# SetUID permissions are not restored

**URL:** <https://forum.duplicati.com/t/setuid-permissions-are-not-restored/10944>\
**Category:** Support\
**Created:** [October 14, 2020, 8:59am UTC](https://forum.duplicati.com/t/setuid-permissions-are-not-restored/10944 "2020-10-14T08:59:27Z")\
**Posts on this page:** 1\
**Showing post:** 5

<div class="post-metadata">

**Author:** ![ts678](https://forum.duplicati.com/letter_avatar_proxy/v4/letter/t/8491ac/32.png) [@ts678](https://forum.duplicati.com/u/ts678)\
**Post date:** [January 27, 2021, 4:55pm UTC](https://forum.duplicati.com/t/setuid-permissions-are-not-restored/10944/5 "2021-01-27T16:55:00Z")

</div>

> [@martineliascz](#):
>
> Sure, once we confirm it should really work like this. OK?

Of course you can file an issue later (or never), but a filed issue potentially gets developers involved.  
If you were planning code-level research, I can suggest files and docs that you might start looking at.  
If you are developers, pull requests are also most welcome, if you can track down the source of this.

> [@martineliascz](#):
>
> I guess the stored metadata should contain 2541 decimal (4755 octal), right?

That would be my guess. The way to see such lines is to flip through files in a dblock.zip of a backup.

> [@martineliascz](#):
>
> I have excluded the following virtual directories from Duplicati backup for such reason:

At least some of these exotic files can be anywhere. It depends on where system and users put them.  
Conventions exist, so you can probably get an approximate restore, but a for-sure-exact may be hard.

> [@martineliascz](#):
>
> please correct me if I am wrong.

You can test this yourself. I have a suspicion that at least /lost+found (being persistent-file-oriented) is persistent. Others are persistent places to do things like mount things (by convention). A mount point persists even if what is mounted on it might come and go. You can try [findmnt](https://linux.die.net/man/8/findmnt) or [df](https://linux.die.net/man/1/df) or [mount](https://linux.die.net/man/8/mount) to study.

> [@martineliascz](#):
>
> Please let me know if you know about any solution which I could utilize to create remote bare metal backups more efficiently - possibly without having to turn off the servers and installing the OS somewhere else in order to get all those permissions.

Is turning off the servers OK at all? Backing up running applications is tricky, as they are likely holding memory-resident data. On Windows, some applications let VSS arrange flush, but Linux lacks VSS…

If you’re willing to be crash-consistent, your options broaden and can be helped with low-level abilities.  
Some backup software (including Duplicati, if Linux is set up right to allow it) can use [LVM snapshots](https://duplicati.readthedocs.io/en/latest/06-advanced-options/#snapshot-policy). Snapshot capabilities are also available through through [filesystems](https://en.wikipedia.org/wiki/Comparison_of_file_systems#Block_capabilities) such as ZFS, if any are available.

[How far is duplicati away from doing a “full” system drive image backup?](https://forum.duplicati.com/t/how-far-is-duplicati-away-from-doing-a-full-system-drive-image-backup/2610) is one of the few discussions relevant to your question that I could find. There are more on the Internet, and maybe you could search.

Questions for you to decide are your consistency needs (e.g. application-consistent, crash-consistent), and restore needs (lost system, lost individual files). There are some image backups, such as Macrium Reflect on Windows that combine an image backup with file-level abilities. I assume it reads its images.

If your applications cooperate, a crash-consistent image might suit your restore if your whole drive dies.  
Individual data file restores could be Duplicati even currently, but still be careful of the consistency need.

---

_[View the full topic](https://forum.duplicati.com/t/setuid-permissions-are-not-restored/10944)._