Snapshots

A snapshot is a copy of the user, system and configuration data stored by snapd for one or more snaps on your system. This data can be found in $HOME/snap/<snap-name> and /var/snap/<snap-name>.

Snapshots are generated manually with the snap save command and automatically when a snap is removed (unless --purge is used with remove, in which case no snapshot is created). A snapshot can be used to backup the state of your snaps, revert snaps to a previous state and to restore a fresh snapd installation to a previously saved state.


Generating a snapshot

The snap save command creates a snapshot for all installed snaps, or if declared individually, specific snaps:

$ sudo snap save
Set  Snap         Age    Version               Rev   Size   Notes
30   core         1.00s  16-2.37~pre1          6229   250B  -
30   core18       886ms  18                    543    123B  -
30   go           483ms  1.10.7                3092   387B  -
30   vlc          529ms  3.0.6                 770   882kB  -

Each snapshot has a unique ID, or revision, shown in the Set column above. This value is unique to each save operation, regardless of the number of snaps it includes. Age is the period of time since the snapshot was created, while Version and Rev refer to the specific snap at the time of the snapshot. Size is the amount of storage used by a snapshot.

ⓘ If you’d rather not wait for the save operation to complete before regaining access to your terminal, add the --no-wait argument.

You can see the state of your system’s snapshots with the snap saved command. Adding --id=<set/unique ID> allows you to query a specific snapshot:

$ snap saved --id=29
Set  Snap             Age    Version               Rev   Size   Notes
29   vlc              2h41m  3.0.6                 770   882kB  -

ⓘ Both the saved and check-snapshot commands accept a –users= option with a comma-separated list of users to filter on.

What a snapshot stores

A snapshot is a copy of the user, system and configuration data stored by snapd for one or more snaps on your system. For each snap, this data can be found in $HOME/snap/<snap-name> and /var/snap/<snap-name>.

More specifically, these are locations that snapped application access through the following environment variables from within the snap:

System-wide locations:

  • SNAP_COMMON ( /var/snap/<snap-name>/common )
  • SNAP_DATA (/var/snap/<snap-name>/<revision>)

User-specific locations:

  • SNAP_USER_COMMON (/home/<username>/snap/<snap-name>/common)
  • SNAP_USER_DATA (/home/<username>/snap/<snap-name>/<revision>)

It’s important to note that SNAP_DATA and SNAP_USER_DATA place their data within a directory specific to each snap revision, whereas SNAP_COMMON and SNAP_USER_COMMON do not.

You need to be aware of what data is copied forward when you move from one revision to the next and which data will be restored if you switch to a previous snapshot.

When you move from one revision to the next, the revision-specific contents of SNAP_DATA and SNAP_USER_DATA are copied into new directories for the new revision. The contents of SNAP_COMMON and SNAP_USER_COMMON remain the same. A snapshot only includes the data for the installed revision, plus the contents of the two COMMON directories.

When a snapshot is restored:

  1. The contents of SNAP_COMMON and SNAP_USER_COMMON will be overwritten and restored to their state when the restored snapshot was created.

  2. The revision-specific contents of SNAP_DATA and SNAP_USER_DATA will be copied into and overwrite the contents of the revision-specific directory of the currently installed revision.

See Data locations for more details on how these locations are intended to be used by a snap, and see Inside a snapshot to see how they’re stored within a snapshot.

Verifying a snapshot

To verify the integrity of a snapshot, use the check-snapshot command:

$ sudo snap check-snapshot 30
Snapshot #30 verified successfully.

Exporting and importing a snapshot

By default, snapshots are maintained and stored on the system that created them. However, to help with backup and recovery, individual snapshots can also be exported and restored.

To export a snapshot, use the snap export-snapshot <set-id> <new-filename> command:

$ sudo snap export-snapshot 30 my-snapshot.zip
Exported snapshot #30 into "my-snapshot.zip"

The resultant snapshot file is a zip archive that contains two json files to validate the snapshot and a zip archive containing the user, system and configuration data for the specific revision of the snap installed when the snapshot was created (see Inside a snapshot for more details).

To import a previously exported snap shot, use the snap import-snapshot command:

$ sudo snap import-snapshot mysnapshot
Imported snapshot as #30
Set  Snap  Age    Version  Rev   Size    Notes
30   vlc   3d02h  1.11.13  4286  255B  -

If the snapshot with the same snapshot identifier exists, the import will overwrite it. If the snapshot doesn’t exist, it will be imported and assigned a new snapshot identifier.

