[SOLVED] XFS Cache drive "metadata corruption"

[Dephcon]

I've been having some issues lately with my fairly new NVME cache disk.  I've had a couple kernel panics and the cache drive going read-only.

 

This is the first time I've been able to access the system to capture a diag, attached.  This stood out imediately:

Sep 20 20:23:15 vault13 kernel: XFS (nvme0n1p1): Metadata corruption detected at xfs_dir3_leaf_check_int+0x93/0xc3 [xfs], xfs_dir3_leaf1 block 0x3e6f0818 
Sep 20 20:23:15 vault13 kernel: XFS (nvme0n1p1): Unmount and run xfs_repair
Sep 20 20:23:15 vault13 kernel: XFS (nvme0n1p1): First 128 bytes of corrupted metadata buffer:
Sep 20 20:23:15 vault13 kernel: 00000000: 00 00 00 00 00 00 00 00 3d f1 00 00 04 3e 49 e8  ........=....>I.
Sep 20 20:23:15 vault13 kernel: 00000010: 00 00 00 00 3e 6f 08 18 00 00 00 25 00 01 b3 b7  ....>o.....%....
Sep 20 20:23:15 vault13 kernel: 00000020: c9 5c 71 28 56 fd 49 6f 91 d3 34 33 ac a8 ba 20  .\q(V.Io..43... 
Sep 20 20:23:15 vault13 kernel: 00000030: 00 00 00 00 40 36 d5 c0 00 a9 00 1a 00 00 00 00  ....@6..........
Sep 20 20:23:15 vault13 kernel: 00000040: 00 00 00 2e 00 00 00 08 00 00 17 2e 00 00 00 0a  ................
Sep 20 20:23:15 vault13 kernel: 00000050: 00 f7 cc bb 00 00 02 2b 01 0f ac 85 00 00 00 00  .......+........
Sep 20 20:23:15 vault13 kernel: 00000060: 06 6c 50 1c 00 00 00 48 06 76 1a 37 00 00 03 58  .lP....H.v.7...X
Sep 20 20:23:15 vault13 kernel: 00000070: 07 bc cb 9b 00 00 02 b0 0c 77 81 da 00 00 00 00  .........w......
Sep 20 20:23:15 vault13 kernel: XFS (nvme0n1p1): Corruption of in-memory data (0x8) detected at __xfs_buf_submit+0xdd/0x172 [xfs] (fs/xfs/xfs_buf.c:1514).  Shutting down filesystem.
Sep 20 20:23:15 vault13 kernel: XFS (nvme0n1p1): Please unmount the filesystem and rectify the problem(s)
Sep 20 20:23:15 vault13 kernel: blk_update_request: I/O error, dev loop2, sector 35378856 op 0x0:(READ) flags 0x80700 phys_seg 2 prio class 0

 

The "corruption of in-memory data" bit caught my eye, last week I did run a 24hours memtest86 to ensure my also fairly new 4x16GB RAM kits wasn't bad.

 

Based on some other threads, it seems I need to boot into maintenance mode to be able to run 'Check Filesystem Status'.  I had to hard reboot to recover so is there anything else i can do while my parity check is running?

 

Thanks!

vault13-diagnostics-20220920-2025.zip

Edited October 24, 2022 by Dephcon

[JorgeB]

5 hours ago, Dephcon said:

I had to hard reboot to recover so is there anything else i can do while my parity check is running?

No, you need to cancel or wait for it to end.