|
[Sponsors] |
Installation on a cluster running a queueing system PBS |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
June 9, 2005, 19:51 |
When I log on the system, the
|
#1 |
Member
diablo80@web.de
Join Date: Mar 2009
Posts: 93
Rep Power: 17 |
When I log on the system, the environment variables are all correct.
When I issue a which blockMesh, for instance, it returns the correct directory. But when I submit through the queueing system, I got "command not found" msgs, and when I include "which blockMesh" in the script to be submitted, it returns nothing (no command). As I am not experienced in clusters (this one is a beowulf) with queues, I am wondering if this is some common and easy problem and if anyone can help me... Note: I am not allowed to log onto the nodes and run an application from there... Thanks a lot, luiz |
|
June 10, 2005, 04:07 |
why not source your .cshrc the
|
#2 |
Super Moderator
Niklas Nordin
Join Date: Mar 2009
Location: Stockholm, Sweden
Posts: 693
Rep Power: 29 |
why not source your .cshrc the first thing you do inthe script
|
|
June 10, 2005, 15:50 |
I did it.
Now, the variables
|
#3 |
Member
diablo80@web.de
Join Date: Mar 2009
Posts: 93
Rep Power: 17 |
I did it.
Now, the variables are correctly set, but the output still contains errors. Here is my script: ################################################## ####### echo Running on host `hostname` echo Time is `date` echo Directory is `pwd` echo This job runs on the following processors: source /home/server/luizebs/OpenFOAM/OpenFOAM-1.1/.OpenFOAM-1.1/cshrc echo `which blockMesh` echo `which mpiexec` blockMesh . GL3 > logblock & ################################################## ####### The logblock file is completely empty... in my script.eXXXXX log file I got: ################################################## ####### [luizebs@redhot oodles]$ cat script.e45957 $: Command not found. $: Command not found. $: Command not found. $: Command not found. [luizebs@redhot oodles]$ ################################################## ####### while in the script.oXXXXX log I got (should I worry about this warning message?): ################################################## ####### [luizebs@redhot oodles]$ cat script.o45959 Warning: no access to tty (Bad file descriptor). Thus no job control in this shell. Running on host n46 Time is Fri Jun 10 10:39:17 PDT 2005 Directory is /home/server/luizebs This job runs on the following processors: /home/server/luizebs/OpenFOAM/OpenFOAM-1.1/applications/bin/linuxOpt/blockMesh /home/server/luizebs/OpenFOAM/OpenFOAM-1.1/src/lam-7.1.1/platforms/linuxOpt/bin/ mpiexec [1] 25710 -------------------------------------- Running PBS epilogue script Killing processes of user luizebs on the batch nodes Doing node n46 Done. [luizebs@redhot oodles]$ ################################################## ####### |
|
June 11, 2005, 12:11 |
ok, so you are standing in $HO
|
#4 |
Super Moderator
Niklas Nordin
Join Date: Mar 2009
Location: Stockholm, Sweden
Posts: 693
Rep Power: 29 |
ok, so you are standing in $HOME according to pwd.
is your GL3 directory located in $HOME? also i would add #!/bin/tcsh (or csh or bash...) on the first line of the script (might help) N |
|
June 13, 2005, 22:32 |
Sorry that was an old log. Eve
|
#5 |
Member
diablo80@web.de
Join Date: Mar 2009
Posts: 93
Rep Power: 17 |
Sorry that was an old log. Even when I was (correctly) on the .../tutorials/oodles dir, the message was the same. (And I did have #!/bin/bashrc in my script).
I tried to execute blockMesh from the node (avoiding qsub) and it issued these msgs: [luizebs@n1 oodles]$ blockMesh . GL3 blockMesh: /lib/i686/libc.so.6: version `GLIBC_2.3' not found (required by /home/server/luizebs/OpenFOAM/linux/gcc-3.4.3/lib/libstdc++.so.6) blockMesh: /lib/i686/libc.so.6: version `GLIBC_2.3.2' not found (required by /home/server/luizebs/OpenFOAM/OpenFOAM-1.1/src/lam-7.1.1/platforms/linuxOpt/lib/ libmpi.so.0) blockMesh: /lib/i686/libc.so.6: version `GLIBC_2.3' not found (required by /home/server/luizebs/OpenFOAM/OpenFOAM-1.1/src/lam-7.1.1/platforms/linuxOpt/lib/ liblam.so.0) blockMesh: /lib/i686/libpthread.so.0: version `GLIBC_2.3.2' not found (required by /home/server/luizebs/OpenFOAM/OpenFOAM-1.1/src/lam-7.1.1/platforms/linuxOpt/lib/ liblam.so.0) [luizebs@n1 oodles]$ which gcc ~/OpenFOAM/linux/gcc-3.4.3/bin/gcc [luizebs@n1 oodles]$ Do you have any clue? Apparently gcc is pointing to the correct location. Thanks, Luiz |
|
June 14, 2005, 05:56 |
your linux version on the clus
|
#6 |
Super Moderator
Niklas Nordin
Join Date: Mar 2009
Location: Stockholm, Sweden
Posts: 693
Rep Power: 29 |
your linux version on the cluster is too old, which version is it?
