Production client status Job submission Production client Production
Production client status • Job submission • Production client • Production worker scripts • Status Andrei Tsaregorodtsev, CPPM, Marseille
Job submission schema CERN Job scripts Production service <XML-RPC> Production center > <X Production client Castor Bookkeeping service <XML-RPC> Storage Bookkeeping DB Monitoring service PC -R ML Data Production DB Bookkeeping XML files <BBFTP>
Client operations 1. Check the availability of the CPU resources (queue); 2. Check …/job/To. Do directory for pending tasks: submit tasks formulated in the old production if any; 3. Request job script(s) from the Production Server; 4. Check for the necessary software, configuration files: install the lacking software if necessary; 5. Submit the jobs obtained, report to the Production Server; 6. Check …/bookkeeping/To. Do directory for bookkeeping update files: - update the bookkeeping database; - initiate bbftp transfer of data files to Castor; - accomplish the bookkeeping update if transfer successful
Job description (1) A *. csh file: setenv SUBMIT_DATE 2002 -10 -09 setenv SUBMIT_TIME 17: 42 setenv PRODUCTION_IDENTIFIER setenv JOB_NAME 400000067 400000370 Production and job identifiers To be fixed ! # Parameters of step Min. Bias. Dec. 2002 setenv MBIAS_QQ_USER_FILE None setenv MBIAS_EVTYPE 61 # ================================ Signalpart description # Parameters of step B 2 pipi. Dec. 2002 setenv B 2 PIPI_QQ_USER_FILE B 2 pipi. dec setenv B 2 PIPI_EVTYPE 411300 # ================================
Job description (2) # Parameters of work flow B 2 pipi. Dec. 2002 setenv SIMULATION_APP SICBMC setenv SIMULATION_VSN v 254 r 2 setenv RECONSTRUCTION_APP Brunel setenv RECONSTRUCTION_VSN v 13 r 1 setenv DBASE_PKG dbase setenv DBASE_VSN v 248 r 4 setenv XMLDDDB_PKG Xml. DDDB setenv XMLDDDB_VSN v 8 r 1 Applications and other packages versions setenv QQ_USER_FILE ${MBIAS_QQ_USER_FILE}, ${B 2 PIPI_QQ_USER_FILE} setenv mcevtype ${MBIAS_EVTYPE}, ${B 2 PIPI_EVTYPE} setenv SPILLOVER on setenv mcmonihisto on # ( optional ) statistics, 500 by default setenv mcstat 500 # ( optional ) output file size, 1000 and 1000 by default setenv mcsize 1000 setenv dstsize 1000 # ================================ A list of different signal datasets to be produced with the same spill-over events Spill-over flag Monitoring histo flag Optional parameters
Production worker scripts • The job scripts are interpreted on a worker node by a set of production scripts: • getting input data; • invoking executables for each step; • saving output datasets and log files; • checking the log files; • preparing bookkeeping update scripts; • reporting job status to PVSS and “new” monitoring services; • The current production scripts are used with the Production Client: • Should we retain compatibility with the whole current production procedure or just the bookkeeping part ?
Client status • Implemented in Python; • Code is functional, some parts to be added: – Logging; – Simple local production manager tool • Tests are being done: – New job script format; – Compatibility with the old production schema; • Proper packaging is to be done; • Easy installation (rpm like) is to be done;
- Slides: 7