Subject: .TRK/.CAL production update
Ioanni.Katsavounidis@lngs.infn.it
Date: Fri Dec 24 1999 - 10:38:39 EST
Ciao a tutti,
The following is a "user's manual" to the production of .TRK and .CAL
files, together with the rest of BMON's files, on the VAXGS ALPHA/VAX
cluster as of 24-DEC-1999.
There are 2 directories where .CAL and .TRK files are produced:
- DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL]
- DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL2]
The first one (TRKCAL) uses the ALPHA executables, as created by
Lynn Miller (regarding the .TRK production) and Alec Habig (regarding
the .CAL production). These executables are DREAM-based and use the
most up-to-date Scintillator and Streamer tube calibration databases
(in particular the geometry part that is essential to the .TRK production)
and also the most up-to-date tracking algorithm (which is also essential
to the .TRK production.)
The above mentioned factors can explain all differences with respect
to the .TRK files produced on VXMACB.
The .CAL production should not have any difference, since it is only
treating RAW ERP information. Yet, as Alec and myself have noticed, these
files are different. This was attributed (after various rounds of debugging)
to some events that are erronously given the wrong "calibration setting"
index *ON THE VXMACB VERSION* and anyways creates a negligible (4th decimal
digit or so) difference in the resulting calibration constants. In addition,
there is no duplication of the REF. ADC. info ("ERP box 90/590") for the
North and South face hits that now make part of SM1/SM6 hits.
The second directory (TRKCAL2) uses VAX executables to produce the
same .CAL and .TRK files, actually re-running the BMON code, as it is
done on VXMACB. In fact, the corresponding BMON.LOG and MAKETRK_RUNxxxxx.LOG
files from the execution of these jobs can be found under
DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL2.BMONITOR]
and the corresponding .LGB, ERP_STATUS, ERP_ERRORS and ??? under
DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL2.BMONITOR.LGB]
The .CAL and .TRK produced under this area should be binary identical
to the ones on VXMACB. I have checked ~ 20 runs or so (including many
calibration runs) and they resulted identical. They only thing I was
not able to check are the .LGB files. If anyone wants to try, just
define your logical to the above mentioned directory (and make sure
the LGB.COM doesn't re-assigns it...)
Regarding the logic that manages the two independent production streams,
there is just a single MAKETRK.COM file under DISK$MACROSCRA3:[MACROUSA0.
MONITOR.TRKCAL] that accepts 4 parameters, of which only the first is
obligatory (= input filename). The second one can be "RAW"/"FZEX"/"DST"
and refers to the type of the input file (defaults to RAW), the third
can be "OLD"/"NEW" and refers to the scintillator/S.T. databases used
(defaults to NEW) and finally the fourth can also be "OLD"/"NEW" and
applies only for the case the job is submitted to a VAX (*not* an ALPHA)
queue: in that case "OLD" means to use the VXMACB executable
(that can only produce .TRK files) or the DREAM-produced one (defaults
to "NEW".) Note that the executable opens 2 files at the beginning
(one .TRK and another .CAL) and determines which one to delete (or maybe
both) at the end based on the run type/file size.
Normally, this MAKETRK.COM file is launched by MR_MONITOR.COM for
a specific file, and since it is given the name "MAKETRK_MONITOR_RUN0xxxxx"
it makes part of QUEUECHECKER when determining when a run is processed.
For that reason, the resulting log files (located also under
DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL]) have the name mentioned
before.
The same MAKETRK.COM file is used to produce the DISK$MACROSCRA3:
[MACROUSA0.MONITOR.TRKCAL2] .CAL and .TRK files, except that the
parameters in this case are <filename>,"RAW","OLD","OLD". In fact
this is done by the execution of DISK$MACROSCRA3:[MACROUSA0.MONITOR.
TRKCAL2.BMONITOR]BMON.COM that
(1) First checks whether the file to be processed is already in the
list DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL2.BMONITOR]STABMON.DAT
(2) Produces all ERP_ERRORS,ERP_STATUS and ??? files under
DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL2.BMONITOR.LGB]
and also any possible (if there are calibration events) .CAL file
under DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL2]
Note that all ERP_ERRORS,ERP_STATUS and ??? files are
also mailed to the .DIS files located under
DISK$MACROSCRA3:[MACROUSA0.MONITOR.TRKCAL2.BMONITOR.LGB]
For the time being, they are only mailed to KATSAVOU, and
(by CC to MACROUSA.)
(3) If it finds tracks, launches MAKETRK.COM on WSGS07_BATCH with the
above mentioned arguments and thus produces the .TRK file.
(4) Adds the file processed to the STABMON.DAT list
Note that BMON.COM is submitted by CL_MONITOR.COM for each run.
BMON at first checks whether another BMON by MACROUSA is already running
on WSGS06_BATCH; in this case it just exits. It then checks the "standard"
locations where new .DAT files may be present (ALL disk$macrogs locations)
and processes them one by one.
CL_MONITOR.COM invokes RAWDAT_MON.COM to determine whether it's OK
to delete a given file - part of RAWDAT_MON's logic is to check
STABMON.DAT before deleting a file.
Hope the above is not too complicated. If you find any problems
or for any suggestions or feedback in general just mail me.
A presto,
Ioannis
This archive was generated by hypermail 2a24 : Fri Dec 24 1999 - 10:38:42 EST