Backups are stored in repositories, so before you can create your first snapshots you need to create a destination repository that will host them.
For the purpose of this tutorial,
the repository will not be encrypted to avoid being prompted for a passphrase on each command.
For encrypted repositories,
simply drop the -no-encryption option and things will work exactly the same:
$ plakar create -no-encryption
$
When no path is provided,
the repository is created in ~/.plakar.
It is possible to create it elsewhere,
in which case the path has to be provided on the command line as such:
$ plakar on /tmp/plakar create -no-encryption
$
There are MANY operations that can be done within a repository,
but we’ll have a look at the main ones: push to create a snapshot, ls to list snapshots and pull to restore a snapshot.
Once a repository is created, backups can be stored in it by pushing snapshots:
$ plakar push /bin
$ plakar push /bin
$ plakar push /bin
$
The push commands will scan /bin and store content in the repository,
performing deduplication to avoid storing redundant data:
the second and third pushes are much faster and do not cause the repository to grow much as they only record structure and metadata informations.
The snapshots are completely independant, it is possible to delete the first one without affecting the two others even though they relied on the same deduplicated data.
The content of a repository can be listed with ls:
$ plakar ls
2022-04-08T00:27:36Z 7ffa415c 13 MB /bin
2022-04-08T00:27:37Z 0fe4c857 13 MB /bin
2022-04-08T00:27:38Z 66052edc 13 MB /bin
$
The output above shows the three snapshots pushed above as well as a few informations regarding them.
Any snapshot can be restored with the pull command:
$ plakar pull 7ffa415c
$
The command above recreates the structure of the snapshot,
in this case a bin directory filled with underlying files:
$ ls -l bin/
total 25072
-rw-r--r-- 1 gilles staff 150624 8 Apr 02:32 [
-rw-r--r-- 1 gilles staff 1326576 8 Apr 02:32 bash
-rw-r--r-- 1 gilles staff 151792 8 Apr 02:32 cat
-rw-r--r-- 1 gilles staff 136960 8 Apr 02:32 chmod
-rw-r--r-- 1 gilles staff 152672 8 Apr 02:32 cp
-rw-r--r-- 1 gilles staff 1153456 8 Apr 02:32 csh
-rw-r--r-- 1 gilles staff 307296 8 Apr 02:32 dash
-rw-r--r-- 1 gilles staff 168112 8 Apr 02:32 date
-rwxr-xr-x 1 gilles staff 185088 8 Apr 02:32 dd
-rw-r--r-- 1 gilles staff 151264 8 Apr 02:32 df
-rw-r--r-- 1 gilles staff 150336 8 Apr 02:32 echo
-rw-r--r-- 1 gilles staff 235296 8 Apr 02:32 ed
-rw-r--r-- 1 gilles staff 151008 8 Apr 02:32 expr
-rw-r--r-- 1 gilles staff 150336 8 Apr 02:32 hostname
-rw-r--r-- 1 gilles staff 150736 8 Apr 02:32 kill
-rw-r--r-- 1 gilles staff 2599008 8 Apr 02:32 ksh
-rw-r--r-- 1 gilles staff 360752 8 Apr 02:32 launchctl
-rw-r--r-- 1 gilles staff 134736 8 Apr 02:32 link
-rw-r--r-- 1 gilles staff 134736 8 Apr 02:32 ln
-rw-r--r-- 1 gilles staff 187040 8 Apr 02:32 ls
-rw-r--r-- 1 gilles staff 134128 8 Apr 02:32 mkdir
-rw-r--r-- 1 gilles staff 135552 8 Apr 02:32 mv
-rw-r--r-- 1 gilles staff 320688 8 Apr 02:32 pax
-rw-r--r-- 1 gilles staff 203504 8 Apr 02:32 ps
-rw-r--r-- 1 gilles staff 150320 8 Apr 02:32 pwd
-rw-r--r-- 1 gilles staff 135408 8 Apr 02:32 rm
-r-xr-xr-x 1 gilles staff 133808 8 Apr 02:32 rmdir
-rw-r--r-- 1 gilles staff 150384 8 Apr 02:32 sh
-rw-r--r-- 1 gilles staff 150288 8 Apr 02:32 sleep
-rw-r--r-- 1 gilles staff 149936 8 Apr 02:32 sync
-rwxr-xr-x 1 gilles staff 1153456 8 Apr 02:32 tcsh
-rw-r--r-- 1 gilles staff 150624 8 Apr 02:32 test
-rwxr-xr-x 1 gilles staff 135408 8 Apr 02:32 unlink
-rw-r--r-- 1 gilles staff 150208 8 Apr 02:32 wait4path
-rw-r--r-- 1 gilles staff 1377760 8 Apr 02:32 zsh
$
Each snapshot is assigned an ID to allow referencing it in subcommands.
To make it easier for humans,
plakar uses prefix-based lookups so that the ID does not need to be typed entirely.
Whenever a snapshot ID is expected,
it is possible to provide the first characters and plakar will complete the missing part as long as the first characters do not resolve to multiple snapshots.
The following command:
$ plakar pull 7ffa415c
$
can be expressed as:
$ plakar pull 7ffa
$
or even:
$ plakar pull 7
$
because no other snapshot has an ID that begins with 7ffa or 7.
A lot of commands operate on files within snapshots.
Snapshots provide a filesystem-like interface which allows browsing content using pathnames that are relative to the snapshot ID.
They are expressed as ID:/pathname,
as can be seen in this command to list the informations regarding /bin/ls within snapshot identified by 7f:
$ plakar ls 7f:/bin/ls
2022-03-26T07:21:13Z -rwxr-xr-x root wheel 187 kB ls
$
It is possible to launch a read-only browser viewer on a repository, encrypted or not, using the following command:
$ plakar browser
lauching browser API at http://localhost:40717
If launching a browser does not work on your system,
simply use the -no-spawn option and point a browser at the displayed URL yourself.