method was build from the source code. I got the tarball from the MFiX website, it requires login credentials but here is the link
From there I was able to load the required dependency; GNU Make, Fortran compiler, CMake and OpenMPI. After that I just needed to compile and build MFiX. These are the websites I used for future reference. The first provides the linux dependencies the second is the compile and build instructions.
It is worth noting that with this method I use the run from the command line which works well given that I am on the cluster. I have found that having mfix installed using the conda method with the GUI is nice on my personal computer for setting up the simulations. I can then copy the directory to the cluster and run from there. Hope this helps anyone else struggling with the same issue and thanks for the support.