February alignment - Krisztian

Email from Roberto about our resposibility and out job: email

Steps:

cd /afs/cern.ch/cms/CAF/CMSALCA/ALCA_TRACKERALIGN/data/commonValidation/CMSSW_5_1_1_patch2/
source /afs/cern.ch/user/k/krajczar/public/TrackerAlignment_setup_env_2.csh
cd /afs/cern.ch/cms/CAF/CMSALCA/ALCA_TRACKERALIGN/MP/MPproduction
./setupNewAlignment.sh

This results in the mp1055 directory. I take my setup used for aligning data from mp1029 and copied here. I also copied here the example 0 T config file from mp1046 (see Roberto's above-mentioned email).

cp ../mp1046/alignment_cosmics_0T.py

Observation: in this config file the the alignment database is overwritten by hand! In earlier config files we used (for instance, alignment_isolated_mu.py) this was not done. As far as understand the comment in the config file, this manual overwriting was done to get around some problem with ESPrefer. Should be checked with Gero! Does it mean that the startgeomerty.txt or the alignables.txt has no effect now, and the starting geometries are to be adjusted directly in the config file?

For the alignment of the 4 T data this will needed to be modified. After these steps I removed the unnecessary files from the directory and made the following modifications to the remaining files:

- several things in setup_align.py (the location of the MillePede scripts, endpoint directories, the config file to be run, global tag, json file)

Check with Gero: global tag (I use GR_P_V28)? JSON?

Test 0T dataset: /Cosmics/Commissioning12-TkAlCosmics0T-v1/ALCARECO

To produce the list of ROOT files corresponding to the above-mentioned dataset do:

dbs search --query='find file where dataset=/Cosmics/Commissioning12-TkAlCosmics0T-v1/ALCARECO and file.numevents>0' | grep -i store > bla.txt

Add "CastorPool=cmscaf" to the beginning of bla.txt to create the final input file list for the setup_align.py script.

I made a simply JSON file, which accepts everyting, and fed it setup_align.py

Run Mille and then Pede by:

perl setup_align.pl
mps_fire.pl 1000
mps_stat.pl
mps_retry.pl FAIL
mps_fetch.pl

mps_fire.pl -mf

The Mille jobs are reported to have failed: "gzip -f jobData/job016/STDOUTsend2nsd: NS009 - fatal configuration error: Host unknown: cnsuser.caf/store/caf/user/krajczar/MPproduction/mp1055: Host not knownmilleBinary???.dat file not found or emptyjob016 217956410 -------- FAIL".

However, the /store/caf/user/krajczar/MPproduction/mp1055 directory seem to contain all files, so it seems that Mille run successfully (?). However, the blindly started Pede jobs fails too: "zip -f jobData/jobm/STDOUTjobm 217962914 Command not foundjobm 217962914 Pede did not end normallyjobm 217962914 -------- FAIL". The jobm/STDOUT says that "/afs/cern.ch/cms/CAF/CMSALCA/ALCA_TRACKERALIGN/MP/MPproduction/mp1055/jobData/jobm/theScript.sh:112: command not found: mps_merge_millepedemonitor.pl".