Error running new spheral build

[rd-contr]

I'm trying to run tests/functional/Hydro/Sedov/Sedov-cylindrical-2d.py in a new build of spheral on our cluster. The "smoke test" works without issue.

[xxxxxx@host ~]$ mpirun -np 2 /opt/ohpc/pub/apps/spheral/latest/bin/spheral -c "import Spheral"
/------------------------------------------------------------------------------
| Spheral version : v2025.06.1 74361a1 develop |
| number of MPI tasks : 2 |
| number of threads per rank: 8 |
------------------------------------------------------------------------------/

But when I run the test case, the code returns a memory issue that effectively looks like it is attempting to "malloc(null)". I ran the case through valgrind trying to get some additional information. That output is attached. I acknowledge that it is entirely possible that I am running the code incorrectly. Any feedback you can provide would be greatly appreciated.

spheral_valgrind_output.txt

[jmikeowen]

That test is slightly out of date, but if you remove line 429 ("output(HEvolution)") it should work. I don't think this one out of date syntax error should resuilt in a memory error, but I'd also recommend running this serially first if you're having problems.

You could also start with one of the tests known to be working such as the Noh-planar-1d.py script.

[rd-contr]

I tried running the Noh-planar-1d.py script in serial mode but end up with effectively the same behavior (a segfault due to a memory allocation error).

[xxxxxx@host Noh]$ /opt/ohpc/pub/apps/spheral/latest/bin/spheral /workdir/xxxxxx/spheral/Spheral/tests/functional/Hydro/Noh/Noh-planar-1d.py
/------------------------------------------------------------------------------
| Spheral version : v2025.06.1 74361a1 develop |
| number of MPI tasks : 1 |
| number of threads per rank: 8 |
------------------------------------------------------------------------------/
---------------- 1-D integrated hydro test -- planar Noh problem ----------------
WT : <SpheralKernel.TableKernel1d object at 0x1467db2556b0>
nodes1 : <SpheralNodeList.FluidNodeList1d object at 0x1467db2524a0>
nodes1.hmin : 0.0001
nodes1.hmax : 0.1
nodes1.nodesPerSmoothingScale : 1.35
distributeNodesGeneric: working on nodes1, (local, global) number nodes 100 100
distributeNodesGeneric: performing initial crappy distribution.
distributeNodesGeneric: nodes1 initially finished
(min, max, avg) nodes per process initially: 100 100 100.0
Total number of nodes: 100
distributeNodesGeneric: calling for redistribution.
distributeNodesGeneric: redistribution done.
nodes1.numNodes : 100
db : <SpheralDataBase.DataBase1d object at 0x1467db2524f0>
db.appendNodeList(nodes1) : None
db.numNodeLists : 1
[host:212873:0:212873] Caught signal 11 (Segmentation fault: invalid permissions for mapped object at address 0x7ffd7b3200a0)
==== backtrace (tid: 212873) ====
0 /opt/ohpc/pub/mpi/ucx-ohpc/1.15.0/lib/libucs.so.0(ucs_handle_error+0x294) [0x146810b98224]
1 /opt/ohpc/pub/mpi/ucx-ohpc/1.15.0/lib/libucs.so.0(+0x2d3dc) [0x146810b983dc]
2 /opt/ohpc/pub/mpi/ucx-ohpc/1.15.0/lib/libucs.so.0(+0x2d688) [0x146810b98688]
3 /lib64/libpthread.so.0(+0x12990) [0x146828ddc990]
4 [0x7ffd7b3200a0]
=================================
/opt/ohpc/pub/apps/spheral/latest/bin/spheral: line 4: 212873 Segmentation fault (core dumped) /opt/ohpc/pub/apps/spheral/latest/.venv/bin/python -B "$@"

[jmikeowen]

That trace looks to me like you're failing in threading for some reason. Try building (from scratch) without threading enabled: configure with -DENABLE_OPENMP=OFF.