BaBar Skimming Commands and Notes

From GridPP Wiki
Jump to: navigation, search


The Skimming Cycle

One Skim Super job corresponds to one input dataset and produces X streams (depending on the SkimCycle). One Merge Super job merges the inputs on 1 to n Skim Super Jobs and produces X output files (One for each stream).

After running BbkCreateTask to set up the initial tasks the basic skim and merge cycles are:

BbkCreateSkims
BbkSubmitSkims
BbkCheckSkims
BbkCreateMerges
BbkSubmitMerges
BbkCheckMerges
BbkTMLoad
BbkTMExport

These run independently apart from the fact that the Merge cycle obviously takes the output of the skim cycle as it's input.

for running on the Grid two new commands BbkTMGridJobStatus and BbkRetrieveGridJob are inserted between the BbkSubmit[Skim|Merge] and BbkCheck[Skim|Merge] commands to check the status of the Skim or Merge job on the grid and to recover the output to the local setup.

BaBar Grid Skimming Commands

BbkTaskManager Commands

BbkCreateTask

Takes the text tasak description file and creates a new task, loading the task settings into the database.

BbkEditTask

Updates and already created task.

BbkCreateSkims and BbkCreateMerges

Create the Skim and Merge SuperJobs respectively, the -n option on BbkCreateSkims creates that many skim super jobs not individual jobs. It appears to be non functional on the BbkCreateMerges command

BbkSubmitSkims and BbkSubmitMerges

Submits the individual skim or merge jobs to the batch system of choice.

BbkSubmitSkims -n 50 MAN-Task01-funny04-R24a1 
BbkSubmitMerges -n 10 MAN-Task01-funny04-R24a1

BbkCheckSkims and BbkCheckMerges

Checks the output of the individual jobs once the jobs has finised either marking it as successful or failed. Also used for reseting failed jobs for resubmission. BbkCheckMerges will also go back and reset the component skims of a merge job if the job failed because of missing input files

BbkCheckMerges MAN-Task01-funny04-R24a1
BbkCheckSkims MAN-Task01-funny04-R24a1

BbkTMLoad

Loads the infomation for the successfully completed merge jobs into the main bookkeepping database at SLAC

BbkTMExport

Exports the successful merge collections to SLAC.

BbkTMSprite

Automation tool, will create, submit and check skims then create, submit, check, load and export skims

BbkTMUser

Used to query the Task database, good for finding out detailed infomation and debugging

BbkTMGridTools Commands

BbkTMGridJobStatus

Used to check the status of skim or merge jobs submitted to the grid.

BbkTMGridJobStatus --check --print MAN-Task01-funny04-R24a1

BbkTMGridJobsStatusDaemon

Daemon that sits in the background updateing the status of skim and merge jobs submitted to the grid

BbkRetrieveGridJob

Once a skim job has runn on the grid the output files (needed to check the status) are stored on a remote server. BbkRetrieveGridJob copies the files back and puts them in the job directory so the Check commands can check the job. Can be run interactively against one or all completed jobs or as a daemon to retrieve jobs as the finish.

BbkRetrieveGridJob --temp /var/tmp <--background> MAN-Task01-funny04-R24a1 <JobID>