I've been getting the "Directory Bread Errors found" message from the Fix Common Problems plug-in, and from what I can find, it seems that this indicates a problem with the boot flash drive. Unfortunately the flash drive is attached to an internal port on the motherboard of the server, so accessing it is a bit troublesome. So far, the server seems to be running fine (7.0.1).

Obviously I could shut down and see what the problem is, but my fear is that I might have to replace the flash drive and with all the reports of the problems finding a flash drive suitable for use as a boot drive, I also worry that this might not be solved easily.

So my question is whether I can live with this error, and if so, for how long? I've seen the Youtube video on booting from an internal drive coming (soon-ish?) - so am really tempted to wait for that. Just don't know how serious the Directory Bread Errors are, and whether I should focus on getting that fixed.

Directory Bread Errors indicate active FAT32 filesystem corruption on the boot drive — not something to live with.

Flash drives in this state tend to fail abruptly rather than gradually. The server running fine now is not a reliable indicator of how long that will continue.

Do this immediately regardless of anything else — Main → Flash → Flash Backup, download the zip, store it somewhere off the server.

If the drive fails before you do this, recovery becomes significantly harder.

On waiting for internal boot — a drive showing active corruption will not wait for a software feature with no confirmed release date.

Replace now, migrate to internal boot when it arrives.

On finding a replacement — the situation is better than general forum anxiety suggests.

There is a guide posted on this forum https://forums.unraid.net/topic/196967-usb-flash-primer/#comment-1606140

that addresses USB boot drive selection in detail.

The short version: a quality USB 2.0 drive from 2008-2013 found in a drawer is likely your best practical option at any price point given current market conditions.

If you don't have one, NOS (New Old Stock) drives of exactly this type are currently available on eBay at very reasonable prices — happy to point you to a specific verified listing if useful.

Thanks, makes sense not to take additional risk on this, and I fortunately had an old (still in packaging) Sandisk Cruzer Blade so I shut down the server and took out the flash drive. A little testing of the flash drive on a Linux machine using fsck.fat showed a serious error, so it was clearly time to set up the new flash drive.

Unfortunately that's been a huge headache. I use the Unraid Flash Creator tool, and tried to install both a backup that I just did from the flash drive as well as the backup from Unraid Connect. In both cases, the drive failed to boot. I suspect that the backups had corrupted files from the failing flash drive, so I reverted to an old backup from March 2025. Unfortunately that failed to boot as well, which was a bit surprising. (So separately I'm wondering how I can check the validity of a flash drive backup.)

I have confirmed that if I download and install a fresh 7.0.1 OS (which is what I was on) the machine boots up, but I didn't proceed because I'd still prefer to avoid redoing too much of the old setup. Drive assignment should be fine, but I'm more worried about the settings I had for my Win 11 VM, which I recall were quite complicated (for me). Currently I'm writing another old flash drive backup to the flash drive, to see if that might work.

But I was also wondering if there's a way to copy specific config files (especially for the VM) to a fresh install OS flash drive, so that at least some things can be recreated automatically on boot up. (To address the risk of using a corrupted config file, I would check the file against other older backups to make sure they're the same.)

Good news on the hardware — a still-packaged Cruzer Blade is a solid interim solution, and catching the corruption with fsck.fat before a crash was exactly the right move.

On the recovery — I'm not deeply versed in Unraid's config recovery specifics, but based on what I've read online your instinct about targeting specific config files is exactly right and looks like the cleanest path forward.

A fresh 7.0.1 install boots cleanly, which means the OS itself is fine.

The problem is restoring your specific configuration without importing corrupted files from the failing drive.

The files worth targeting for your VM setup specifically are in the config folder on the flash drive.

The key ones appear to be:

vm.conf — contains your VM definitions including all the settings you configured for the Win 11 VM. This is the file most worth recovering carefully.

domain.cfg — VM global settings.

network.cfg — network configuration.

shares/ folder — individual share configurations.

plugins/ folder — list of installed plugins that auto-reinstall on boot.

Your approach of cross-checking the same file across multiple backups is exactly correct.

If the vm.conf from your March 2025 backup matches an earlier backup, it's almost certainly clean.

The practical workflow based on what I've found — start with the fresh 7.0.1 install that boots cleanly, then copy only the verified config files into the config folder one at a time rather than restoring the entire backup at once.

Boot after each addition to confirm stability before adding the next. This isolates any corrupted file immediately.

Worth waiting for more expert eyes on the specific recovery steps — but this approach seems to be the right direction based on what I've been able to find.

config/super.dat is your array disk assigments, config/pools is your pool assignments.

Might as well just reinstall plugins instead of trying to recover them.

Your docker templates are in config/plugins/dockerMan/templates-user. If you have those and your appdata, containers are easily reinstalled from Previous Apps on the Apps page.

https://docs.unraid.net/unraid-os/troubleshooting/common-issues/docker-troubleshooting/

3 hours ago, Lolight said:

vm.conf — contains your VM definitions including all the settings you configured for the Win 11 VM. This is the file most worth recovering carefully.

Are you seeing this file mentioned in some doc or forum post? AFAIK, it doesn't exist.

22 minutes ago, JorgeB said:

Are you seeing this file mentioned in some doc or forum post? AFAIK, it doesn't exist.

Thank you for the correction — that's exactly the kind of specific knowledge I was lacking and precisely why I flagged that my advice was research-based rather than hands-on experience.

So to correct what I posted — VM definitions in Unraid are stored as individual XML files rather than a vm.conf file.

The global VM manager settings live in /boot/config/domains.cfg on the flash drive.

Could you confirm the exact path where the individual VM XML definition files are stored on the flash drive?

That would complete the recovery picture for the poster trying to preserve their Win 11 VM settings — and correct the record properly for anyone finding this thread later.

13 hours ago, Lolight said:

VM XML definition files are stored

https://docs.unraid.net/unraid-os/using-unraid-to/manage-storage/shares/#default-shares

/mnt/user/system/libvirt/libvirt.img for example

@sonofdbn

Honestly for the Win 11 VM specifically — recreating the VM configuration in the Unraid VM manager from scratch might actually be the cleaner path forward.

Your actual Windows installation lives in the vdisk image file and is completely separate from the libvirt configuration.

The VM manager settings you'd be recreating are just the wrapper that tells Unraid how to run it — CPU cores, memory allocation, network settings, USB passthrough and so on.

Tedious to redo if you had a complex setup, but infinitely less risky than importing a potentially corrupted libvirt database onto a fresh installation.