I am running a simulation using hdf5 output in parallel using mpirun, built with CXX=g+±8.
Everything compiles and runs fine right until the end, where I get this error:
HDF5-DIAG: Error detected in HDF5 (1.10.3) MPI-process 1:
#000: H5E.c line 1619 in H5Eget_auto2(): wrong API function, H5Eset_auto1 has been called
major: Error API
minor: Can’t get value
[myproc1] *** Process received signal ***
[myproc1] Signal: Abort trap: 6 (6)
[myproc1] Signal code: (0)
[myproc1] [ 0] 0 libsystem_platform.dylib 0x00007fff60037f5a _sigtramp + 26
[myproc1] [ 1] 0 ??? 0x0000000000000000 0x0 + 0
[myproc1] [ 2] 0 libsystem_c.dylib 0x00007fff5fdd51ae abort + 127
… with further stack trace.
The error message is output once for each processor i.e. twice for -np 2.
Does anybody know whether this is a bug I can fix, or whether it is a problem with hdf5? Please let me know if you need any more info.