Thank you for the quick response!

Hmm, actually it seems the build got stuck on 32 bits:
https://bugzilla.redhat.com/show_bug.cgi?id=1653142

Not sure what's broken there. It got stuck in the %check where it's trying to run the benchmark.

(In reply to Jaroslav Škarvada from comment #2)
> Hmm, actually it seems the build got stuck on 32 bits:
> https://bugzilla.redhat.com/show_bug.cgi?id=1653142
> 
> Not sure what's broken there. It got stuck in the %check where it's trying
> to run the benchmark.

https://koji.fedoraproject.org/koji/taskinfo?taskID=31129461

The same problem encountered when recompiling hpl-2.1, so probably not HPL problem, but regression somewhere else. It seems the benchmark is much much slower on the 32 bit. I do not have time to debug, so I will probably just disable the %check.

It seems to work with f29, but not with f30.

f29 has:
openmpi-devel-2.1.1-14.fc29

f30 has:
openmpi-devel-2.1.5-1.fc30

I am not sure whether it's openmpi related, but I will check.

Yeah, I'm seeing some similar hangs with hdf5 tests.