Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.atnf.csiro.au/people/pulsar/pmsurv/docs/parkes_birdiehunt.txt Дата изменения: Fri Oct 4 11:55:20 2002 Дата индексирования: Wed Dec 26 01:37:56 2007 Кодировка: Поисковые слова: jet

Written by: Froney Crawford Email: crawford@mit.edu
Created: July, 20, 1998 Location: pmsurv/docs/parkes_birdiehunt.txt
Updated: December 1, 1998

REVISED VERSION BASED ON HANDWRITTEN NOTES ON SCRIPT IN PARKES:
Nov, 19th, mk


Use the pulsar account for the processing.

This assumes the DLT device is /dev/rmt/1hn -- check that this is the
attached device on Orion as it may have changed. In any case, check
pmenv.pl (at Parkes!) to make sure it will read /dev/rmt/1hn. If you
change pmenv.pl, be sure remake it. Also make sure $PM_HOME in pmlogin
points to the proper Parkes directory.

DO NOT use the pmproc java interface as this will read whats going on
at Epping, not Parkes.

Check the disk space in $PM_HOME/bdhunt to make sure that > 2.0
GB are available for the birdiehunting.

The bdhunt directory is located on ORION_4.

[NOTE: In the old write-up you changed to ~/pm_home_orion/bdhunt which
is located on ORION_1, ie. a local disk on orion. Perseus is the only
other machine which can read this dir, so we are restricted to using
orion and two preseus processors for the birdiehunting. Perseus can
see this disk via a temporary mount, meaning that every time perseus
crashes, the disk must be remounted. Be sure to talk to Gina Spratt
about this is you find you cannot access the bdhunt dir from perseus
(from orion you should always be able to access the bdhunt dir).
It works however with the setup using the ORION_4 disk!!]

Setup:

Orion = master machine (with DLT) for reading data from tape and for
doling jobs to slaves
Orion = slave machine (1 processor)
Perseus = slave machine (2 processors, use perseus0 and perseus1)

0) Login as "pulsar"
Make sure crontab is running on each machine you want to use. Check
using "crontab -l". If the PM crontab is not runnning, type
"crontab $PSRHOME/bin/sun4sol/pmcron"
You may have to update data dir here to if necessary

1) In orion window (as pulsar):
Insert DLT and rewind.
"cd $PM_HOME/bdhunt"
remove exit_status_sc_td if it is there.
make digitizer.offsets file:
"sc_td -u 1n -D -n 1 -c 3.5 TAPE" (make sure first file on tape is a grid ptg!)
copy relevant line to file "digitizer.offsets"
rewind the tape and remove exit_status_sc_td
"echo TAPE >! $PM_HOME/conductor/tape_label.txt" (e.g. "echo PM0100 >! ...")
"touch $PM_CONDUCTOR/slave_names.txt"
"touch $PM_CONDUCTOR/bigsister.data"
"touch $PM_HOME/bdhunt/pm_bad_pointings"
"pmtape b"

(This will start the tape read processing. If it stops without doing
anything, just type pmtape b again to remove the last stop command.
That should work.)

2) In another orion window:
"cd $PM_CONDUCTOR"
"pmchief_hunter"

(This starts up the birdie hunter master, which doles out jobs to
slaves.)

3) In a slave window (e.g., perseus):
make sure crontab is running.
"cd $PM_HOME"
"mkdir SLAVENAME_slave" (e.g., sagitta_slave)
"cd SLAVENAME_slave"
If dir bird doesnt exist, "mkdir bird"
"cd bird"
"pmbirdslave SLAVENAME $PM_HOME/SLAVENAME_slave/bird/" (must include
last slash in dir name)

(This starts up a slave which will do the actual birdie
hunting. Repeat for other slaves.)

4) To stop processing: Ctrl-C the various windows.

5) After birdie hunting:
For some reason the birdie summary is not done automatically here.
Do this by hand.
"cd $PM_HOME/bdhunt"
"pmbird TAPE NSTART NEND" where NSTART and NEND are start and
end data files.

(This produces two plots, one is beam mode and one in pointing mode
which you can print.)

6) Make a storage dir for the resulting files:
"mkdir TAPE"
"mv TAPE_* TAPE/."
"mv digitizer.offsets TAPE/."
"mv TAPE.pmbs TAPE/."

Rewind unload the DLT. Done!

If you need to redo the birdiehunt, be sure to remove
bird_allocated_beams.log in $PM_CONDUCTOR.

Note:
PM software to keep updated in /psr1/soft/search/pmfind/ (at Parkes):
pmauto/ (scripts only)
pmbirdie/ (all)
pmcommon/