ILCDIRAC for users

Getting Started: Registration, Installation, etc.

File access, downloading, uploading etc. with the DiracFileCatalog

First read the "Data Management" DIRAC-Tutorials.

Storage Elements

The following storage elements can be used as output location:
CERN-SRM (default)
DESY-SRM
RAL-SRM
KEK-SRM
IN2P3-SRM
PNNL-SRM
SLAC-SRM

We recommend you use a storage element that's close enough geographically to you.

Introduction to the ILCDirac interface to the LC Software Applications

Available Software Versions

Use the script: dirac-ilc-show-software that dumps the current available software. Please look at this website: DiracAvailableSoftware

Run jobs with your own libraries (e.g. Marlin Processor)

Getting the output

There are several ways of getting the job outputs: if you have 1 job, if you have a job repository, and if you use OutputData, things are different. In case of 1 job, use

dirac-wms-job-get-output JID

If you have a job repository repo.rep:

dirac-repo-retrieve-output -r repo.rep (-O)
The -O option gives you the possibility to download also the output data.

Finally, you can also download the job output from the web portal by left-clicking on a job, Sandobx, Get output file(s). You can also get the job input that way.

Warning, important The maximum size of an output sandbox is 10Mb. If the sum of the file sizes in there is bigger, the sandbox is not sent to the DIRAC sandbox store, but they are added to an archive send to the GRID. The Storage Element chosen depends on the site (usually a SE that's close to the site where the job ran). In this case, the retrieval from the web portal will fail with an error (missing sandbox), but the command line work as it retrieves the sandbox automatically from the GRID (not possible via the portal).

Monitoring the jobs, Webinterface

Warning, important Outdated, there is a completely different webinterfaces

The web monitoring of jobs happen here: https://voilcdiracwebapp.cern.ch/DIRAC/?view=tabs&theme=Grey&url_state=1|*DIRAC.JobMonitor.classes.JobMonitor: Connect to the ILCDIRAC monitoring page and you'll get something NOT (because the picture is outdated but somewhat similar) like the image below.

webMonitoring.png

The corresponding menu actions are:

  1. Main Menu: This menu offers options for systems, jobs, tools and help.
  2. Selections: Shows a set of selectors than permits generate customs selections.
  3. Buttons to open/collapse panels: Permit open or collapse left menu.
  4. Actions to perform for job(s): These actions permits select all, select none, reset, kill or submit
  5. Menu to change DIRAC setup: Users can change between different setups.
  6. Current location: Indicates where the user is located inside the portal.
  7. Buttons to submit or reset the form: After options are selected its possible to submit and execute the selection or reset the selectors.
  8. Pagination controls: Permits navigate between the pages, and also show in which page the user is navigating.
  9. Refresh table: Reload the page without loose the previous selection and show the new status.
  10. Items per page: This option allow the users to specify how many items are going to be displayed by page.
  11. User DIRAC login: Login assigned to the user connected to DIRAC web portal.
  12. DIRAC Group: The user could belong to different groups and perform actions depending of the group previously selected.
  13. Certificate DN: Web portal shows the distinguish name of user certificate what is being used to realize the connection.
  14. Index items displayed: Display the range of items displayed in the page.

Monitoring the jobs is also possible via a command line utility, dirac-wms-job-status JID, where JID is an individual job ID. If you use the job repository file, you can monitor all the jobs in there with

dirac-repo-monitor -r repo.rep

F.A.Q. and Frequently Encountered Problems

My Job Submission is Very Slow

If the job repository gets too large your job submission becomes very slow. Use different job repository files, for example name the repository file after your job group

jobGroup = "resonableDescription_take1"
dirac = DiracILC(True, jobGroup+".rep")

My jobs keep failing to upload output data

If an outputfile already exists on the grid, your job will not be able to overwrite it. You have to either delete your outputfiles before submitting your jobs again or use, for example, the jobgroup as a subfolder to differentiate different job groups Use the jobgroup to separate outputfiles in subdirectories.

jobGroup = "jetReco_take1"
...
job.setOutputData(["somefile1","somefile2"],"some/path/"+jobGroup,"CERN-SRM")
Change jobGroup whenever there is a new set of steering files, parameters or whatever to avoid trying to overwrite your outputfiles

If you no longer need a set of output files, please remove them from the storage.

