[CIG-SHORT] Problem on the installation of Pylith

Ikuo Cho ikuo-chou at aist.go.jp
Mon Mar 16 19:27:26 PDT 2009 

Lief, Brad;

As seen from the cofig file, I actually set the variable as 
PYTHON=/home/beowulf/local/bin/python
when compiling Pylith. 

LS> Alternatively, you can explicitly tell 'configure' where 'nemesis' is 
LS> located:
LS> 
LS> ./configure --prefix=/home/beowulf/local --enable-cubit 
LS> PYTHON=/home/beowulf/local/bin/nemesis

According to your suggestion, I did set the variable PYTHON to nemesis
explicitly and complide Pylith agan. Finally, I was able to ran the
example. 
Thank you for your insightful comment!

On Mon, 16 Mar 2009 17:48:45 -0700
Leif Strand <leif at geodynamics.org> wrote:

LS> Ikuo,
LS> 
LS> I suspect that your PYTHON environment variable is set, and this 
LS> confused the PyLith 'configure' script.  Normally, for the PyLith build, 
LS> PYTHON is automatically set to the path of 'nemesis'.  But if the PYTHON 
LS> environment variable is set, the 'configure' script will use that 
LS> instead (I think).
LS> 
LS> So, before you run PyLith 'configure', make sure that PYTHON is not set:
LS> 
LS> unset PYTHON
LS> ./configure --prefix=/home/beowulf/local --enable-cubit
LS> 
LS> This will cause the 'configure' script to search for 'nemesis' on your 
LS> PATH, and set PYTHON accordingly.
LS> 
LS> Alternatively, you can explicitly tell 'configure' where 'nemesis' is 
LS> located:
LS> 
LS> ./configure --prefix=/home/beowulf/local --enable-cubit 
LS> PYTHON=/home/beowulf/local/bin/nemesis
LS> 
LS> Using "PYTHON=/home/beowulf/local/bin/nemesis" on the command line will 
LS> override the environment setting.
LS> 
LS> --Leif
LS> 
LS> 
LS> Brad Aagaard wrote:
LS> > Ikuo-
LS> > 
LS> > If the import mpi worked, then I think the examples should run. Please try 
LS> > them again. There is a chance some environment variable changed.
LS> > 
LS> > If they do not work, check to see if the environment variables are any 
LS> > different in the config.log when configuring PyLith compared to the output of 
LS> > running the env command in the directory where you tried to run the example.
LS> > 
LS> > Brad
LS> > 
LS> > 
LS> > On Monday 16 March 2009 5:27:12 pm Ikuo Cho wrote:
LS> >> Brad,
LS> >>
LS> >> Thanks for the reply.
LS> >> I did "nemesis & import mpi" and obtained
LS> >>
LS> >> [beowulf at localhost ~]$ nemesis
LS> >> Python 2.5.2 (r252:60911, Mar 16 2009, 16:42:20)
LS> >> [GCC 4.1.2 20070925 (Red Hat 4.1.2-33)] on linux2
LS> >> Type "help", "copyright", "credits" or "license" for more information.
LS> >>
LS> >>>>> import mpi
LS> >> It seems that nemesis imported mpi correctly.
LS> >> I append two config files that you requested to this email, hoping that
LS> >> the problem will be solved.
LS> >>
LS> >> Ikuo Cho
LS> >>
LS> >>
LS> >> On Mon, 16 Mar 2009 08:35:34 -0700
LS> >> Brad Aagaard <baagaard at usgs.gov> wrote:
LS> >>
LS> >> BA> Ikuo-
LS> >> BA>
LS> >> BA> It looks like PyLith built okay, but there is probably something wrong
LS> >> in your BA> environment variables. "import mpi" should fail in a python
