Backup strategy? my .local/ is 30Go ; my .config/ has cache data, my ~/.* is a mess

Is there any rational regarding application config files? Some rule that would allow to fill bugs for apps that do inconvenient things. Currently here is my situation: I am trying to do backup to start a clean Fedora 34 install.

For example, MuseScore has some configuration files in ~/.local/share/MuseScore/... but ~/.local/share/ is also where Gnome Boxes or VirtualMachine Manager or libVirt are storing their image, so the folder is about 30Go … I do not plan to backup that, I just copied my 3 XML files manually.

There are all kinds of “misbehaviors” like logs files in .config/VirtualBox/ or huge cache in .config/google-chrome/.

And the Home folder is also full of hidden files and folders.

Not mentioning that when you remove an application (via RPM or Flatpak) the config files stay in their hidden directory forever. On Android this kind of things are managed.

Does everybody carefully pick and choose the “relevant” folders and remove unused folders as the applications are removed? or do you just run huge and long backup?

1 Like

I typically backup the home directory excluding some caches and data like this:

rsync -P -r -l -p -t -b --suffix="~$(date +%Y%m%d)" \
--exclude-from=- "${HOME}/" "${RSYNC_DST}/" << EOI
/.cache/
/.config/chromium/
/.config/libvirt/qemu/lib/
/.config/libvirt/qemu/save/
/.config/pulse/
/.local/share/gnome-boxes/images/
/.local/share/gvfs-metadata/
/.local/share/libvirt/images/
/.local/share/tracker/
/.local/share/Trash/
EOI
4 Likes

Thanks for the tips. It is a carefully excluding some folders. It is kind of a workaround.

  • Would it make sense to fill bugs to application that use .config for their cache, logs and what not?
  • Could this .config have like a per application quota :slight_smile: ?
  • or at least some alert(?) could be raised if their size grow beyond 50ko?

I guess I am searching for a system solution … as in Freedesktop, Gnome or Fedora system.

I’m afraid this mainly depends on the apps and their developers.
If you want to raise an issue, each case should be treated separately.
So, the issues should be addressed to the respective upstream bug trackers.

On the other hand, the need to back up specific data is mostly personal preference.
You can save time and space by excluding some subjectively unimportant data.
The data may be stored in accordance with the standards, but it’s just irrelevant for you.

There’re no app-specific quotas in general case.
But you can utilize some sort of disk usage analyzer:
https://wiki.gnome.org/action/show/Apps/DiskUsageAnalyzer

I am just trying to see what could be done globally. So that each user does not have to manage this mess.

I guess .config/ was meant to hold … configuration, but the message has been lost or never clearly stated or it is just my personal interpretation. I will start a crusade to make $HOME/.config/ THE configuration backup folder, free from logs, cache data or anything that is more than a few ko basically.

No doubt that issues should be raised for each application. Cache data is really not a matter of personal preference. But yes, at worse you will just lost time and disk space.

1 Like

Hi Bruno.

I was wondering the same thing. Back in the day I was happy that Déjà Dup was offering this great, convenient way to backup your data on Linux. Their default is to back up /home/$USER with the Trash excluded – seems legit.

But then you get shocked when realising that it would backup your Steam games along with 40 GB Windows boxes (minimum) on your precious cloud storage! Not even mentioning the time it’ll take for your files to get prepared.

I always considered /home to contain user-generated files only. But then again, isn’t a virtual machine image user generated? (;

Maybe opening a ticket at Déjà Dup might be more productive than a discussion here. Folks over there probably discussed this issue already.

Right now, I’m trying to configure an opt-in list for myself, and hoping I don’t forget anything important.

Documents
Images
Videos (self-made ones)
.gnupg (GPG keys)
.ssh (SSH keys)
.thunderbird (I archive emails once a year, the rest is on IMAP any way)

1 Like

Currently:

  • I use a specific script for backup, hand-made and regularly updated to include/exclude specific folders …
  • I removed some applications (minicom, …) that are not good FreeDesktop citizens (and there are alternatives)
  • I fill bugs for applications that cannot be easily replaced (Mozilla, VSCode, JetBrains …). Usually there are existing bugs, Mozilla’s bug is 17 years old for example … but maybe one day …

Now I just found this bug report for Déjà Dup: Flatpak doesn't ignore .var/app/*/cache (#145) · Issues · World / Déjà Dup Backups · GitLab

There the developer acknowledges that the tool is supposed to exclude cache directories by default. The maxim seems to be that adding /home/$USER should suffice as a sane default.

But then again, another documented hypothesis of Déjà Dup is that storage space is cheap.