On Wed, Nov 17, 2010 at 5:34 PM, Linus Johannissen <
Linus.Johannissen.manchester.ac.uk> wrote:
> Thanks Tru, but I did try both with and without first installing AmberTools
> (sorry I forgot to mention that!), and I get the same error when I run
> "make" in $AMBERHOME/AmberTools/src/.
>
> This time I started over with a clean slate, trying just to install
> AmberTools. One thing I find perplexing is that even though I run
> "./configure intel", the first few lines when running make seem to suggest
> it's using f90, although later on it's using icc and ifort (no idea if this
> is normal):
>
It's not *using* f90. The message says "making install in f90", and it then
moves into an "f90" directory where it runs "make install". This is a
netcdf issue, so it may help to look at the netcdf build log, because
something is going wrong.
I'm going to side with Dave Case on this one. What you should do before you
install Amber is *unload* your Gaussian module, since it's plopping an
"install" into your path that I'm guessing is mucking up the install
process. I have never seen an error in the NETCDF build (which is where
this is occurring) before, and I've installed it on a lot of different
systems/clusters.
As long as you see the "phrase": /opt/software/gaussian/d02/g03/bsd/install
-c -m 644
then the compilation process is using the *wrong* install which is likely
breaking stuff.
As a note to general users: Do NOT source Gaussian files that set up your
environment in normal shell environments. ONLY source them in PBS scripts
that specifically run Gaussian. The reason is that their profile scripts
that set up environment variables are effectively poison to every other
package that uses these variables (notably LD_LIBRARY_PATH). These scripts
set up the environment variables properly for Gaussian, so g03/g09 will run
properly, but it could easily break other packages (this drove me insane for
some time before I figured it out). Therefore, source the Gaussian scripts
and/or load the Gaussian module only in sessions that you are actually
running a Gaussian job.
Hope this helps,
Jason
> cd netcdf/src && make install
> make[1]: Entering directory
> `/home/linusj/amber/amber11/AmberTools/src/netcdf/src'
> Making install in f90
> make[2]: Entering directory
> `/home/linusj/amber/amber11/AmberTools/src/netcdf/src/f90'
>
> Anyway, the only error message I can spot is the same as before:
> chmod: cannot access
> `/home/linusj/amber/amber11/AmberTools/src/netcdf/lib/libnetcdf.a': No such
> file or directory
> make[3]: *** [install-libLTLIBRARIES] Error 1
> make[2]: *** [install-am] Error 2
> make[1]: *** [install-recursive] Error 1
> make: *** [../netcdf/lib/libnetcdf.a] Error 2
>
> As always, thank you for your suggestions
> Linus
>
> ____________________
> Dr Linus Johannissen
> Post-doctoral research associate
> Manchester Interdisciplinary Biocentre
> University of Manchester
> UK
>
> ________________________________________
> From: Tru Huynh [tru.pasteur.fr]
> Sent: 17 November 2010 18:40
> To: AMBER Mailing List
> Subject: Re: [AMBER] Amber 11 installation problems
>
> On Wed, Nov 17, 2010 at 05:44:21PM +0000, Linus Johannissen wrote:
> > Thanks for the quick reply!
> >
> > Here are some more details:
> > 1. I untarred Amber 11 and AmberTools 1.4 in /home/linusj/amber, and set
> $AMBERHOME to /home/linusj/amber/amber11
> > 2. I applied the bug fixes for both Amber 11 and AmberTools 1.4
> > 3. The directory /home/linusj/amber/amber11/AmberTools/src/netcdf/ was
> created, but it does not contain libnetcdf.a
> > 4. Step-by-step, this was my procedure: 1) untar Amber and AmberTools; 2)
> > apply the patches for Amber (apply_bugfix.x bugfix.all); 3) apply the
> patch
> > for AmberTools (patch -p0 -N < bugfix.all) -- and no, I didn't use the
> same
> > "bugfix.all" for both; 4) use "configure intel" in
> $AMBERHOME/AmberTools/src/
> > to generate the config.h files;
>
> STOP
> you forgot to build ambertools
> make && make install
> (cd $AMBERHOME/test && make test)
>
> before continuing
>
> Tru
>
> --
> Dr Tru Huynh | http://www.pasteur.fr/recherche/unites/Binfs/
> mailto:tru.pasteur.fr | tel/fax +33 1 45 68 87 37/19
> Institut Pasteur, 25-28 rue du Docteur Roux, 75724 Paris CEDEX 15 France
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 17 2010 - 15:30:04 PST