You need to recompile all of the source on the cluster. Submit the Allwmake script. N |
|
June 14, 2005, 13:48 |
2.4.20-30.7
When I recompil
|
#7 |
Member
diablo80@web.de
Join Date: Mar 2009
Posts: 93
Rep Power: 17 |
2.4.20-30.7
When I recompile (with ./Allwmake from $HOME/OpenFOAM/OpenFOAM-1.1 dir) it still complains about GLIBC_2.3 Here is the output: SOURCE_DIR=. SOURCE=foamDebugSwitches.C ; g++ -m32 -Dlinux -Wall -W -Wno-unused-parameter -O3 -ffast-math -DNoRepository -ftemplate-depth-30 -I/home/server/luizebs/OpenFOAM/OpenFOAM-1.1/src/OpenFOAM/lnInclude -IlnInclude -I. -fPIC -pthread -c $SOURCE -o Make/linuxOpt/foamDebugSwitches.o g++: /lib/i686/libc.so.6: version `GLIBC_2.3' not found (required by g++) make: *** [Make/linuxOpt/foamDebugSwitches.o] Error 1 + wmake foamInfoExec Making dependency list for source file foamInfoExec.C SOURCE_DIR=. SOURCE=foamInfoExec.C ; g++ -m32 -Dlinux -Wall -W -Wno-unused-parameter -O3 -ffast-math -DNoRepository -ftemplate-depth-30 -I/home/server/luizebs/OpenFOAM/OpenFOAM-1.1/src/OpenFOAM/lnInclude -IlnInclude -I. -fPIC -pthread -c $SOURCE -o Make/linuxOpt/foamInfoExec.o g++: /lib/i686/libc.so.6: version `GLIBC_2.3' not found (required by g++) make: *** [Make/linuxOpt/foamInfoExec.o] Error 1 [luizebs@n22 OpenFOAM-1.1]$ Thanks again, luiz |
|
June 14, 2005, 15:40 |
Sounds like you first need to
|
#8 |
Senior Member
Mattijs Janssens
Join Date: Mar 2009
Posts: 1,419
Rep Power: 26 |
Sounds like you first need to compile gcc for your os version.
Information on how to do that should be somewhere on the site. (we do: ./configure --prefix=/usr/tmp/gcc-3.4.3 --enable-languages=c,c++ --enable-shared --enable-__cxa_atexit --disable-checking --with-system-zlib) |
|
June 15, 2005, 04:58 |
Yes and you will need to recom
|
#9 |
Super Moderator
Niklas Nordin
Join Date: Mar 2009
Location: Stockholm, Sweden
Posts: 693
Rep Power: 29 |
Yes and you will need to recompile everything else too.
But now your in a tight spot, how will you do the post-processing? I think you need to have two binary-versions on for the old cluster and one for the new workstation. N |
|
June 17, 2005, 02:11 |
I tried to compile gcc and Foa
|
#10 |
Member
diablo80@web.de
Join Date: Mar 2009
Posts: 93
Rep Power: 17 |
I tried to compile gcc and Foam (with Allwmake), but it still complains about GLIBC-2.3.2.
Apparently gcc was installed with no problem, but ./Allwmake outputs a lot of messages like this: /home/server/luizebs/OpenFOAM/OpenFOAM-1.1/src/lam-7.1.1/platforms/linuxOpt/lib/ liblam.so.0: undefined reference to `pthread_cond_signal@GLIBC_2.3.2' Should I install this library separetely? (I thought it should come with gcc-3.4.3) Where should it be? What I have in /lib dir is: $ ls /lib/libc* /lib/libc-2.2.5.so /lib/libcap.so.1.10 /lib/libcrypt-2.2.5.so /lib/libcrypt.so.1 /lib/libcap.so /lib/libcom_err.so.2 /lib/libcrypto.so.0.9.6b /lib/libc.so.6 /lib/libcap.so.1 /lib/libcom_err.so.2.0 /lib/libcrypto.so.2 $ Thanks a lot again, Luiz |
|
June 17, 2005, 20:12 |
Also, which version of GLIBC i
|
#11 |
Member
diablo80@web.de
Join Date: Mar 2009
Posts: 93
Rep Power: 17 |
Also, which version of GLIBC is required by Foam?
I googled about installing GLIBC and got a little scared... So, firstly I would like to be sure it is really the case, rather than anything else that i am missing... Thanks, luiz |
|
June 18, 2005, 12:04 |
It looks to me like you have s
|
#12 |
Super Moderator
Niklas Nordin
Join Date: Mar 2009
Location: Stockholm, Sweden
Posts: 693
Rep Power: 29 |
It looks to me like you have some old libraries hanging around.
Did you remove the OpenFoam installation and install just the sources or did you just try to recompile with the old binaries installed? The glibc-version is not something foam specific, but machine specific. What it means is that you have compiled your library on a certain system and then running it on an other system where these libraries are missing/different. N |
|
June 20, 2005, 18:16 |
Hi Niklas,
I did not remove t
|
#13 |
Member
diablo80@web.de
Join Date: Mar 2009
Posts: 93
Rep Power: 17 |
Hi Niklas,
I did not remove the OpenFoam installation. I just ran ./Allwmake from top dir. What exactly should I do? (which directories should I leave and which ones should I remove before running ./Allwmake?) If you can provide me more details, since I am not linux expert, I will be very very gratefull... Thanks a lot again, luiz |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
What is the best cluster file system to go with OF | sampaio | OpenFOAM Installation | 4 | March 17, 2011 05:10 |
Installation on IBM machine running AIX | gellert | OpenFOAM Installation | 2 | October 14, 2005 05:26 |
Running FineTurbo on a Linux Cluster | Philipp Höfliger | Fidelity CFD | 3 | May 6, 2004 09:47 |
Who is running Phoenics with a cluster? | Maria Morabito | Phoenics | 2 | January 4, 2004 06:07 |
Running CFX on a cluster | jvk | CFX | 9 | September 19, 2002 23:22 |