Rsync'ing the /snap directory - backup is bigger than source

Hello,

My Problem
I am using rsync to backup my Ubuntu 22.04 server running on an SD Card in a Raspberry Pi. I run the backup when Ubuntu is also running with this command:

> rsync -aDH --partial --numeric-ids --delete --force --exclude "${MNTPATH}" --exclude '/dev' --exclude '/lost+found' --exclude '/media' --exclude '/mnt' \
> --exclude '/proc' --exclude '/run' --exclude '/sys' --exclude '/tmp' --exclude '/var/swap' --exclude '/etc/udev/rules.d/70-persistent-net.rules' \
> --exclude '/var/lib/asterisk/astdb.sqlite3-journal' "${OPTIONS[@]}" / "${MNTPATH}/"

So all (I think) of the directories whose content appears at runtime only are excluded from the backup.

HOWEVER… my resulting backup is always a lot bigger than I would expect. The backup is created using rsync on to a image file (the truncate command) that is partitioned and formatted (boot partition as fat32 and root partition as ext4).

Trying to Debug
When I loopmount this image file and also take out the Ubuntu server sd card and mount them both separately in my laptop I get this:

The root partition on the backup image… it has been loop device mounted in my laptop running Manjaro.

[flex@flex~]$ sudo du -hs /media/writable/
4.8G    /media/writable/

The root partition on the SD Card containing the OS that the backup was made from with that SD Card removed from the Pi.

[flex@flex~]$ sudo du -hs /media/writable1/
3.3G    /media/writable1/

Why is the backup 1.5GB bigger than the original it was made from? The backup should be smaller because it excluded a lot of runtime directories from Ubuntu running on the sd card.

More Debugging
I ran this rsync command to double check if some mysterious extra files were being created on the destination that are not on the source or were the same files getting bigger when they were on the destination compared with them on the source:

sudo rsync -nrlpgoDv --delete --exclude-from='/home/ubuntu/exclude-list.txt' / /mnt/dst_root/

/home/ubuntu/exclude-list.txt
dev
lost+found
media/4TB1
mnt
proc
run
sys
tmp

But I don’t see any big differences between the source and destination in terms of numbers of files or the size they occupy in source vs destination.

Snap is installed

ubuntu@ubuntu:/snap$ ls -l
total 20
-r--r--r-- 1 root root  548 Apr 19 11:11 README
drwxr-xr-x 2 root root 4096 May 21 21:30 bin
drwxr-xr-x 4 root root 4096 May 27 17:19 core20
drwxr-xr-x 4 root root 4096 May 21 21:30 lxd
drwxr-xr-x 4 root root 4096 May 27 17:18 snapd

ubuntu@ubuntu:/snap$ snap list
Name    Version        Rev    Tracking       Publisher   Notes
core20  20220512       1498   latest/stable  canonical*  base
lxd     5.0.0-b0287c1  22927  5.0/stable/…   canonical*  -
snapd   2.55.5         15909  latest/stable  canonical*  snapd

ubuntu@ubuntu:/snap$ df -h
Filesystem      Size  Used Avail Use% Mounted on
tmpfs           379M  4.7M  375M   2% /run
/dev/mmcblk0p2   30G  3.7G   25G  13% /
tmpfs           1.9G     0  1.9G   0% /dev/shm
tmpfs           5.0M     0  5.0M   0% /run/lock
/dev/mmcblk0p1  253M  148M  105M  59% /boot/firmware
/dev/sda1       3.6T  1.3T  2.2T  37% /media/4TB1

ubuntu@ubuntu:/snap$ sudo losetup -a
/dev/loop1: [45826]:1392 (/var/lib/snapd/snaps/core20_1498.snap)
/dev/loop4: [45826]:7068 (/var/lib/snapd/snaps/snapd_15541.snap)
/dev/loop2: [45826]:77516 (/var/lib/snapd/snaps/lxd_22915.snap)
/dev/loop0: [45826]:7407 (/var/lib/snapd/snaps/core20_1437.snap)
/dev/loop5: [45826]:7773 (/var/lib/snapd/snaps/snapd_15909.snap)
/dev/loop3: [45826]:7426 (/var/lib/snapd/snaps/lxd_22927.snap)

My Theory
Is this because of snap? Maybe because when I rsync from Ubuntu I copy snap installation and data files and settings which exist in squashfs file system which is compressed and so on the destination they take up less space?

Any advice much appreciated.

Cheers,

flex

snap packages are gpg signed, read-only, compressed squashfs image files that do not get unpacked but only get mounted read-only on your system …

the mount points for these files are located underneath /snap

snaps do have a built-in rollback mechanism, which means you always at least have the last as well as the current version on disk …

what you are doing when rsyncing the /snap directory is effectively “unpacking” the snap content into your target location … what you are also doing is to unpack the redundant “roll-back-snaps” from the former version … so you end up with duplicated and uncompressed content that should never have been unpacked in the first place :slight_smile:

do not sync the content of /snapif you feel the need to back up that directory, do not go deeper than two levels …

writable system data of snaps lives underneath /var/snap, writable user data in ~/snap and the actual snap image files live in /var/lib/snapd

2 Likes

@ogra
Thanks so much for your really helpful reply which cleared things up for me nicely :slight_smile: