INSTALLATION OF THE PRODUCTION RELEASE PMC02.01
 

IT 'S DONE WITH THE TOOL BUILDREL WITHOUT PROBLEMS. HOWEVER WE OBTAIN A SLIGHTLY
DIFFERENT SIZE FOR THE EXECUTABLE ON LYON  LINUX-PLATFORM
 
 

D0GSTAR2.X        LYON     46984599  MBytes

D0GSTAR2.X        FNAL     44447208   MBytes  ====> This executable will be used at LYON
 

 (AT LYON, WE ARE USING CERNLIB 99 ?)

 A VERY FAST COMPARAISON HAS BEEN DONE BY MICHEL JAFFRE THAT HAS SIMULATED  A

SAMPLE OF 20 EVENTS IN THE TWO FOLLOWING CHANNELS :
 

PROTON ANTIPROTON ----> TOP ANTI-TOP

AND

PROTON ANTIPROTON ----> JETS JETS  (ONE OF JET MUST HAVE A PT BETWEEN 20 AND 40 GEV)

WITH A NOISE A 2 EVTS  PROTON ANTIPROTON ----> JETS JETS  WITH LOW PT.
 

THE GEOMETRY OF THE CALORIMETER WAS MIXED.

THE CPU-TIME AND THE SIZE ARE COMPATIBLE WITH THE DATA GIVEN BY G.GRAHAM  AND
THE BASIS  MAY BE :
 

                            10 MINUTES IN REAL TIME AND 5 MBYTES FOR THE  EVENT  SIZE.
 

  SO  AT OUR COMPUTING CENTER , AT LYON, A REASONABLE PROFILE FOR A JOB BATCH  WOULD BE
  A JOB RUNNING  UP TO  THE  TIME  LIMIT  (17 H FOR THE MOMENT)  AND  SIMUlATING A SAMPLE OF 100 EVENTS WITH A TOTAL SIZE OF    500 MBYTES.

 THE DATA WILL BE WRITTEN ON REDWOOD MEDIA, IN AN INTERNAL TAPE FORMAT THEN REWRITTEN IN
ENSTORE-FORMAT ( SEE  BELOW)

  FOR INSTANCE,  ON THE BASIS OF 70000 EVENTS QCD,  THOSE PREVIOUS DATA  WOULD GIVE A FULL TIME OF 16   MONTHS  FOR ONE WORKER-CPU.   SO IF WE HAVE 16 WORKERS WE 'LL BE  ABLE TO DO THIS SIMUATION   IN A MONTH.
 

B/ THE MEANS OF THE C2IN2P3
 
 

FOR THE MOMENT THERE ARE 10 WORKERS PENTIUM II , 450 MHZ, 512MO,    RUNNING LINUX, REDHAT 5.2.

 AN ORDER OF 10 BI-PROC  PENTIUM III 450 MHZ, 512MO  HAS BEEN DONE AND THESE PC ARE EXPECTED AT THE  END OF OCTOBER.

HOWEVER, THE LINUX PLATFORM  AT LYON IS NOT RESERVED ONLY TO THE D0-PEOPLE IN FRANCE .
FOR THE MOMENT, FEW EXPERIENCES HAVE MIGRATED ON LINUX, BUT AS THE EXECUTABLE IS READY, WE

HAVE TO FILL THE  BATCH QUEUE ,   VERY FAST.

C/  HOW  WILL WE MANAGE THE PRODUCTION ?

 AT LYON, WE HAD HAVE A BASIC TOOL, WITHOUT GUI INTERFACE TO SUBMIT   AUTOMATICALLY A SET OF

JOBS. IT WAS USEFUL BUT  ON THE LAST FRIDAY  GREGORY GRAHAM GAVE A VIDEO-CONFERENCE SHOWING

THE MC_RUNJOB PACKAGE. IT IS A SET OF PYTHON TOOLS, WITH TEMPLATE FILES AND A GRAPHIC USER

INTERFACE IN    TCL/TK.  BESIDES  AN IMPORTANT THING IS THAT TOOL PRODUCES THE SAM OUTPUT,

MANDATORY TO EXPORT OUR DATA.
 

 SO WE HAVE INSTALLED IT, YESTERDAY : WE HAVE GOT THE FIRST SCREEN. WE'LL HAVE LOCAL

ADAPTATIONS TO USE OUR BATCH SYSTEM AND WE ARE VERY INTERESTED BY FURTHER DEVELOPMENTS.

WE THINK ABOUT A POSSIBILITY TO SUBMIT A RANGE OF JOBS GIVEN IN A JOB_LIST THROUGH A SINGLE  CLICK
 

D/  EXPORT PROCEDURES
 

AT THE BEGINNING OF THE WEEK, MONDAY AND TUESDAY, WE HAD A VERY PRODUCTIVE  MEETING , HOLD

AT LYON WITH LEE LUEKING AND THE EXPERTS OF THE COMPUTING CENTER.

FIRST, WE TALK ABOUT THE RESPECTIVE FORMAT OF TAPES ,AT FNAL  AND LYON.ANAND LYON.AN

AGREEMENT HAS BEEN FOUND.  LEE HAS GOT AN ACCOUNT, AT LYON, TO FACILITATE THE  UNDERSTANDING

OF  FURTHER PROBLEMS .

 SECONDLY, WE HAD A CONTACT WITH THE DATABASE-MANAGER AT LYON  CONCERNING   ORACLE

BECAUSE SAM SYSTEM USES THAT. AT LYON , ORACLE 7.0.X IS  IN USE AND  THE VERSION 8.0X IS NOT YET

INSTALLED AND CHECKED.  WE HOPE THAT WILL BE DONE AT THE END OF OCTOBER.  ON AN OTHER HAND A

CONTACT WILL BE ESTABLISHED BETWEEN  THE 2 DB , AT FNAL AND LYON.  IN ORDER TO MAKE EASY THE

ADAPTATIONS ,THE DB FNAL MANAGER WILL GET AN ACCOUNT ON DB-MACHINES .

 CONCERNING THE EXPORT PROCEDURE, SOME  DEVELOPMENT HAS TO BE DONE BY THE EXPERTS  OF

THE TAPE-SYSTEM AT LYON. THEY WILL TRY TO AVOID A DISK STAGING TO CONVERT THE MEDIA - IN

INTERNAL FORMAT - INTO THE D0 -ENSTORE FORMAT.

IT REMAINS  THE PROBLEM TO KEEP THE KNOWLEDGE OF WHAT IS ON TAPE AND WHERE  IS  THAT.

FOR THE MOMENT, WE HAVEN'T  ACCESS TO  A  TOOL  AS HPSS  TO LINK FILENAMES AND VOLUME ID AND FILE SEQUENCE.