FLUKA
Summary
This guide illustrates how to submit a FLUKA parametric job using IGP (Italian Grid Portal).
Preliminary Steps
It is mandatory a valid X509 certificate and VO GRIDIT membership.
We assume that you are familiar with concepts like certificate proxies, job submission, storage elements. If this is not the case, better read this
introduction. Other than that, you don't need to access an User Interface, since your grid interactions will be performed through IGP. Of course you will need to be registered to it.
Each portal user has a personal storage area, accessed through IGP Data Management (DM). The storage area path has a common prefix (
/grid/gridit/spes_project/fluka) and a suffix depending from the user name (/username). Users might create directory and files under this path.
This personal storage area will be used to load input files, using an archive, either plain (extension
tar) or compressed (extension will be
tgz). The archive must necessarily contain FLUKA input files, according to the following convention:
archiveName_N.inp
with N positive integer. FLUKA input files have also to be consecutive within the same archive. For instance, if your archive is called
silicon2, the archive should be looking as follows:
> tar tfz silicon2.tgz
silicon2_1.inp
silicon2_2.inp
silicon2_3.inp
silicon2_4.inp
silicon2_5.inp
On your own machine, you can create in advance the archive, that will be eventually uploaded to IGP DM. To create locally the archive you should copy on a directory all the input files (already renamed properly, with the archive name and consecutive indexes), then enter the directory and then run the command
> tar cf archiveName.tar archiveName_*
or
> tar zcf archiveName.tgz archiveName_*
depending if you wish to create a plain or a compressed archive.
Please make sure that the chosen archive name (
silicon2 in the example above) is the same for each
inp file, and it corresponds also to the archive file name.
Input file upload
Once created locally (on your own machine) the archive with the input files, you can upload it to IGP, with the
DM tools accessible in the
Storage section.
Change current path to the common
/grid/gridit/spes_project/fluka and then to the personal directory within. If the input directory does not exist, create it.
Then click on Upload, which prompts for the archive file selection. As target, you can leave
Random Destination, or select a Storage Element you trust. Then press start to trigger the transfer, and wait for the
Transfer OK message.
At the end you should see the archive file in the path you have specified.
Parametric job submission
You are now ready to submit a FLUKA job to the grid. A FLUKA job is implemented as a
Parametric job, i.e. a compound job made of N simple jobs, each of them identified by an integer. Each of the jobs differ from the other for few
parameters, like the index, the input or output files, etc..
To submit such a job, move to the
Job section in IGP. Although FLUKA users have not a dedicated graphic interface, it is available a template which simplifies the submission procedure. Click on
Manage Job Template, then
Job Template and you will get
Job Template Fluka. For the actual submission, still in the
Job section, click on
Submit New Job and then
Use Template. This will redirect you to the list of template, either personal or shared with other users. Click on the
Action button corresponding to
Job_Template_Fluka and then the button
Edit. This shows up the template and allows you to customize it
The following customization is suggested:
- Change Job Name as desired
- Leave unchanged Executable
- Modify Arguments like this:
- first argument: %s, don't change it, indicates to the portal the parametric jobs order
- second argument: change it to the archive name being used (silicon2 in the example before)
- third argument: /alex/silicon2/, change it to the remote path in DM where you uploaded the archive
- fourth argument: leave it blank to store the output files in the same path specified with third arguments, insert a valid remote path to store the output on a different location
- Choose GRIDIT as VO
- Standard Output : here the value %s returns by default the index of actual job part of the parametric submission. Change if you want a more meaningful output file name
- Standard Error : here the value %s returns by default the index of actual job part of the parametric submission. Change if you want a more meaningful error file name
- Change the Parameters attribute, inserting the number N of jobs being submitted (it must correspond to the highest index in the input files contained in the archive)
- Parameter Start and Parameter Step can be left unchanged, although IGP supports custom values for these attributes.
The template obtained can be saved as a new template, or it can be used for the submission to the grid.
Note
- If your personal certificate is expired, IGP will ask you to load a new valid certificate. Once uploaded, you need to wait some minutes before a new job submission
- Changes to other parameters is considered an advanced topic and it's not being discussed here.
Job Monitor
From the
Job section you can check the status of your submitted jobs
- Waiting: IGP has accepted the job for the submission
- Running: job is being executed in the grid site specified from the Site field
- Failed: job execution is failed
- Done: the job has been correctly executed
Tips
- IGP considers a job Done when there are no failures due to IGP itself or the grid. For instance, if the execution fails because of a wrong input, but exits correctly to the the operating system, the grid, and therefore IGP, have no way to detect the failure, which is only at the application level.
- Single jobs, or multiple selections, can be submitted again or canceled just selecting the job and applying for the desired option.
- Clicking on the Action button in the row representing the job, and then Get Output, you can check the job Standard Output or Standard Error if available
- In IGP section My Data*-> *Advanced Settings you can enable mail notifications related to job status change, and proxy expiration.
Output retrieve
If everything went fine, the job will assume the
Done status. Move to the IGP section
Storage and you will see the job output files in the directory you have specified customizing the Fluka job template file
For each job there is a tar, following this convention for the name :
archiveName_N_out_MMDDYY-HHMMSS-10_digits_UTC.tgz
This should prevent conflicts resulting from one or more files having the same name, for instance it would happen with repeated submission of the same, identical, job template.
Selecting an output file and choosing the
Download action from the pop up, you will download the job output files to your machine. Multiple selection is allowed, but this may increase the time needed for download, consider it if there are huge output files.
--
EmidioGiorgio - 2013-11-05