HI all,
Happy New Year!

I see this message under the disks tab

pool: disk2

state: ONLINE

status: One or more devices has experienced an error resulting in data

corruption. Applications may be affected.

action: Restore the file in question if possible. Otherwise restore the

entire pool from backup.

see: https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-8A

config:

NAME STATE READ WRITE CKSUM

disk2 ONLINE 0 0 0

md2p1 ONLINE 0 0 28

errors: 35 data errors, use '-v' for a list

I did a -v and checked the files it says are damaged - the files work fine.

There is a button saying SCRUB - what does it do?

What should I do at this point?

Thank you

Daytona

This is a very common ZFS situation, and the wording of the message is scarier than what’s usually going on....

From your output:

state: ONLINE
status: One or more devices has experienced an error resulting in data corruption.
...
md2p1 ONLINE 0 0 28
errors: 35 data errors

Key facts:

• The pool is ONLINE

• There are no read or write errors

• There are checksum (CKSUM) errors

• ZFS detected 35 blocks whose checksums did not match

• ZFS cannot silently fix them because this is a single-disk pool

This is detection, not failure — and ZFS is doing exactly what it is supposed to do.

Why the files “work fine”

ZFS errors are block-level, not file-level.

That means:

• A corrupted block might be:

  • unused padding

  • metadata that isn’t currently read

  • part of a compressed block that still decompresses

  • part of a file that hasn’t been touched yet

• Many apps do not notice minor corruption immediately

• ZFS never guesses — if a checksum mismatches, it records it permanently

So “the file opens” ≠ “the data is correct”.

ZFS is more strict than most filesystems — that’s a feature.

What the SCRUB button does (important)

A scrub:

• Reads every block in the pool

• Verifies every checksum

• If redundancy exists (mirror / RAID-Z):

  • Automatically repairs bad blocks

• If redundancy does NOT exist (single disk):

  • Confirms and logs corruption

  • Cannot repair

A scrub does not:

• delete data

• rewrite files

• fix corruption without redundancy

Think of scrub as a full audit, not a repair tool.

What you should do — step by step

*Run a scrub (yes, press the button)

Or from CLI:

zpool scrub disk2

Check progress:

zpool status disk2

This confirms:

• whether the error count increases

• whether corruption is stable or ongoing

• As you may have a bad disk inteh zfs setup that needs replaced...

List exactly which files ZFS thinks are bad

You already ran -v, but for completeness:

zpool status -v disk2

You’ll see paths like:

/mnt/disk2/some/path/file.ext

If the same files reappear after scrub → confirmed corruption.

If the files are non-critical

If you don’t care about those files:

rm /mnt/disk2/path/to/file
zpool clear disk2

Then scrub again.

This often resolves the warning permanently.

If the files matter

ZFS cannot repair without redundancy.

Your options are:

• Restore the affected files from backup

• Re-copy them from a known good source

• If unsure → delete and re-ingest

After replacement:

zpool clear disk2
zpool scrub disk2

---

What zpool clear actually does

zpool clear disk2

This:

• clears the error counters

• does not erase history

• does not hide future errors

• is safe to run after addressing the cause

If corruption still exists, it will reappear.

Important diagnostic checks (do these)

*SMART check (very important)

smartctl -a /dev/sdX

Look for:

• Reallocated sectors

• Pending sectors

• CRC errors

• Read errors

Checksum errors often come from:

• marginal sectors

• bad SATA cables

• power instability

• RAM errors (ECC strongly recommended)

see openzfs docs:
https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-8A/

more info on the scrub:
https://openzfs.github.io/openzfs-docs/man/v2.4/8/zpool-scrub.8.html

Bottom line

• SCRUB = full integrity scan

• Single-disk ZFS cannot repair, only detect

• “Files work” does not mean “no corruption”

• ZFS is behaving correctly

• This is not panic-level unless errors grow

• Potential bad disk needs replacement

ultimately this tells me that you set a zfs in the unriad array where zfs should be added to unraid as pool devices.

as you don't have parity, zfs nor a good setup fo zfs implementations on unraid...

Good idea to run memtest, since that's the #1 reason for data corruption.

Thanks folks Appreciate all the help.
Shall run memtest as well.

I had the old XFS file system and converted to ZFS one disk at a time. Had no idea I should have done some other stuff.
Can someone please point me to the best practice when using ZFS?

I have a 2 parity drive UNRAID Version 6.12.4 2023-08-31 with 13 data drives.

9 hours ago, daytona235 said:

Can someone please point me to the best practice when using ZFS?

Nothing extra typically needs to be done; you can scrub the disks to confirm all is good, but that should only be needed if there are sync errors after doing a parity check.

so this is wrong.

if all theses 12 disks (2 parity) ... each disk made there own zfs zpool with 1 disk. you lose all the zfs benefits such as rad z mirroring, raidz 1 (equivalent to raid 5 miror and parity stripe)
parity done't fuction as it should when the disk is lost.

usualy you would pare mutiple disk togetr in a zpool 3 disk min to make a raidz1 which strip[es for speed and mirors for redundacny/protection. you can lose 1 disk and still maintain all your data.
or do a raidz2 same strip and mirror, but cand lose 2 disk and keep all your data. (min of 5 disks)

see the zfs docs:
https://docs.unraid.net/unraid-os/advanced-configurations/optimize-storage/zfs-storage/

and a few youtube videos:
https://www.youtube.com/watch?v=XQC6lXOXGT4
https://www.youtube.com/watch?v=UR5RCItyCsw

and forum post:

I highly recommend fixing your storage by pulling 3 disk of equivalent size and making a zfs raidz1 reformatting them to make a zfs pool

the hybrid approach ( the setup your currently using)

*The hybrid ZFS-in-array approach is helpful for specific backup or replication scenarios but is not a replacement for a full ZFS pool. ZFS disks in the array are managed individually; you do not get the combined performance, redundancy, or self-healing of a true multi-disk ZFS pool. For full ZFS functionality, always use dedicated ZFS pools.

https://unraid.net/blog/zfs-guide?srsltid=AfmBOor7XJrR8tcD76hGK2pYvWyDKAUcGg1gvI3YkL3jsHH2N01zatXq

Edited January 6Jan 6 by bmartino1