JIM Job Submission For Monte Carlo Requests

• Introduction
• Prerequisites
• Monte Carlo Job Submission
• Job Description File
• Needed JIM Submission Info
• JIM Merge Jobs
• JIM Structured Jobs
• JIM McFarm Submission

A job submitted via JIM takes the path schematically illustrated below.

Jim_client site ==> Scheduler site ==> Execution site

• The user who submits a request to Samgrid via JIM must have a valid grid certificate. The certificate can be obtained from the DOE Grids Certificate Service page. To complete the form use PPDG as the affiliation and put the comment "Personal certificate for SAM-Grid activities for the DZero experiment". The contact info must be somebody that knows the requester and is known by the Certificate Authority (CA). Presumably the comment is enough for the CA to ask sam-oncall for help. Once the certificate is obtained one obtains from it the Globus certificate and private key which are then placed in the user's ~/.globus directory on the machine where request submission will occur. The grid certificate contains the user's grid identity known as the certificate subject. Also note that a requester with a valid kerberos principal automatically has a certificate from the FNAL KCA.
• The machine where the user will submit the request must have the jim_client FNAL product installed and properly tailored.
• The user's subject must be in the grid-mapfile of the execution sites where the jobs will run and the scheduler sites handling the requests. The subject in the grid-mapfile must be mapped to the specific account at the execution site that runs Samgrid jobs. A central grid-mapfile with everybody's authorized subject is presently maintained. Tools exist for site admins to pull an updated grid-mapfile to their site. To get in the central grid-mapfile send a request and your subject to sam-admin@fnal.gov.

• runjob_requestid = request Id to be processed
• jobfiles_dataset = name of dataset in SAM with needed runtime files
• minbias_dataset = name of overlap dataset in SAM for the digitized phase
• phase_dataset = name of input dataset from SAM (if needed)
  This is used to get generated files from SAM instead of running pythia, when the request specifies it.
• runjob_numevents = number of events requested
• d0_release_version = D0RunII release version
• station_name = execution site station name where job will run
  This is optional as a random site will be selected if this is not specified.

• Release, request Id, events, data input file
  Found in the request file of the request assigned by Queue.py. The request details can be gotten by clicking on the request Id on the MC Request page or with this command:
  $ sam get request details --request-id=13669 --dictfmt
• Execution sites
  The station names can be found at http://samgrid.fnal.gov:8080/list_of_resources.php?.
• What job files are available?
  Check with commands like:

  $ sam translate constraints --dim="file_name cardFile%"
  $ sam translate constraints --dim="file_name d0_p%.tar.gz"
  $ sam translate constraints --dim="file_name mc_runjob_v%.tar.gz"
  

  Create the jobfiles dataset with a command like:

  $ sam define dataset \
    --file_name=cardFile_v00-06-00.tar.gz,d0_p14.05.01.tar.gz,mc_runjob_v06-02-02-jim-01.tar.gz,MagField_v00-01-00.tar.gz \
    --defname=jms_set5 --group=dzero

• To submit the request with JDF file name sg-request-13669 and obtain a Global Job Id:

  $ setup sam
  $ setup jim_client
  $ grid-proxy-init -hours 200 #(this is optional: if not done and a kerberos ticket is present, the KCA will be used)
  $ samg submit sg-request-13669
  

• Web based monitoring tools
  List of schedulers at http://samgrid.fnal.gov:8080/list_of_schedulers.php?
• Click on scheduler configured for user jim_client site
  • Click on Request number of Global Job Id
    • SAM query of all files or data tier
      Total event count, file size, average file size, list of files
    • Can be done with sam commands
  • Click on Status of Global Job Id
    • Download output gives logs of stderr and stdout of subjobs (zipped and tar'ed)
    • For diagnostics by experts
• Email sent to submitter upon completion (+notify_user attribute)