Hi all,

yesterday suddenly files on my share went missing and after looking at Unraid it said that one drive is disabled (contents emulated). when i tried to look into that directory over cli i got this error:
ls: reading directory '.': Structure needs cleaning

After some back and forth the drive went missing so i swapped the slots in my case. On discord they recommend me then to run the rebuild. After the rebuild the drive now reports as:

Unmountable: wrong or no file system

in fdisk -l i dont see the drive anymore but in lsblk i see it.

I wanted to try to repair the filesystem but the drive is not selectable in Unraid itself.

Now i dont know if its safe to reboot or not.

towerkgn-diagnostics-20251015-1029.zip

Syslog rotated; can't see the beginning of the issue, but there were read errors on multiple disks:

Oct 14 20:01:05 TowerKGN kernel: md: disk2 read error, sector=5155657784

Oct 14 20:01:05 TowerKGN kernel: md: disk0 read error, sector=5155657784

Reboot and post new diags after array start.

Hey, that started during the rebuild. Is it safe to restart even if one disk is currently missing for unraid? @JorgeB

i did a reboot now all drives say Unmountable: wrong or no file system....

here is the new diag

towerkgn-diagnostics-20251015-1029.zip

Diags are the same.

towerkgn-diagnostics-20251015-1221.zip

sry dragged the wrong file

There are errors for multiple disks, e.g.:

Oct 15 12:20:27 TowerKGN kernel: sd 12:0:0:0: [sde] tag#4 Sense Key : 0x2 [current]

Oct 15 12:20:27 TowerKGN kernel: sd 12:0:0:0: [sde] tag#4 ASC=0x4 ASCQ=0x22

Oct 15 12:20:27 TowerKGN kernel: sd 12:0:0:0: [sde] tag#4 CDB: opcode=0x88 88 00 00 00 00 00 00 00 00 40 00 00 00 08 00 00

Oct 15 12:20:27 TowerKGN kernel: I/O error, dev sde, sector 64 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0

Oct 15 12:20:27 TowerKGN kernel: md: disk2 read error, sector=0

Oct 15 12:20:27 TowerKGN kernel: sd 9:0:0:0: [sdd] tag#6 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=DRIVER_OK cmd_age=0s

Oct 15 12:20:27 TowerKGN kernel: sd 9:0:0:0: [sdd] tag#6 Sense Key : 0x2 [current]

Oct 15 12:20:27 TowerKGN kernel: sd 9:0:0:0: [sdd] tag#6 ASC=0x4 ASCQ=0x22

Oct 15 12:20:27 TowerKGN kernel: sd 9:0:0:0: [sdd] tag#6 CDB: opcode=0x88 88 00 00 00 00 00 00 00 00 40 00 00 00 08 00 00

Check power/connections for all disks, especially any power splitters, then new diags.

towerkgn-diagnostics-20251015-1416.zip

Reseated all Sata connections, Power is Molex powered backplane. Cable sit firm in there. One drive is back. Could it be the PCIe Sata Controller?

No disk errors so far, but disk 1 has filesystem issues, check filesystem

THANKS! Did the fix filesystem and drives are back up. Do you have an idea what this was?

I installed lancache last week and i found this https://www.reddit.com/r/unRAID/comments/zygjzr/reading_directory_structure_needs_cleaning_in/. Could Lancache have caused this?

One folder has this error "ls: cannot access 'abc': Structure needs cleaning". Do you know how to fix that?

Also should i start a parity check now?

Edited October 15, 2025Oct 15 by CrispyBearcon

45 minutes ago, CrispyBearcon said:

Do you have an idea what this was?

It looked like the issue was caused by a power/connection issue, or the controller/backplane.

46 minutes ago, CrispyBearcon said:

Could Lancache have caused this?

Don't think so.

46 minutes ago, CrispyBearcon said:

One folder has this error "ls: cannot access 'abc': Structure needs cleaning". Do you know how to fix that?

That suggests there are still fs issues, could be in a different disk, diags may have more info.

towerkgn-diagnostics-20251015-1718.zip

here is the most recent diag.

if i go in that share that had problems in the first place its empty but on the 2 disk i see all data. How can i fix that?

yeah if i navigate to the folder it says "ls: reading directory '.': Structure needs cleaning"

that folder is 4 folders deep into the share. how could just one folder have a problem?

Edited October 15, 2025Oct 15 by CrispyBearcon

this is the DMESG error after ls:

[16318.723526] XFS (md2p1): Metadata corruption detected at xfs_buf_ioend+0xa5/0x380, xfs_inode block 0x395edd3e8 xfs_inode_buf_verify

[16318.723538] XFS (md2p1): Unmount and run xfs_repair

[16318.723541] XFS (md2p1): First 128 bytes of corrupted metadata buffer:

[16318.723545] 00000000: 24 1a 9c 92 6d 85 ce 6d 6f 93 4c d2 44 fc dc 3b $...m..mo.L.D..;

[16318.723549] 00000010: 98 25 67 cf 3f 77 ea c1 c5 b7 19 f8 16 ee f9 6f .%g.?w.........o

[16318.723552] 00000020: 76 48 30 f5 c9 61 87 35 a3 da 2a ee a0 d8 92 c3 vH0..a.5..*.....

[16318.723555] 00000030: cc 6c cd 9b 9b 53 a0 69 79 01 e4 94 72 d5 4f 37 .l...S.iy...r.O7

[16318.723559] 00000040: aa 93 9e 81 25 4c 5d dd d7 25 b1 ba 1c c7 68 6b ....%L]..%....hk

[16318.723562] 00000050: 00 b6 ab b7 f7 3e 76 31 ad 48 42 a3 ae b1 05 df .....>v1.HB.....

[16318.723565] 00000060: de da 64 5c 81 28 13 65 0b 6f 1f 49 78 a2 3e 33 ..d\.(.e.o.Ix.>3

[16318.723568] 00000070: b4 01 36 42 53 25 cc d8 e1 93 28 7f 0a 9c db 66 ..6BS%....(....f

[16318.723577] XFS (md2p1): metadata I/O error in "xfs_imap_to_bp+0x60/0xd0" at daddr 0x395edd3e8 len 32 error 117

can i do this "xfs_repair /dev/md2p1" down to the folder or do i need to do this on the complete drive (or is it share)

Edited October 15, 2025Oct 15 by CrispyBearcon

48 minutes ago, CrispyBearcon said:

(md2p1): Unmount and run xfs_repair

Check filesystem on this disk as well (disk2).

Thanks for the help! it seems everything is back in line.

hey. i hope its ok to reopen this thread. my xfs filesystem again crashed the drive. i ordered a replacement in advance but the smart tests looks fine for the drive. Is there some way to see what this is causing?towerkgn-diagnostics-20251101-2025.zip

Diags are after rebooting, so we can't see what happended, but SMART looks fine, so most likely a power/connection issue.

yeah dont have a diag when it happend. the server was offline without me noticing.... i will check all cabling again.