amvault — Copy Amanda dumps from one volume to another
amvault
[-o
configoption
...] [-q] [--quiet] [-n] [--dry-run] [--fulls-only ] [--export ] [--src-timestamp
src-timestamp
]
--label-template
label-template
--dst-changer
dst-changer
[--autolabel
autolabel-arg
...]
config
[
hostname
[
disk
[
date
[
level
[
hostname
[...]
]
]
]
]
]
This application is not yet in its final form, and is subject to major revision in subsequent versions of Amanda. Backward compatibility is not guaranteed.
Note that Amanda restore/recover operations will request tertiary media by label when dumpfiles are not found on secondary media, but there is no provision to automatically fetch such media from a different changer
Feedback on and patches to this application are invited and encouraged!
Amvault is conceptually equivalent to "amfetchdump | taper". That is, it reads specified dumps from secondary media and re-writes them on tertiary media.
Amvault Copies data from the run with
timestamp src-timestamp
onto volumes using the changer
dst-changer
, and labeling new volumes with
label-template
. If src-timestamp
is
"latest", then the most recent amdump or amflush run will be used. If
--fulls-only
is given, then only full (level-0) dumps are
copied.
The --quiet
(-q
) option will
eliminate non-error messages, and is useful when running amvault from
cron. The --dry-run
(-n
) option will
cause amvault to print the dumps it would vault, but not actually perform
any vaulting operations.
The dumps to be read from secondary media can be specified by any
combination of dump specifications, --fulls-only
, and
--src-timestamp
. At least one must be specified, lest
amvault attempt to vault all dumps in the catalog. See amanda-match(7) for more information on dump
specifications.
Note that the datestamp given in the dumpspec is the
dump datestamp - the run in which the backup was taken
on the Amanda client. The --src-timestamp
, on the other
hand, is the write timestamp - the run in which the
dump was written to secondary media. The latter option facilitates
duplicating the results of an entire backup run, including any dumps that
might have been flushed from holding disk.
The --dst-changer
must be specified, and names the changer
in which tertiary media are stored. In general, this should be different
from the secondary changer, to eliminate the possibility of overwriting
secondary media with tertiary data.
The changer parameter should specify the name of a changer defined in amanda.conf(5). For example:
define changer vaulting_tape { tapedev "/dev/rmt/1n" tpchanger "chg-zd-mtx" changerdev "/dev/sg0" changerfile "vaulting-changer.conf" }
The --label-template
option is required, and specifies a
label template which is used to generate new labels for tertiary volumes.
The --autolabel
option works just like the
autolabel parameter in amanda.conf(5), and can be
specified multiple times if necessary. The default is 'empty'.
If amanda.conf(5) contains the new part-size splitting parameters, then amvault will use them without any additional configuration. However, if the configuration still uses the old splitting parameters (tape_splitsize, split_diskbuffer, and fallback_splitsize), then amvault will need some additional configuration in order to properly split dumps to tertiary media. To do so, specify a new tapetype in amanda.conf(5), say "TERTIARY", and set the part-size and other appropriate parameters there. Then reference that tapetype in the amvault invocation:
amvault -otapetype=TERTIARY ...
The --export
option will cause amvault to attempt to move
completed tertiary volumes to import/export slots, where they can be
more easily removed by an operator.
This manual page was written by
.