LS> >> shell, but not in BA> a python shell started via nemesis.
LS> >> BA>
LS> >> BA> What happens when you run the following?
LS> >> BA>
LS> >> BA> nemesis
LS> >> BA> import mpi
LS> >> BA>
LS> >> BA> Please send the config.log files that were created when installing
LS> >> nemesis and BA> PyLith.
LS> >> BA>
LS> >> BA> Brad
LS> >> BA>
LS> >> BA>
LS> >> BA>
LS> >> BA> On Monday 16 March 2009 4:27:26 am Ikuo Cho wrote:
LS> >> BA> > Dear pylith developers,
LS> >> BA> >
LS> >> BA> > I have installed Pylith from the source code into a Linux machine
LS> >> with BA> > specification:
LS> >> BA> >
LS> >> BA> > OS: Fedora Core 8
LS> >> BA> > CPU: Intel(R) Pentium(R) D CPU 3.00GHz
LS> >> BA> > uname -m = i686
LS> >> BA> > uname -r = 2.6.24.5-85.fc8
LS> >> BA> > uname -s = Linux
LS> >> BA> > uname -v = #1 SMP Sat Apr 19 12:39:34 EDT 2008
LS> >> BA> >
LS> >> BA> > As a note, I installed Python2.5.2 to /home/beowulf/local and export
LS> >> BA> > PYTHONPATH=/home/beowulf/local/lib/python2.5/site-packages
LS> >> BA> > although I have another python version 2.5.1 in the system directory
LS> >> BA> > /usr. I installed pylith so that it does NOT depend on MPICH but
LS> >> MPICH2. BA> >
LS> >> BA> > No noticeable warning was found in the installation process of
LS> >> Pylith, BA> > but I was not able to run it correctly. To be specific, I
LS> >> tried to run BA> > Pylith in the directory
LS> >> BA> > /home/beowulf/Shoyu_cho.1.3.1/pylith-1.3.1/examples/3d/hex8 by typing
LS> >> BA> > “ pylith dislocation.cfg”. However, I only obtained the following
LS> >> BA> > messages.
LS> >> BA> >
LS> >> BA> > [beowulf at localhost hex8]$ pylith dislocation.cfg
LS> >> BA> > Traceback (most recent call last):
LS> >> BA> >   File "/home/beowulf/local/bin/pylith", line 31, in <module>
LS> >> BA> >     from pylith.PyLithApp import PyLithApp
LS> >> BA> >   File
LS> >> BA> >
LS> >> "/home/beowulf/local/lib/python2.5/site-packages/pylith/PyLithApp.py", line
LS> >> BA> > 17, in <module> from mpi import Application
LS> >> BA> >   File
LS> >> BA> >
LS> >> "/home/beowulf/local/lib/python2.5/site-packages/pythia-0.8.1.7-py2.5.egg/m
LS> >> BA> >pi/__init__.py", line 14, in <module> from _mpi import *
LS> >> BA> > ImportError: No module named _mpi
LS> >> BA> >
LS> >> BA> > Could you tell me what happened?
LS> >> BA> > I append two make.log files to this maiil and I think that the
LS> >> following BA> > information obtained from the python shell is useful to
LS> >> make a diagnosis. BA> >
LS> >> BA> > [beowulf at localhost ~]$ python
LS> >> BA> > Python 2.5.2 (r252:60911, Mar 16 2009, 16:42:20)
LS> >> BA> > [GCC 4.1.2 20070925 (Red Hat 4.1.2-33)] on linux2
LS> >> BA> > Type "help", "copyright", "credits" or "license" for more
LS> >> information. BA> >
LS> >> BA> > >>> import sys
LS> >> BA> > >>> sys.path
LS> >> BA> >
LS> >> BA> > ['',
LS> >> '/home/beowulf/local/lib/python2.5/site-packages/merlin-1.6.egg', BA> >
LS> >> '/home/beowulf/local/lib/python2.5/site-packages/pythia-0.8.1.7-py2.5.egg',
LS> >> BA> >
LS> >> '/home/beowulf/local/lib/python2.5/site-packages/Cheetah-2.0rc8-py2.5-linux