I keep getting Could not verify credential errors

You see errors like:
[SE][GetSpaceTokens][] httpg://srm-public.cern.ch:8443/srm/managerv2: CGSI-gSOAP running on pclcd15.cern.ch reports Error initializing context
GSS Major Status: Authentication Failed

GSS Minor Status Error Chain:
globus_gsi_gssapi: SSLv3 handshake problems
globus_gsi_callback_module: Could not verify credential
globus_gsi_callback_module: Could not verify credential
globus_gsi_callback_module: Invalid CRL: The available CRL has expired
This means that the local copy of the certificate revocation list (CRL) expired. Please run dirac-admin-get-CAs to update you copy, or if you don;t have the right to run the command yourself, ask your local dirac administrator.

How do upload files to the Grid?

Have a look at the "Data Management" DIRAC-Tutorials.

How do I use this wonderful system?

Well, start by looking at the documentation at DiracForUsers and DiracByExample. Then contact ilcdirac-register@cernNOSPAMPLEASE.ch for more info.

I need to use my own processors

See also here: IlcdiracUserLibraries

It's fully taken in account in dirac. For that, you'll need to compile them against a version that dirac knows. And we defined a directory containing those version on afs under /afs/cern.ch/eng/clic/software/x86_64-slc5-gcc41/.

So simply setup the env, use cmake including the ilcsoft.cmake in the directory of your choice from the available ones, and put your processor/libraries in the proper directories as mentioned in the HeadFirstTalk above. And normally you should be going. We try to keep the version up to date, so you should be able to find always the latest release of ILCSOFT. We can also define new versions if needed, as long as one admin has access to the software binary location.

I want to specify an input file using for e.g. srm://srm-ilc.gridpp.rl.ac.uk/castor/ads.rl.ac.uk/ ?

You need to use the GetSRMFile class of ILCDIRAC as first job application. See DiracForUsers for more details, and the ilcdiracdoc page, UserJob and GetSRMFile specifically.

Running behind a proxy?

At RAL for example. I don't know what that changes. Need Marcel Stanitzki's advice.

Some commands in the web interface don't work?

The Get LogFile method is reserved to production jobs. It's never set for user jobs. It's normally pointing to a web location. To get the logs, make sure you put *.log in the OutputSandbox parameter of the UserJob.

How do I specify data coming from the Lcg File Catalog?

Most data is now copied in the DIRAC File Catalog browsable using dirac-dms-filecatalog-cli. A web view is also available (check the data menu in the interface). In case the file you are looking for is not in the DFC, you may use the GetSRMFile feature (above) given that you have some replica information..

How do I get the list of files that were uploaded on a SE by my jobs?

dirac-repo-create-lfn-list repository.rep
This will print on screen the list of files for each job, so you would probably want to redirect the output to a text file.

That command might take some time, depending on how many jobs there are.

How do I retrieve the uploaded output data of my jobs?

dirac-repo-retrieve-jobs-output-data repository.rep

How to check for replicas of a file

dirac-dms-lfn-replicas /ilc/...

Only works with files added in the DFC.

How to replicate a file

dirac-dms-replicate-lfn /ilc/your/file CERN-SRM 
dirac-dms-create-replication-request CERN-SRM <LFN|LocalFileWithLFNs> 

Or use the replicate command in the filecatalog-cli.

CERN-SRM is one of the storage elements configured. See above for list of other StorageElements.

Topic attachments
I Attachment History Action Size Date Who Comment
PNGpng webMonitoring.png r1 manage 421.1 K 2013-04-09 - 13:38 StephanePoss  
Edit | Attach | Watch | Print version | History: r61 < r60 < r59 < r58 < r57 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r60 - 2016-10-30 - AndreSailer
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CLIC All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback