[gmx-users] gmx_mpi works in command line but not in script

Marc Hoemberger hoemberg at brandeis.edu
Wed Apr 4 14:29:36 CEST 2018


Hi Alex,

from the looks of the module name it seems that what you are looking for is
the intel  Composer XE 2015.1.133. What could have happened is that the
architecture of your login node is different from some of the compute nodes
(almost certainly the case), so that when this composer was compiled the
person in charge of the cluster did not compile it for use on the compute
nodes.
That is why I did not have access to some of the compilers I wanted to use.
To fix this I ended up compiling the necessary software myself but making
sure that they are compiled with either (a) the target system being the
architecture of the compute node, (b) directly compiled on the compute
node, which seems difficult if you cannot access the nodes directly.

If you are using one of the standard scheduler you usually should have an
option to list all nodes and sometimes might be able to request a node
specifically (e.g. Grid Scheduler uses "qstat -f" to list all nodes, and
with qrsh you can ask for access to a compute node). If all else fails you
could always ask the person in charge for software deployment of your
cluster, although I would first try to exhaust the other paths.

-Marc


On Tue, Apr 3, 2018 at 5:01 PM, Alex <alexanderwien2k at gmail.com> wrote:

> Hi Marc,
>
>
> On Tue, Apr 3, 2018 at 4:35 PM, Marc Hoemberger <hoemberg at brandeis.edu>
> wrote:
>
> > Hi Alex,
> >
> > I have had a similar issue on our cluster (different module files etc.)
> for
> > me it turned out that on the actual compute nodes this module was not
> > available. Thus if you run it from the command line on e.g. your login
> node
> > you might have access to this compiler / modulefile, but when moving to a
> > compute node it might not be there.
> >
>
> Yes, it think that is the reason. How did you solve it?
>
> >
> > Did you try running the command line invocation on the same node you are
> > submitting the script to?
> >
> Actually, I can not find the list of node, otherwise I could log in inside
> one of them (if I was allowed) to invoke gromacs there.
>
> The rest of my run script is juts gmx_mpi grompp and mpirun gmx_mpi mdrun .
>
> Best regards,
> Alex
>
> >
> > -Marc
> >
> > On Tue, Apr 3, 2018 at 4:13 PM, Alex <alexanderwien2k at gmail.com> wrote:
> >
> > > On Tue, Apr 3, 2018 at 3:46 PM, Mark Abraham <mark.j.abraham at gmail.com
> >
> > > wrote:
> > >
> > > > Hi,
> > > >
> > > > On Tue, Apr 3, 2018 at 8:56 PM Alex <alexanderwien2k at gmail.com>
> wrote:
> > > >
> > > > > Thanks Marc.
> > > > >
> > > > > I removed the two EXPORT line in the script but still missing the
> > > library
> > > > > (removing only the second line also did not help):
> > > > >
> > > > > ModuleCmd_Load.c(208):ERROR:105: Unable to locate a modulefile for
> > > > > 'intel/composerxe/2015.1.133'
> > > > >
> > > >
> > > > Why isn't this module being found?
> > > >
> > > I do not know. The funny point is that, the intel/composerxe/2015.1.133
> > is
> > > needed when I invoke gmx_mpi in command line, and after loading it, the
> > > gmx_works fine. But I do not know why it can not be found inside
> script.
> > >
> > > >
> > > >
> > > > > ------------------------------------------------------------
> > > > --------------
> > > > > mpirun was unable to find the specified executable file, and
> > therefore
> > > > > did not launch the job.  This error was first reported for process
> > > > > rank 0; it may have occurred for other processes as well.
> > > > >
> > > >
> > > > If this was the error message, then you won't get a log file from the
> > > > executable that wasn't found, so I presume that below you are looking
> > at
> > > an
> > > > old log file...
> > > >
> > > > No, it can generate .log file, I ran it again and a new log file was
> > > generated.
> > >
> > > >
> > > > > And .log file:
> > > > >
> > > > > gmx_mpi: error while loading shared libraries:
> libmkl_intel_lp64.so:
> > > > cannot
> > > > > open shared object file: No such file or directory
> > > > >
> > > >
> > > > That's a library that has to come from the module that didn't load,
> so
> > > look
> > > > there.
> > > >
> > > >
> > > Best regards,
> > > Alex
> > >
> > >
> > > > Mark
> > > >
> > > >
> > > > > ------------------------------------------------------------
> > > > --------------
> > > > > mpirun was unable to find the specified executable file, and
> > therefore
> > > > > did not launch the job.  This error was first reported for process
> > > > > rank 0; it may have occurred for other processes as well.
> > > > >
> > > > > Regards,
> > > > > Alex
> > > > >
> > > > >
> > > > > On Tue, Apr 3, 2018 at 2:47 PM, Mark Abraham <
> > mark.j.abraham at gmail.com
> > > >
> > > > > wrote:
> > > > >
> > > > > > Hi,
> > > > > >
> > > > > > It looks like you are choosing to mismanage LD_LIBRARY_PATH. Try
> > > > letting
> > > > > > the modules do it, since it is their job.
> > > > > >
> > > > > > Mark
> > > > > >
> > > > > > On Tue, Apr 3, 2018, 20:29 Alex <alexanderwien2k at gmail.com>
> wrote:
> > > > > >
> > > > > > > Dear all,
> > > > > > >
> > > > > > > I have installed the gromacs2018.1 parallel in a cluster, it
> > works
> > > > > > > perfectly in the command line, of course after loading some
> > module.
> > > > > > > But when I put the commands along with loading module commands
> > in a
