Best Practices: Which folders/directories to exclude?

[jotteerr]

From the recent Duplicacy Backup Failure thread, one can see that Duplicacy documentation should make clearer which purpose the software has (basically incremental backups of user folders, not cloning a whole hard drive incl. the OS). Arq, which I evaluated, too, but ditched in favor of Duplicacy, does this, and furthermore it has a preconfigured set of excludes for the most frequent use case of backing up the user folder.

For Mac OS these are certain subfolders in the ~/Library folder:
~/Library/Caches,
~/Library/Logs
~/Library/Saved Application State
~/.Trash

Other OSes have other directories/folders which shouldn’t be included, as /dev and /proc for Linux, but I leave that for others to recommend.

[towerbr]

Remembering what was mentioned by @Droolio in the other topic linked above, that Duplicacy is a file-based backup software.

If the user wants to back up the entire drive (C:\), excluding unwanted folders by filter, I think that on Windows, the folders to be exclude would be these:

C:\ProgramData
C:\Users\[user]\AppData\Roaming
C:\Users\[user]\AppData\Local
C:\ProgramData
C:\Program Files
C:\Program Files (x86)
C:\Users\Public
C:\WINDOWS

Which is basically what is in the environment variables:

%ALLUSERSPROFILE%
%APPDATA%
%LOCALAPPDATA%
%ProgramData%
%ProgramFiles%
%ProgramFiles(x86)%
%PUBLIC%
%TEMP%
%TMP%
%windir%

But I think this is an inadequate way to configure backup, as it is very subject to errors in configuring which folders to exclude (different drive letters, settings, etc).

For example, the recycle bin folder is a bit tricky to find:

$Recycle.Bin\%SID%

where %SID% is the SID of the user

It is much easier to set up repositories pointing directly to the folders that contain files that you want to back up.

[saspus]

And I’ll link this here because I think it is still not Implemented:

Honoring com_apple_backup_excludeItem on MacOS

On MacOS an extended attribute com.apple.metadata:com_apple_backup_excludeItem is used to tell backup software that the particular file does not need to be backed up. This is a more robust solution compared to separately maintained global exclusion list as the metadata is kept physically close to the the data it describes; that way applications, who have much better idea about how to manage their files, can decide what gets backed up and what is transient and shall be skipped. Example: myimac:…

Ideally, you should be able stick the whole ~ into duplicacy and it should do the right thing. The only exclusions would needed to be specified are for apps who don’t set those extended attributes.

It’s unrealistic to expect users to do that manually when applications know best what needs to backed up and what does not.

[saspus]

I take different approach. Instead of excluding stuff I don’t need – I only include stuff that I do and then exclude everything else.

A good starting point I published here – https://blog.arrogantrabbit.com/backup/Duplicacy-CL-setup-on-macOS/ – it’s may be outdated now, with the Catalina Beta; maybe I’ll update that article when Catalina is released.