Hi Lawrence, I'm running Ubuntu 17.10, and I get: gcc --version gcc (Ubuntu 7.2.0-8ubuntu3.2) 7.2.0 mpicc -show gcc -I/usr/lib/x86_64-linux-gnu/openmpi/include/openmpi -I/usr/lib/x86_64-linux-gnu/openmpi/include/openmpi/opal/mca/event/libevent2022/libevent -I/usr/lib/x86_64-linux-gnu/openmpi/include/openmpi/opal/mca/event/libevent2022/libevent/include -I/usr/lib/x86_64-linux-gnu/openmpi/include -pthread -L/usr//lib -L/usr/lib/x86_64-linux-gnu/openmpi/lib -lmpi OpenMPI output is in attached .txt file. Thanks, Owen ________________________________________ From: firedrake-bounces@imperial.ac.uk [firedrake-bounces@imperial.ac.uk] on behalf of Lawrence Mitchell [lawrence.mitchell@imperial.ac.uk] Sent: 19 March 2018 17:06 To: firedrake@imperial.ac.uk Subject: Re: [firedrake] Firedrake test script not advancing Dear Owen,

On 19 Mar 2018, at 17:01, Owen Pembery <O.R.Pembery@bath.ac.uk> wrote:

Dear Firedrakers,

I've just installed Firedrake on an Ubuntu virtual machine, and I'm finding that the firedrake test script seemingly stops advancing (sometimes at different points in the tests).

By `stops advancing' I mean `stays on the same test for at least an hour'.

I've attached the console output for a couple of attempts running the test script.

Is there something I'm doing wrong, is this a bug, or is it an issue with the virtual machine? (Or something else?)

Looking at the output, it looks like lots of parallel tests are causing segfaults. Perhaps eventually the test suite is giving up. Can you provide a little more detail on which ubuntu version? We test on ubuntu and haven't seen any such failures, but perhaps our ubuntu is too old. In particular, I'd be interested in knowing about: gcc --version mpicc -show Assuming you have OpenMPI then also: ompi_info Cheers, Lawrence _______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake

Hi Lawrence,

Ugh, I think this might be this bug:

https://github.com/firedrakeproject/firedrake/issues/1153

Seemingly no change has occurred since then.

Can you try any of the workarounds described there? We can provide more details on how to do this.

I'll try the workarounds there, and ask for help if I get stuck. Thanks, Owen ________________________________________ From: firedrake-bounces@imperial.ac.uk [firedrake-bounces@imperial.ac.uk] on behalf of Lawrence Mitchell [lawrence.mitchell@imperial.ac.uk] Sent: 19 March 2018 17:30 To: firedrake@imperial.ac.uk Subject: Re: [firedrake] Firedrake test script not advancing Hi Owen,

On 19 Mar 2018, at 17:19, Owen Pembery <O.R.Pembery@bath.ac.uk> wrote:

Hi Lawrence,

I'm running Ubuntu 17.10, and I get:

gcc --version gcc (Ubuntu 7.2.0-8ubuntu3.2) 7.2.0

mpicc -show gcc -I/usr/lib/x86_64-linux-gnu/openmpi/include/openmpi -I/usr/lib/x86_64-linux-gnu/openmpi/include/openmpi/opal/mca/event/libevent2022/libevent -I/usr/lib/x86_64-linux-gnu/openmpi/include/openmpi/opal/mca/event/libevent2022/libevent/include -I/usr/lib/x86_64-linux-gnu/openmpi/include -pthread -L/usr//lib -L/usr/lib/x86_64-linux-gnu/openmpi/lib -lmpi

OpenMPI output is in attached .txt file.

Ugh, I think this might be this bug: https://github.com/firedrakeproject/firedrake/issues/1153 Seemingly no change has occurred since then. Can you try any of the workarounds described there? We can provide more details on how to do this. Everyone: should we just ditch OpenMPI? Cheers, Lawrence _______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake

On 19 Mar 2018, at 17:52, Ham, David A <david.ham@imperial.ac.uk> wrote:

I’m in favour of dumping OpenMPI. It also causes us parallel test issues cf https://github.com/firedrakeproject/firedrake/issues/1199

Are there any downsides with going to mpich?

All the ubuntu packages that link against MPI are built against OpenMPI. We need to be very careful that when we install on such systems that we don't accidentally end up getting the wrong one. Lawrence