I digged a bit the configury logic and found
./configure --prefix=/usr/local/ --with-cuda --with-slurm
--with-pmi=/usr/local/slurm
should do the trick, if not
./configure --prefix=/usr/local/ --with-cuda --with-slurm
--with-pmi=/usr/local/slurm --with-pmi-libdir=/usr/local/slurm/lib64
There is a linux utility program `locate` that may be installed on
your system. You could try
$ locate ibv_devinfo
Thus, mine returns
$ locate ibv_devinfo
/usr/bin/ibv_devinfo
/usr/share/man/man1/ibv_devinfo.1.gz
That should find it if it is on local disk and not in a network
filesystem, and i
In our config the "--with-pmi" points to the slurm “prefix” dir not the slurm
libdir. The options below work for us with SLURM installed in “/opt/slurm”.
I’ll note that after sharing this config with regard to another issue, it was
recommended to drop the “/usr” in the “—with-foo=/usr” options
Dear John,
I see, thank you for your reply. Unfortunately the cluster support is of poor
quality, and it would take a while to get this information from them. Is there
any way in which I can check this by myself? Also, it looks like ibv_devinfo
does not exist on the cluster
$ ibv_devinfo
-bash:
On Oct 9, 2018, at 8:55 PM, Noam Bernstein wrote:
>
>> That's basically what the output of
>> ompi_info -a
>> says.
You actually probably want:
ompi_info | grep btl
That will show you the names and versions of the "btl" plugins that are
available on your system. For example, this is what I
find /usr/local/slurm/ -name pmi.h
/usr/local/slurm/include/slurm/pmi.h
From: users On Behalf Of Bennet Fauber
Sent: Wednesday, October 10, 2018 2:19 PM
To: users@lists.open-mpi.org
Subject: Re: [OMPI users] issue compiling openmpi 3.2.1 with pmi and slurm
Yes, I was just wondering about the pa
Yes, I was just wondering about the path as specified,
--with-pmi=/usr/local/slurm/include/slurm
What do you get from
$ find /usr/local/slurm -name pmi.h
On Wed, Oct 10, 2018 at 2:07 PM Ross, Daniel B. via users <
users@lists.open-mpi.org> wrote:
> This is a slurm build and you point it
This is a slurm build and you point it to the slurm pmi directories.
From: users on behalf of Bennet Fauber
Sent: Wednesday, October 10, 2018 1:14 PM
To: users@lists.open-mpi.org
Subject: Re: [OMPI users] issue compiling openmpi 3.2.1 with pmi and slurm
I thou
I thought the --with-pmi=/the/dir was meant to point to the top of a
traditional FHS installation of PMI; e.g., /opt/pmi with
subdirectories for bin, lib, include, man, etc. It looks like this is
pointing only to the header file, based on the name.
On Wed, Oct 10, 2018 at 11:27 AM Ralph H Castai
Yeah, the CPPFLAGS didn’t get set - I suspect because the location being
pointed to starts with “/usr/local” and therefore looks like a standard one.
What you might do is set CPPFLAGS yourself to be
“/usr/local/slurm/include/slurm” before running configure - should fix the
problem. Might also n
Here is the best I can do. The config.log is too large to email.
grep pmi config.log
$ ./configure --prefix=/usr/local/ --with-cuda --with-slurm
--with-pmi=/usr/local/slurm/include/slurm --with
pmi-libdir=/usr/local/slurm/lib64
configure:71744: result: '--prefix=/usr/local/' '--with-cuda' '--
> On Oct 10, 2018, at 4:51 AM, Dave Love wrote:
>
> RDMA was just broken in the last-but-one(?) RHEL7 kernel release, in
> case that's the problem. (Fixed in 3.10.0-862.14.4.)
I strongly suspect that this is it. In the process of getting everything
organized to collect the info various people
It appears that the CPPFLAGS isn’t getting set correctly as the component
didn’t find the Slurm PMI-1 header file. Perhaps it would help if we saw the
config.log output so we can see where OMPI thought the file was located.
> On Oct 10, 2018, at 6:44 AM, Ross, Daniel B. via users
> wrote:
>
I have been able to configure without issue using the following options:
./configure --prefix=/usr/local/ --with-cuda --with-slurm
--with-pmi=/usr/local/slurm/include/slurm
--with-pmi-libdir=/usr/local/slurm/lib64
Everything compiles just fine until I get this error:
make[3]: Leaving directory
Well, good question. To be fair, the test passes if you run it with a lower
number of processes. In addition, I had a couple of years back a discussion on
that with one of the HDF5 developers, and it seemed to be ok to run it this way.
That being said, after thinking about it a bit, I think the
On that system please tell us what these return:
ibstat
ibstatus
sminfo
ibdiagnet
On Wed, 10 Oct 2018 at 12:49, John Hearns wrote:
>
> Noam, what does ompi_info say - specifically which BTLs are available?
> Stupid question though - this is a single system with no connection to a
> switch?
>
Noam, what does ompi_info say - specifically which BTLs are available?
Stupid question though - this is a single system with no connection to a switch?
You probably dont have an OpenSM subnet manager running then - could
that be the root cause?
On Wed, 10 Oct 2018 at 09:53, Dave Love wrote:
>
>
RDMA was just broken in the last-but-one(?) RHEL7 kernel release, in
case that's the problem. (Fixed in 3.10.0-862.14.4.)
___
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users
"Gabriel, Edgar" writes:
> Ok, thanks. I usually run these test with 4 or 8, but the major item
> is that atomicity is one of the areas that are not well supported in
> ompio (along with data representations), so a failure in those tests
> is not entirely surprising .
If it's not expected to wo
19 matches
Mail list logo