Title: Alex Sousa-MINOS Meeting
1The MINOS Farm Processing Infrastructure
Work done in collaboration with Howie, Liz and
Jon.
Overview
- Develop an infrastructure for the MINOS
Production Data Processing - Automate the Production Data Processing setting
up for two modes of operation - Run through the full data sample
- Run on data files as they become available
- System to be setup using the Fermilab Fixed
Target Farm (fnsfo) NuMi, KTeV,
HyperCP(E871)...
http//mixcoatl.phy.tufts.edu/asousa/production/pr
oduction.html
2The MINOS Farm Processing Infrastructure
The Farm Processing Block Diagram
3The MINOS Farm Processing Infrastructure
A simple Farm model
CPU
DISK
NETWORK
FBSNG
http//www-isd.fnal.gov/fbsng/
4The MINOS Farm Processing Infrastructure
Farms Batch System Next Generation (FBSNG)
- Job Control
- Submission
- Monitoring
- Killing/Canceling
- Holding/Releasing
- Scheduler
- Guaranteed scheduling
- Load Balancing
- Job queue priority changes dynamically as its
sections or sections of other jobs start/complete
- Application Programmers Interface (API)
- Allows development of custom tools for job
monitoring/management (Python binding)
5The MINOS Farm Processing Infrastructure
FBSNG Job is submitted via a Job Description File
(JDF), e.g.
SECTION input_phase EXECinput.exe arg1
arg2 QUEUEinput_queue NUMPROC1
SECTION main EXECanalysis.exe
arg1 arg2 arg3 QUEUEmain_queue
NUMPROC5 DEPENDstarted(input_phase)
STDERRJ.startup.err
STDOUTJ.startup.out NEED1 SECTION
output_phase EXECoutput.exe arg1 arg2
arg3 arg4 QUEUEoutput_queue
DEPENDdone(input_phase) done(main)
NUMPROC1
/gt fbs submit test.jdf /gt fbs status -q
queue (ltjobidgtltsectionidgt) /gt fbs
kill -s section (ltjobidgtltsectionidgt)
6The MINOS Farm Processing Infrastructure
FBSNG Web Monitoring
(http//www-isd.fnal.gov/cgi-bin/fbsng/fbswww/fbsw
ww)
List of Queues
CPU Utilization by Project
7The MINOS Farm Processing Infrastructure
Plans for the Immediate Future
- Start running test scripts on the WorkerNuMi
nodes - Study and implement the automation of the Farm
Processing Infrastructure. - Obtain input from the Reconstruction Group and
Software Group on - What production job to run.
- What ntuples to produce.
- What calibration constants to use.
- ...