> > > > > bash
> > > > > > > script to be submitted in the cluster, it crashes with the
> below
> > > > > massage:
> > > > > > >
> > > > > > > ++++++++++++++++++
> > > > > > > ModuleCmd_Load.c(208):ERROR:105: Unable to locate a modulefile
> > for
> > > > > > > 'intel/composerxe/2015.1.133'
> > > > > > > gmx_mpi: /cm/shared/apps/gcc/4.8.1/lib64/libstdc++.so.6:
> version
> > > > > > > `GLIBCXX_3.4.20' not found (required by
> > > > > > > /home/alex23/.local/gromacs-2018.1/bin/../lib64/
> > > libgromacs_mpi.so.3)
> > > > > > >
> > > > > > > And part of the log file:
> > > > > > >
> > > > > > > gmx_mpi: /cm/shared/apps/gcc/4.8.1/lib64/libstdc++.so.6:
> version
> > > > > > > `GLIBCXX_3.4.20' not found (required by
> > > > > > > /home/alex44/.local/gromacs-2018.1/bin/../lib64/
> > > libgromacs_mpi.so.3)
> > > > > > > gmx_mpi: /cm/shared/apps/gcc/4.8.1/lib64/libstdc++.so.6:
> version
> > > > > > > `GLIBCXX_3.4.20' not found (required by
> > > > > > > /home/alex44/.local/gromacs-2018.1/bin/../lib64/
> > > libgromacs_mpi.so.3)
> > > > > > > gmx_mpi: /cm/shared/apps/gcc/4.8.1/lib64/libstdc++.so.6:
> version
> > > > > > > `GLIBCXX_3.4.20' not found (required by
> > > > > > > /home/alex44/.local/gromacs-2018.1/bin/../lib64/
> > > libgromacs_mpi.so.3)
> > > > > > > gmx_mpi: error while loading shared libraries:
> > > libmkl_intel_lp64.so:
> > > > > > cannot
> > > > > > > open shared object file: No such file or directory
> > > > > > > ++++++++++++++++++
> > > > > > >
> > > > > > > Here is part of my script.sh file
> > > > > > > ++++++++++++++++++
> > > > > > > . /etc/profile.d/modules.sh
> > > > > > > module load intel/composerxe/2015.1.133
> > > > > > > module load shared
> > > > > > > module load clusterA
> > > > > > > module load gcc/4.8.1
> > > > > > > module load sge/univa
> > > > > > > module load clusterA-openmpi/intel/2015/1.8.1-mlnx-ofed
> > > > > > > export PATH=$HOME/.local/gromacs-2018.1/bin:$PATH
> > > > > > > export
> > > > > > >
> > > > > > > LD_LIBRARY_PATH=$HOME/.local/gromacs-2018.1/lib64${LD_
> > > > > > LIBRARY_PATH:+:$LD_LIBRARY_PATH}
> > > > > > > ++++++++++++++++
> > > > > > >
> > > > > > > Do you know what I am missing?
> > > > > > > Thanks
> > > > > > > Alex
> > > > > > > --
> > > > > > > Gromacs Users mailing list
> > > > > > >
> > > > > > > * Please search the archive at
> > > > > > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List
> > before
> > > > > > > posting!
> > > > > > >
> > > > > > > * Can't post? Read http://www.gromacs.org/
> Support/Mailing_Lists
> > > > > > >
> > > > > > > * For (un)subscribe requests visit
> > > > > > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx
> > -users
> > > > or
> > > > > > > send a mail to gmx-users-request at gromacs.org.
> > > > > > >
> > > > > > --
> > > > > > Gromacs Users mailing list
> > > > > >
> > > > > > * Please search the archive at http://www.gromacs.org/
> > > > > > Support/Mailing_Lists/GMX-Users_List before posting!
> > > > > >
> > > > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > > > > >
> > > > > > * For (un)subscribe requests visit
> > > > > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_
> gmx-users
> > > or
> > > > > > send a mail to gmx-users-request at gromacs.org.
> > > > > >
> > > > > --
> > > > > Gromacs Users mailing list
> > > > >
> > > > > * Please search the archive at
> > > > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > > > > posting!
> > > > >
> > > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > > > >
> > > > > * For (un)subscribe requests visit
> > > > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users
> > or
> > > > > send a mail to gmx-users-request at gromacs.org.
> > > > >
> > > > --
> > > > Gromacs Users mailing list
> > > >
> > > > * Please search the archive at http://www.gromacs.org/
> > > > Support/Mailing_Lists/GMX-Users_List before posting!
> > > >
> > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > > >
> > > > * For (un)subscribe requests visit
> > > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users
> or
> > > > send a mail to gmx-users-request at gromacs.org.
> > > >
> > > --
> > > Gromacs Users mailing list
> > >
> > > * Please search the archive at http://www.gromacs.org/
> > > Support/Mailing_Lists/GMX-Users_List before posting!
> > >
> > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >
> > > * For (un)subscribe requests visit
> > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > > send a mail to gmx-users-request at gromacs.org.
> > >
> >
> >
> >
> > --
> > Marc Hoemberger
> > Ph.D. Candidate, Biochemistry and Biophysics,
> > Laboratory of Dorothee Kern
> > MS-009, 415 South St.
> > Brandeis University
> > Waltham, MA, 02453
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at http://www.gromacs.org/Support
> > /Mailing_Lists/GMX-Users_List before posting!
> >
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> > * For (un)subscribe requests visit
> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > send a mail to gmx-users-request at gromacs.org.
> >
> --
> Gromacs Users mailing list
>
> * Please search the archive at http://www.gromacs.org/
> Support/Mailing_Lists/GMX-Users_List before posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>



-- 
Marc Hoemberger
Ph.D. Candidate, Biochemistry and Biophysics,
Laboratory of Dorothee Kern
MS-009, 415 South St.
Brandeis University
Waltham, MA, 02453


More information about the gromacs.org_gmx-users mailing list