| CFHT
archive manual - processes |
 |
Archive Processes
The archive system generally consists of a pipeline of daemons which pass
data to each other, performing a process at each step. Each daemon employs
handlers which make the various software and hardware interactions generic.
The daemons are modular and the pipeline extendable so new processes and
pipelines can be added or removed rapidly and without interfering with
the existing data path.
The current daemons are:
Copyd
Distd
Tarexad
Stripheadd
Aiqed
Dbd
The Copy Daemon
Copyd is responsible for intelligently moving files from computers at the
summit to the archive host. It can check for files on any number of hosts
of any platform and copy files using a number of methods. When copying
cfh12k mosaics it concatenates the individual files into a singe MEF (Multi
Extension Fits) format as it copies the data. When copying Megacam data it
splices the per amplifier mosaics into per chip mosaics.
Parfile: /h/archive/sw/par/copy2d.par
Working directory: /data/kapu/copy2d
Associated utilities: mef runon
The Distributor Daemon
Distd receives data from copyd and distributes it to the tape daemons.
It also handles
the communication to and from the media-writing daemons when one reaches
the end of it's media. The distributor empties the archiving daemon's working
directories and gives a signal for the daemon to finish up once it is done
writing the current set. Once both daemons have changed their media, distd
repopulates their working directories with the files that it cleaned up
thus ensuring that each piece of archive media begins with the same file.
Parfile: /h/archive/sw/par/dist2d
Working directory: /data/kapu/dist2d
Associated utilites: change_auto_media
The Header Daemon
Stripheadd also gets data from the copy daemon and extracts information about each file from its fits header. It
catalogs a complete copy of the primary header internally as well as on
the ftp site. It also manages various lists that are used by other processes
and people. It maintains a sybase database which contains basic information
about each file and it manages the copy of each file to the distributed
storage system. It takes a list of the available filesystems and evaluates
which one has the most free space. It then copies the file to this storage
area and updates the database with the current location and creates a symbolic
link to the file in the virtual filesystem of archive data in /h/archive/current.
Parfile: /h/archive/sw/par/stripheadd.par
Working directory: /data/kapu/stripheadd
Associated utilities: sql_update extracth log_sheet_line.cfh12k.tcl
log_sheet_line
The Image Quality Daemon
Aiqued scans files for various features and maintains lists which are its
output.
Parfile: /h/archive/se/par/aiqued.par
Working directory: /data/kapu/aiqued
Associated utilities: IRAF
The Tar Daemon
Tarexad's name is actually a little confusing. Tarexa initially referred
to "tar exabyte" though it no longer writes exabyte tapes and it is no
longer necessary that it write data in tar format. At any rate, this daemon
is responsible for writing the archive data to media in a sensible format
and for re-reading the data and checking it for errors. It also updates
the sybase database with information about the location of each file on
tape for retrieval. Tarexa makes extensive use of handlers to keep its
function as administrative and device independent as possible. In order
to produce two copies of archive media, two copies of tarexad are concurrently
run, they communicate to each other via distd to keep their tapes identical.
parfile: /h/archive/sw/par/tarexa3d /h/archive/sw/par/tarexa4d
Working directory: /data/kapu/tarexa3 /data/kapu/tarexa4
Associated utilities: tarexa3h tarexa4h tarexa_init sql_update
The Database Daemon
This daemon is still part of the pipeline though it has been replaced by
calls to the database handler in both stripheadd and tarexad.
Parfile: /h/archive/sw/par/dbd.par
Working directory: /data/kapu/dbd
Kanoa
Last
modified: Tue Jul 17 14:36:21 HST 2001