Alternatively, might this be the PETSc memory leak? On Mon, 30 May 2016 at 15:18 Mitchell, Lawrence < lawrence.mitchell@imperial.ac.uk> wrote:

...

On 30 May 2016, at 13:20, Fryderyk Wilczynski <scfw@leeds.ac.uk> wrote:

Dear Firedrakes,

My computations get killed with the following exception:

COFFEE finished in 0.0201869 seconds (flops: 92 -> 92) COFFEE finished in 0.0305271 seconds (flops: 92 -> 92) COFFEE finished in 0.0411251 seconds (flops: 92 -> 92) [1] pyop2:INFO Solving nonlinear variational problem... [2] pyop2:INFO Solving nonlinear variational problem...

...

...

[1]PETSC ERROR: to get more information on the crash.

--------------------------------------------------------------------------

...

mpirun noticed that process rank 3 with PID 6682 on node comp-pc3234 exited on signal 9 (Killed).

Probably this happens because of a segfault somewhere in the code. Are you using Expressions in your code that maybe are not safe (overwriting memory somewhere). Does this happen in serial? It is hard to know more without knowing exactly what you are doing.

Lawrence