Atom topic feed | site map | contact | login | Protection des données personnelles | Powered by FluxBB | réalisation artaban
You are not logged in.
The same applies for the path of the files in the export file.
Define them as if they are located at "/home/aster/shared/"
Aether Engineering
- aethereng.com -
Offline
ok i have got the points
but it seems that i have to run twice as_run to get the results file written to the directory i started the problem at first
or
it looks like there is fairly long timme (several teens of second) before the results are writtem to the directory i started the problem at first
one more question why does the .mess file ends with
<I>_EXIT_CODE = 2
i still have to get confidence to this way of running code_aster
consider reading my book
freely available here https://framabook.org/beginning-with-code_aster/
Offline
I am afraid I cannot help on this one.
I get the results real-time in my folder and the mess file is as expected.
If you want to share the export file for a check, maybe there is something wrong there but I don't really know
Aether Engineering
- aethereng.com -
Offline
Ioannis
thanks for your valuable help
i found out that i have to type a Ctrl C in the terminal to really end up the study and get the results properly saved
compared to a regular linux code_aster run of the same problem the results are ok
the speed is a few percent slower, quite good
i have to make more test but i am not really enthusiastic about docker code_aster at the time
main draw back is the black box side and the lack of a proper reliable doc
consider reading my book
freely available here https://framabook.org/beginning-with-code_aster/
Offline
Thanks Ioannis for your help, and also many thanks also to Jean-Pierre for your valuable remarks.
Like Code_Aster, Docker itself as well as this image are open-source: their source codes are all deposited on Github.
Docker itself is rather well documented, see docs.docker.com
Concerning this Docker image for Code_Aster..it is currently just a personal project and of coruse lacks a comprehensive documentation and may also contain bugs or performance lags..I'm not myself a purely IT geek (like many others just a computational mechanical guy) or an expert on Docker. So I decided to open-source this project so that others can share their experiences.
Offline
Thanks Ioannis for your help
You are welcome Tianyi, it is a pleasure to support and promote this work!
I find the value of this contribution to be very high because:
1. I cannot compare the potential problems one can have when trying to install code_aster and its prerequisites with the luxury of just downloading an image of the software. I mean the installation process is ok but for some distros or when you are re-installing another version or an additional component there are cases where one can face serious problems that are difficult to understand the origin and to overcome.
2. Code_aster does not come precompiled for Windows, let alone Mac OS. So it is a really great opportunity to promote its dissemination among the Windows users. And of course, I am really happy to be able to run even the parallel version in Windows!
Best,
Ioannis
Aether Engineering
- aethereng.com -
Offline
To your 1st question, make sure that you allocate enough CPU to your Docker environment. You can go to the settings and allocate the number of CPU's you want to have in your Docker containers.
To your 2nd question, you need to add
P version stable_mpi
in your export file.
Thank you, now seems that is running in more than 1 processor, but when reaches the analysis part, the next error message appears:
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
! <S> Exception utilisateur levee mais pas interceptee. !
! Les bases sont fermees. !
! Type de l'exception : error !
! !
! Erreur de programmation lors de l'assemblage : !
! Les quantités que l'on cherche à assembler (MATR_ELEM ou VECT_ELEM) ont été !
! calculées avec au !
! moins 2 partitions différentes : _0000002 et _0000004 !
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Should I change the solver to mumps? In the test cases they uses the mult_front solver...
thank you so much and regards!
Offline
What is for sure is that the MULT_FRONT solver doesn't support MPI parallelism, while MUMPS does. See [u4.50.01]
MULT_FRONT only consider OpenMP parallelism.
So you could indeed try using MUMPS in parallel. Please let me know if it works. Anyway in the latest development version MUMPS is now the default solver due to its robustness.
Offline
What is for sure is that the MULT_FRONT solver doesn't support MPI parallelism, while MUMPS does. See [u4.50.01]
MULT_FRONT only consider OpenMP parallelism.
So you could indeed try using MUMPS in parallel. Please let me know if it works. Anyway in the latest development version MUMPS is now the default solver due to its robustness.
I already try it but it doesnt work, I dont know what else could be...
Thank you in advance! Regards!
Offline
I had try a set of configurations but nothing seems to work, even the test cases for the parallel version gets the same error when I raise the number of processors (the default is 1) in the export file. I was wondering if maybe you have some testcases for me to run so I can check if the error belongs to my system configuration or image settings or something else. Thank you so much! I am really excited about this topic because I have spent almost 2 months trying to compile the parallel aster version.
I really appreciate any help you can provide me.
Regards!
Last edited by MateoMartinez123 (2018-07-03 23:31:17)
Offline
Hi Mateo,
To verify that the parallel version works, the simplest way is to run the so-called unit tests. So you can for instance run one or several tests that are specific for the parallel version.
