schedule(F)

Format

# media entries
media drive k size [ size ... ] [ format ]
media drive d density size [ size ... ] [ format ]

# specification of backup level
[0-9] size savetime importance marker

# filesystem backup schedule
/dev/rfilesys cycle method

Description

This cyclical schedule (or ``cycle'') is a list of backup levels to perform (including no backup at all) and a pointer to the last-used element of that list. The pointer is advanced to the next element of the list on a regular basis (each time fsphoto is run, usually once per day), starting at the beginning each time it falls off the end. It is advanced, however, only on success -- the desired backup must have been successful.

Each entry in the file is on a separate line. Blank and comment lines (beginning with ``#'') may be placed anywhere. Several keywords are recognized:

site sitename

sitename is passed to fsave as a description to place on each tape label. Usually, sitename is the name of the company or a building number.

media drive k size [ size ... ] [ format ]

Device drive is a floppy disk or tape drive capable of handling volumes with any of the listed sizes (in kilobytes). If specified, format is the command used to format the described floppies. This also applies to standard cartridge tapes.

media drive d density size [ size ... ] [ format ]

Device drive is a density BPI magtape drive capable of handling tapes of any of the indicated sizes (in feet). As with floppy drives, format is the optional command used to format the described tape.

[0-9] size savetime importance marker

Specification of each backup level, as described in fsave(ADM). The defaults are:

# Level   Size    Savetime        Importance      Marker
  0       -       "1 year"        critical        none
  1       -       "3 months"      necessary       none
  2...7   -       "1 month"       important       none
  8       -       "2 weeks"       useful          none
  9       -       "1 week"        precautionary   none

All four fields must be specified. A size means to use the first size listed in the appropriate media sizes list.

/dev/rfilesys cycle method

The filesystem resident on device /dev/rfilesys is to be backed-up according to cycle, which is a space-separated list of backup levels (the digits 0 to 9, passed to backup), or the letter x, meaning no backup should occur.

A backup cycle must have at least one member, but it may be of any length. Different filesystems may have cycles of different lengths.

The method field defines the backup utility to be used; cpio works for both XENIX and UNIX filesystems, but xbackup works only on XENIX filesystems.

Here is the default schedule file:

# SYSTEM BACKUP SCHEDULE
site machinename

# Media Entries
#
# 96 tpi 1.2 MB floppy 0
media /dev/rfd096ds15 k 1200 format /dev/rfd096ds15
# 96 tpi 1.2 MB floppy 1
media /dev/rfd196ds15 k 1200 format /dev/rfd196ds15
# 135 tpi 1.44 MB floppy 0
media /dev/rfd0135ds18 k 1440 format /dev/rfd0135ds18
# 135 tpi 1.44 MB floppy 1
media /dev/rfd1135ds18 k 1440 format /dev/rfd1135ds18
# Cartridge tape 1
# media /dev/rct0 k 60000 125000 150000 tape erase
# Mini cartridge drive (10 MB)
# media /dev/rctmini k 8800 format /dev/rctmini
# Mini cartridge drive (20 MB)
# media /dev/rctmini k 17200 format /dev/rctmini
# Mini cartridge drive (40 MB)
# media /dev/rctmini k 37500 format /dev/rctmini
# 9-track tape drive
# media /dev/rmt0 d 1600 2400 1200 600
#

# Backup Descriptor Table
#        Backup   Vol.   Save for    Vitality        Label
#        level    size   how long    (importance)    marker
         0        -      "1 year"    critical        "a red sticker"
         1        -      "4 months"  necessary       "a yellow sticker"
         2        -      "3 weeks"   useful          "a blue sticker"
         3        -      "1 week"    precautionary   none
# Schedule table
#                 1 2 3 4 5    6 7 8 9 0    1 2 3 4 5    6 7 8 9 0
# Filesystem      M T W T F    M T W T F    M T W T F    M T W T F Method
/dev/rroot        0 3 3 3 3    2 3 3 3 3    1 3 3 3 3    2 3 3 3 3 cpio
# Alternative Schedule for systems with /u filesystems
# /dev/rroot      0 x 3 x 3    2 x 3 x 3    1 x 3 x 3    2 x 3 x 3 cpio
# /dev/ru         3 0 3 3 3    3 2 3 3 3    3 1 3 3 3    3 2 3 3 3 cpio

In the alternate example, /dev/rroot is backed-up using a level 0 backup the first time fsphoto is run (on a Monday), and if that backup is successful, the next (second) time it runs (Tuesday), no backup is performed. If doing nothing is successful, the third time it runs (Wednesday) a level 3 backup occurs. If that backup succeeds, no backup occurs the fourth time (Thursday), but the fifth time fsphoto is run (Friday), a level 3 backup is made.

Each time a successful backup at the specified level happens, the pointer advances so that the next run of fsphoto (on the next weekday) will do the next backup scheduled for that filesystem. If however, a backup fails (or is interrupted or postponed by the operator) the pointer is not advanced; hence, the next time fsphoto is attempted, the same level backup will again be tried so the sequence will not be broken (but the timing may be off).

The larger and more rapidly changing filesystem /dev/ru is backed up more frequently (each time fsphoto is run -- once a day -- instead of every other time), and the levels used are staggered to prevent having to perform two full-scale backups (like levels 0 or 1) of the large filesystems on the same day. The backup cycle period is also shorter, two weeks instead of four.

Limitations

It is not necessary to specify the name of the ``raw'' (/dev/r) device for each filesystem, but the backups are faster if this is done.

See also