Hi Gilles,
Thanks, and yes, you are right:
ompi_info --all | grep "MCA io" | grep priority
MCA io: parameter "io_romio_priority" (current value: "20",
data source: default, level: 9 dev/all, type: int)
MCA io: parameter "io_romio_delete_priority" (current value:
"20", data source: default, level: 9 dev/all, type: int)
MCA io: parameter "io_ompio_priority" (current value: "10",
data source: default, level: 9 dev/all, type: int)
MCA io: parameter "io_ompio_delete_priority" (current value:
"10", data source: default, level: 9 dev/all, type: int)
So it seems we are indeed using ROMIO. Any suggestions what that means with
respect to our file coherence issue?
Regards,
Michael
On 23 Jul 2015, at 14:07 , Gilles Gouaillardet
<gilles.gouaillar...@gmail.com<mailto:gilles.gouaillar...@gmail.com>> wrote:
Michael,
ROMIO is the default in the 1.8 series
you can run
ompi_info --all | grep io | grep priority
ROMIO priority should be 20 and ompio priority should be 10.
Cheers,
Gilles
On Thursday, July 23, 2015, Schlottke-Lakemper, Michael
<m.schlottke-lakem...@aia.rwth-aachen.de<mailto:m.schlottke-lakem...@aia.rwth-aachen.de>>
wrote:
Hi Gilles,
> are you running 1.8.7 or master ?
1.8.7. We recently upgraded our cluster installation from OpenSUSE 11.3/OpenMPI
1.6.5 to OpenSUSE 12.3/OpenMPI 1.8.7. Before the upgrade, we did not encounter
this problem.
> if not default, which io module are you running ?
> (default is ROMIO with 1.8 but ompio with master)
We did not specify anything at configure time, so I guess we are using the
default. But if you tell me how, I can check.
> by any chance, could you post a simple program that evidences this issue ?
As of this time, unfortunately no. We only experience this issue
intermittently, and only when running our suite of regression tests. It *seems*
to occur only with a handful of the ~40 tests, but if we run only a subset of
the tests (instead of all of them), it may not occur at all, depending on the
subset. I tried using a MWE program but could not reproduce the issue with it.
Sorry for not being more helpful, but we are also scratching our heads trying
to understand what is going on and I just thought that maybe someone here has
had a similar experience in the past (or might give us some pointers at what to
look at).
Regards,
Michael