You can generate a list of these parallel units tests using
as_run --vers stable_mpi --list --all --filter='"parallel" in testlist'
Then just pick a random test to run, for instance the rather involved mumps05c
as_run --vers stable_mpi --test mumps05c
I got the following results
-- CODE_ASTER -- VERSION : EXPLOITATION (stable) --
Version 13.4.0 modifiée le 29/06/2017
révision 503b50e0ecda - branche 'v13'
Copyright EDF R&D 1991 - 2018
Exécution du : Wed Jul 4 07:49:04 2018
Nom de la machine : d4c0762afd16
Architecture : 64bit
Type de processeur : x86_64
Système d'exploitation : Linux Ubuntu 16.04 xenial 4.9.87-linuxkit-aufs
Langue des messages : en (UTF-8)
Version de Python : 2.7.12
Version de NumPy : 1.11.0
Parallélisme MPI : actif
Rang du processeur courant : 0
Nombre de processeurs utilisés : 8
Parallélisme OpenMP : actif
Nombre de processus utilisés : 2
Version de la librairie HDF5 : 1.8.14
Version de la librairie MED : 3.2.1
Version de la librairie MFront : 3.0.0
Version de la librairie MUMPS : 5.1.1
Version de la librairie PETSc : 3.7.7p0
Version de la librairie SCOTCH : 6.0.4
Mémoire limite pour l'exécution : 1274.00 Mo
consommée par l'initialisation : 319.56 Mo
par les objets du jeu de commandes : 0.00 Mo
reste pour l'allocation dynamique : 952.79 Mo
Taille limite des fichiers d'échange : 12.00 Go
<I> <FIN> FERMETURE DE LA BASE "GLOBALE" EFFECTUEE.
<FIN> Arrêt normal dans "FIN".
<I> <FIN> ARRET NORMAL DANS "FIN" PAR APPEL A "JEFINI".
---------------------------------------------------------------------------------
cpu system cpu+sys elapsed
---------------------------------------------------------------------------------
Preparation of environment 0.00 0.00 0.00 0.00
Copying datas 0.00 0.01 0.01 0.03
Code_Aster run 1505.91 421.43 1927.34 246.58
Code_Aster run - rank #0 92.22 26.79 119.01 0.00
Code_Aster run - rank #1 89.25 29.34 118.59 0.00
Code_Aster run - rank #2 89.45 29.64 119.09 0.00
Code_Aster run - rank #3 96.95 22.62 119.57 0.00
Code_Aster run - rank #4 89.74 28.97 118.71 0.00
Code_Aster run - rank #5 91.54 26.38 117.92 0.00
Code_Aster run - rank #6 94.98 23.85 118.83 0.00
Code_Aster run - rank #7 98.10 20.41 118.51 0.00
Copying results 0.00 0.00 0.00 0.00
---------------------------------------------------------------------------------
Total 1505.96 421.49 1927.45 246.79
---------------------------------------------------------------------------------
---------------------------------------------------------------------------------
(*) cpu and system times may be not correctly counted using mpirun.
as_run 1.13.10
------------------------------------------------------------
--- DIAGNOSTIC JOB : <A>_ALARM
------------------------------------------------------------
So it runs fine (only ALARM, not ERROR), even with 8 MPI cores mixed with 2 OpenMP threads...while I only allocated 4 CPU's to Docker.
As reported on github.com/tianyikillua/code_aster_on_docker, all the unit tests are passing, which include all parallel ones, except those that require a specific module (europlexus, etc).
Last edited by tianyikillua (2018-07-04 08:57:24)
Offline
Hello,
thank you for the code_aster docker image, will be very usefull.
I have been able to run the container and execute a test with as_run inside it.
Could you provide an example command to execute aster container passing arguments from the command line?
something probably close to
docker run -ti --rm -v $pwd:$pwd quay.io/tianyikillua/code_aster as_run --test forma02a
but my command currently does not work.
Regards,
Offline
Hello
You can do
docker run --rm quay.io/tianyikillua/code_aster "/home/aster/aster/bin/as_run --test forma02a"
To directly run a test without opening an interative shell
Or if you want to run `test.export` in your current directory (pwd), do
docker run --rm -v $(pwd):/home/aster/shared -w /home/aster/shared quay.io/tianyikillua/code_aster "/home/aster/aster/bin/as_run test.export"
Currently you have to specify the absolute path to as_run (/home/aster/aster/bin/as_run). The bashrc file is not loaded by doing this...maybe I need to modify the `.profile` file??? To investigate.
Last edited by tianyikillua (2018-07-07 11:21:45)
Offline
thank you for your precise answer:
docker run --rm -v $(pwd):/app -w /app quay.io/tianyikillua/code_aster "/home/aster/aster/bin/as_run export_calculContraintes.txt"
first try: success!