LS> >> BA> >-i686.egg', '/home/beowulf/local/lib/python2.5/site-packages', BA> >
LS> >> '/home/beowulf/local/lib/python25.zip',
LS> >> BA> > '/home/beowulf/local/lib/python2.5',
LS> >> BA> > '/home/beowulf/local/lib/python2.5/plat-linux2',
LS> >> BA> > '/home/beowulf/local/lib/python2.5/lib-tk',
LS> >> BA> > '/home/beowulf/local/lib/python2.5/lib-dynload']
LS> >> BA> >
LS> >> BA> > >>> import mpi
LS> >> BA> >
LS> >> BA> > Traceback (most recent call last):
LS> >> BA> >   File "<stdin>", line 1, in <module>
LS> >> BA> >   File
LS> >> BA> >
LS> >> "/home/beowulf/local/lib/python2.5/site-packages/pythia-0.8.1.7-py2.5.egg/m
LS> >> BA> >pi/__init__.py", line 14, in <module> from _mpi import *
LS> >> BA> > ImportError: No module named _mpi
LS> >> BA> >
LS> >> BA> > Also,
LS> >> BA> >
LS> >> BA> > [beowulf at localhost ~]$ nemesis
LS> >> BA> > Python 2.5.2 (r252:60911, Mar 16 2009, 16:42:20)
LS> >> BA> > [GCC 4.1.2 20070925 (Red Hat 4.1.2-33)] on linux2
LS> >> BA> > Type "help", "copyright", "credits" or "license" for more
LS> >> information. BA> >
LS> >> BA> > >>> import _mpi
LS> >> BA> >
LS> >> BA> > After all, the nemesis  seems to have been installed correctly and I
LS> >> BA> > have a single MPI installation on this machine (I installed MPI for
LS> >> BA> > Pylith). Also, I made sure the correct 'mpiexec' and 'python' in my
LS> >> path BA> > when installing nemesis.
LS> >> BA> >
LS> >> BA> > Ikuo Cho
LS> >> BA> > ----------------------------------------------------------
LS> >> BA> > Geological Survey of Japan,
LS> >> BA> > National Institute of Advanced Industrial Science and Technology
LS> >> BA> > Tsukuba Central 7, Tsukuba 305-8567 Japan
LS> >> BA> > Tel +81-29-861-3891, Fax +81-29-861-3682
LS> >> BA> > ----------------------------------------------------------
LS> >> BA>
LS> >> BA>
LS> >> BA> _______________________________________________
LS> >> BA> CIG-SHORT mailing list
LS> >> BA> CIG-SHORT at geodynamics.org
LS> >> BA> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-short
LS> >>
LS> >>
LS> >> 長　郁夫 (Ikuo Cho)
LS> >> ----------------------------------------------------------
LS> >> 独立行政法人　産業技術総合研究所
LS> >> 地質情報研究部門　地震発生機構研究グループ
LS> >> 〒305-8567　つくば市東1-1-1　つくば中央第七
LS> >> Tel 029-861-3891
LS> >>      029-861-3688,3534(つながらない場合はこちら迄)
LS> >> Fax 029-861-3682, e-mail ikuo-chou at aist.go.jp
LS> >> ----------------------------------------------------------
LS> >> ※  転送先の個人アドレスから容量オーバーのメッセージが届く場合
LS> >>     がありますが，オフィスには届いておりますので再送は不要です。
LS> > 
LS> > 
LS> > _______________________________________________
LS> > CIG-SHORT mailing list
LS> > CIG-SHORT at geodynamics.org
LS> > http://geodynamics.org/cgi-bin/mailman/listinfo/cig-short
LS> 


長　郁夫 (Ikuo Cho)
----------------------------------------------------------
独立行政法人　産業技術総合研究所
地質情報研究部門　地震発生機構研究グループ
〒305-8567　つくば市東1-1-1　つくば中央第七
Tel 029-861-3891
     029-861-3688,3534(つながらない場合はこちら迄)
Fax 029-861-3682, e-mail ikuo-chou at aist.go.jp
----------------------------------------------------------
※  転送先の個人アドレスから容量オーバーのメッセージが届く場合
    がありますが，オフィスには届いておりますので再送は不要です。

More information about the CIG-SHORT mailing list