Restoring a snapshot

The restore command replaces the current user, system and configuration data with the corresponding data from the specified snapshot:

$ sudo snap restore 30
Restored snapshot #30.

Before using this command, make sure you have the snap application installed, and that its services, if any, are stopped (snap stop <snap-name>). Start them again once the snapshot has been successfully restored.

By default, this command restores all the data for all the snaps in a snapshot. You can restore data for specific snaps by simply listing them after the command and for specific users with the --users=<usernames> argument.

Excluding a snap’s system and configuration data from snap restore is not currently possible.

Deleting a snapshot

The forget command deletes a snapshot. This operation removes a snapshot from local storage and can not be undone:

$ sudo snap forget 30
Snapshot #30 forgotten.
$ snap saved --id=30
No snapshots found.

By default, this command deletes all the data for all the snaps in a snapshot. You can delete the data for specific snaps by listing them after the command.

Automatic snapshots

Apart from on Ubuntu Core devices, where the feature is disabled by default, a snapshot is generated automatically when a snap is removed. These snapshots are retained for 31 days before being deleted automatically.

To see which snapshots are generated automatically, look for auto in the Notes column output from snap saved:

$ snap saved
Set  Snap              Age    Version               Rev   Size   Notes
30   go                25d5h  1.10.7                3092   387B  -
30   vlc               25d0h  3.0.6                 770   882kB  -
31   vlc               529ms  3.0.6                 770   882kB  auto

As with manual snapshots, automatically generated snapshots can be manually deleted with snap forget <set-id>.

Automatic snapshot retention time is configured with the snapshots.automatic.retention system option. The value needs to be greater than 24 hours:

$ snap set system snapshots.automatic.retention=30h

To disable automatic snapshots, set the retention time to no:

$ snap set system snapshots.automatic.retention=no

Disabling automatic snapshots will not affect pre-existing automatically generated snapshots, only those generated by the removal of subsequent snaps.

Automatic snapshots require snap version 2.39+.

Inside a snapshot

On Ubuntu-based systems, snapshots are stored in the /var/lib/snapd/snapshots directory and are both stored and exported as a zip file. This zip file contains the following:

<snap-snapshot-zip>
├── <snapshot-number>_<snap-name>_<revision>.zip
└────── archive.tgz
    │   ├── <revision-number>
    │   └── $SNAP_DATA (/var/snap/<snap-name>-<revision>)
    │   └── common
    │       └── $SNAP_COMMON (/var/snap/<snap-name>/common)
    ├── meta.json
    ├── meta.sha3_384
    └── user
        └── <username>.tgz
            ├── <revision-number>
            │   └── $SNAP_USER_DATA (/home/<username>/snap/<snap-name>/<revision>)
            └── common
                └── $SNAP_USER_COMMON(/home/<username>/<snap-name>)

  • meta.json: describes the contents of the snapshot, alongside its configuration and checksums for the archives.
1 Like

I might add that both subcommands “saved” and “check-snapshot” take the option “–users=” which accepts a comma-separated list of users to filter on.

1 Like

“Snapshots are generated manually with the snap save command and automatically when a snap is removed.” Parenthetically, unless you remove with the --purge option.

Thanks for the suggestion. I’ve updated the text.

What happens if I

  • generate a snapshot S for snap X revision 1 in stable,
  • refresh the snap X to revision 3 in channel edge, and
  • restore the snapshot S

? Will revision 1 replace revision 3, or will revision 3 remain but now with potentially incompatible data files? Will the tracked channel change?

@nteodosio Let’s find out!