Offline
Hello
This is a great new thanks for sharing your work this is so helpfull!
for your information I tried the salome_meca and they where some missing lib I solved it with:
apt-get install libglu1-mesa libGLU libxmu6 net-tools
regards
fred
Offline
Hi,
Thanks for the information. Which test cases failed because of these packages?
Using the latest packages (stable and testing) (code-aster.org/spip.php?article1078), I updated the stable Docker image (to 13.6) and also added a new image for the testing version 14.2. They should be functional in 2 hours (around 14 August 2018 at 18:00 UTC+2 Paris time); currently they are being built for the last time.
Again, both the sequantial and the parallel versions are available for the stable version.
Only the sequential version is current available for the testing version though.
Cheeeers,
Tianyi
Offline
Out of interest, have you tried doing this for the whole Salome-Meca package or just Code_Aster?
Offline
1. If you mean using the officially compiled (using intel compilers) code_aster contained in the salome_meca package, there is already such Docker image quay.io/tianyikillua/salome_meca, where you can directly use the "as_run" command
2. If you mean using the GUI interface of salome_meca through Docker, there are lots of graphical issues (see for instance github.com/tianyikillua/code_aster_on_docker/issues/4, but also reported elsewhere) and there may be not a robust way to do this
Offline
Dear All,
Congratulations to you all for this great work and useful feed back. Though I have not used it yet, I think, it is a great tool and has the potential to be a widely used one. I am saying this due to the following reasons, based on my experience with many of my students:
1) Many students, initially, have a phobia for Linux, which is a usual phenomenon.
2) For introductory FEM course at undergraduate level, many students hesitate to venture into the beautiful Linux world.
3) Using VM puts heavy load on common PCs/Laptops normally used by students, and make the solution pretty slow; the system hangs many a times.
4) Code-aster and Salome-Meca for windows (http*//code-aster-windows.com) are promising, but at present has some issues. Few of my students could not get results for STAT_NON_LINE in a parametric study, using astk in Windows 10. However, the same astk file (with path adjusted) gave appropriate results in Linux (UBUNTU18.04).
5) Even in Linux, change of Distribution creates problems due to changed library, default python version etc., as mentioned earlier in this thread.
These clearly indicates the advantages of docker, to run Code-aster in all platforms.
Now, I would like to know, whether only as_run option is available or also astk way of running a study is possible.
With thanks and regards.
Sukumar
Offline
Dear Sukumar,
Thank you for your remarks. As a matter of fact you can use `astk` with this Docker image, even under Windows.
1. For Windows users, download and open a X11 server via MobaXterm or vcXsrc, etc.
2. Get your IP address
3. Run (here for Windows users, note %cd%) in the directory containing simulations files
docker run -ti --rm -e DISPLAY=[YOUR IP ADDRESS]:0 -v %cd%:/home/aster/shared -w /home/aster/shared quay.io/tianyikillua/code_aster
Note: please replace
[YOUR IP ADDRESS]
by your actual IP address.
4. Run `astk` in your terminal. Voilà
Last edited by tianyikillua (2019-09-19 12:46:55)
Offline
I stuck in a problem while trying to run a file from my folder in Docker in windows
I have mad E: as Shared Drive
my working file is in E:/CODEASTER/example
C:\Users\praf_>docker run -ti --rm -v $%cd%:/home/aster/shared -w /home/aster/shared quay.io/tianyikillua/code_aster
docker: Error response from daemon: Mount denied:
The source path "$C:\\Users\\praf_"
is not a valid Windows path.
Last edited by praful (2019-09-19 08:52:33)
Prafulla Malla
Offline
1. Go to your working folder
cd /d E:/CODEASTER/example
2. Open Docker (you also accidentally added a $ before %cd%)
docker run -ti --rm -v %cd%:/home/aster/shared -w /home/aster/shared quay.io/tianyikillua/code_aster
Offline
Lot of thanks to you Titany. I was trying whole day but in vain.
when you don't know the solution, its like climbing Everest and, when there is solution, its too simple
2nd problem but not so important. I got problem while trying to run GUI. I uesd vcXsrc
aster@0f313077e016:~/shared$ astk
aster@0f313077e016:~/shared$ application-specific initialization failed: couldn't connect to display "[192.168.44.176]:0"
Error in startup script: invalid command name "wm"
while executing
"wm withdraw ."
(file "/home/aster/aster/lib/astk/astk.tcl" line 28)
Last edited by praful (2019-09-19 11:05:25)
Prafulla Malla
Offline
Could you please remove the left and right brackets?
docker run -ti --rm -e DISPLAY=192.168.44.176:0 -v %cd%:/home/aster/shared -w /home/aster/shared quay.io/tianyikillua/code_aster
Offline
Thanks, finally the problem got solved. Also, it worked only with MobaXterm but not with vcXsrc
Prafulla Malla
Offline