Hello everyone,
I recently switched from using Debian stable to Ubuntu 14.10 on my development laptop, and I'm now having trouble getting the Einstein Toolkit to run. It compiles fine, but even the simplest simulations segfault consistently. Does anyone have any advice on how to approach debugging this? I've attached the error output from my attempt to run the example static tov star in the simplified tutorial for new users (https://docs.einsteintoolkit.org/et-docs/Simplified_Tutorial_for_New_Users) as well as the two sets of build options that I've tried using (the first, ubuntu.cfg, is the default, the second, yugoth.cfg, is a set of options that I've used successfully in the past).
I know your time is precious and I appreciate any that you can spare.
Thanks in advance!
Best, Jonah Miller
On Sat, Jan 24, 2015 at 04:41:09PM -0500, Jonah Miller wrote:
I know your time is precious and I appreciate any that you can spare.
Without more information I would guess that a broken MPI installation is to blame. You don't by any chance use another mpirun than the one that comes with the mpich library you link against? (I know - but that also happened to me once). Other than that - did you try running on just one thread instead of 2?
Frank
Hi Frank,
Thanks for your reply.
On 15-01-24 11:02 PM, Frank Loeffler wrote:
You don't by any chance use another mpirun than the one that comes with the mpich library you link against? (I know - but that also happened to me once).
Actually that turned out to be exactly the issue. Apparently the Anaconda Python (which is disconnected from the package manager) installs its own version of MPI for Python's mpi module. I didn't know this and I stupidly added the Anaconda directory to my search path and it overrode the system files.
Thanks again for your help!
Best, Jonah
users@lists.einsteintoolkit.org