Introduction

Search

Getting Started

Overview Generate fcl Upload fcl Prestage data Submit jobs Postprocess

Introduction to Mu2e grid jobs

Mu2e uses the jobsub_client package provided by Fermilab Computing Division to run jobs on the grid. Therefore general information about running grid jobs and jobsub applies.

The mu2egrid package provides Mu2e-specific code required for using jobsub. All scripts in the package support the --help option. If it is present on the command line, all other options will be ignored. All scripts also support --dry-run and --verbose options to show what will be done without performing the action. A basic support for G4beamline and MARS jobs is described here. The rest of this documentation focuses on running framework (art) jobs.

Please consider the following points before running grid jobs:

• Is there a standard dataset that you can use instead of starting from scratch? Some are listed here.
• The length of the job. It is recommended to have jobs shorter than about 8 hours but longer than 15 minutes.
• The total size of its input and output files. Worker nodes at Fermilab have 30GB of disk space per core.
• The size of those outputs files that are intended for tape storage. It is inefficient to store small files on tape. Good file sizes are from a few hundred MB to a few GB. If jobs are too slow to produce sufficiently large files, one should consider concatenating outputs before writing them to tape.
• Memory required to run the job. See Estimating per-job resources.
• If the current job writes out framework files for use by subsequent job stages, the above criteria should be applied to analyze the whole processing chain. It is important to remember that later stages can easily read multiple input files per job, but can not "split" an existing art file. For example, the configuration to produce standard g4s4 conversion electron datasets cnf.mu2e.....fcl runs only 1000 events per job, resulting in short (XXX minutes) jobs that produce small (XXX MB) output files. However digi+tracking jobs with background mixing would be too slow for singificantly larger g4s4 inputs.
• Before submitting any jobs to the grid, make sure that tape-based input datasets, if any, are pre-staged to disk.

To run some quick tests it is sufficient to

1. Generate a set of fcl files.
2. Submit mu2eprodsys grid jobs.

1. Generate a set of fcl files that completely define the jobs to be run.
2. Register the fcl dataset with SAM, and make fcl files available on /pnfs.
3. Make sure that input data files, if any, are pre-staged to disk.
4. Submit mu2eprodsys grid jobs.
5. Check results, identify and re-run failed jobs.
6. Store outputs on tape:

Next: generate fcl files.

Security, Privacy, Legal