First, let’s install a snap that has different revisions available to install, and that’ll create some files in ~/snap. The nethack snap comes to mind, as it’ll also be reasonably small to play around with:

~$ snap info nethack 
name:      nethack
summary:   The popular nethack console adventure
publisher: Oliver Grawert (ogra)
store-url: https://snapcraft.io/nethack
contact:   ogra@ubuntu.com?subject=nethack
license:   Other Open Source
description: |
  Play the nethack cli version
snap-id: i2ba1vb7DvsIzb8R987xvPGMQWNHiARe
channels:
  latest/stable:    3.6.6 2020-09-07  (87) 20MB -
  latest/candidate: 3.6.2 2019-08-24  (79) 13MB -
  latest/beta:      3.6.2 2019-08-24  (79) 13MB -
  latest/edge:      3.6.6 2023-08-31 (113) 15MB -

so let’s install that:

~$ snap install nethack
nethack 3.6.6 from Oliver Grawert (ogra) installed

note you don’t have any files yet:

~$ find ./snap/ | grep nethack
~$ 

so let’s run it once, quit at the character creation screen, and look again

~$ nethack
# here nethack takes over the terminal and shows the character creation screen; press 'q' to quit.
~$ find ./snap/ | grep nethack
./snap/nethack
./snap/nethack/common
./snap/nethack/87
# a few more files in 87/ elided
./snap/nethack/87/.nethackrc
./snap/nethack/current

just to be more obvious, let’s drop our own file in there:

~$ touch ./snap/nethack/current/HELLO
~$ find ./snap/nethack | grep HELLO
./snap/nethack/87/HELLO

Now we take a snapshot:

~$ snap save nethack
Set  Snap     Age    Version  Rev  Size    Notes
1    nethack  183ms  3.6.6    87   1.09MB  -

So let’s refresh to edge:

~$ snap refresh nethack --edge
nethack (edge) 3.6.6 from Oliver Grawert (ogra) refreshed

note how our HELLO file has been copied over:

~$ find ./snap/nethack | grep HELLO
./snap/nethack/87/HELLO
./snap/nethack/113/HELLO

Note also that the current link isn’t yet pointing at the new revision until you run something from the snap as this user. So let’s do that.

~$ readlink snap/nethack/current
87
~$ nethack

~$ readlink snap/nethack/current
113

now, let’s pretend that nethack migrates HELLO to HI somehow:

~$ mv snap/nethack/current/HELLO snap/nethack/current/HI
~$ find ./snap/nethack | grep -E 'HELLO|HI'
./snap/nethack/87/HI
./snap/nethack/113/HELLO

Ok. Now the question is, what happens if I tell snapd to restore the snapshot we have saved for revision 87, now that revision 113 is current?

~$ snap restore 1
Restored snapshot #1.
~$ find ./snap/nethack | grep -E 'HELLO|HI'
./snap/nethack/87/HELLO
./snap/nethack/113/HELLO

HTH, HAND.

1 Like

HI, thanks for the answer! (It seems something went wrong when you pretended HELLO->HI, the result should be reversed, but one can get the idea.)

But it didn’t really get to the questions I asked. I followed your procedure though and after snap restore I found that edge is still being tracked:

% snap info nethack
name:      nethack
summary:   The popular nethack console adventure
publisher: Oliver Grawert (ogra)
store-url: https://snapcraft.io/nethack
contact:   ogra@ubuntu.com?subject=nethack
license:   unset
description: |
  Play the nethack cli version
commands:
  - nethack
snap-id:      i2ba1vb7DvsIzb8R987xvPGMQWNHiARe
tracking:     latest/edge
refresh-date: today at 13:42 CEST
channels:
  latest/stable:    3.6.6 2020-09-07  (87) 20MB -
  latest/candidate: 3.6.2 2019-08-24  (79) 13MB -
  latest/beta:      3.6.2 2019-08-24  (79) 13MB -
  latest/edge:      3.6.6 2023-08-31 (113) 15MB -
installed:          3.6.6            (113) 15MB -

Therefore the answer is yes: Restoring to a snapshot created in a different version of the snap will not restore the snap to that previous version, which can possibly lead to incompatible or, if the snapped program itself isn’t sufficiently smart, corrupt data.

I think such behavior should be documented in the main post.

hi!

i was going through the canonical k8s tutorial, and read that when deleting the snap, this creates a snapshot by default, which can be after used to restore its user/system data.

i read about the snap restoring process , but it does not explain much on the requirements for it to work, which, from what I’ve checked on this forum, is a frequent question.

  1. first of all the snap has to be installed, since the snapshot only handles user/system data.
  2. secondly, the snap daemons of the app have to be stopped snap stop <snap_app>, previous to restore. then you can start them again snap start <snap_app>

having satisfied these two points, the restore process works correctly!

hope this can be clarified on the doc of snap, and maybe add a cross-ref link from the k8s tutorial.

note: i was granted permission and now it’s edited