[6.12.10] Help! XFS repair needed on disk? [SOLVED]

[jaso]

I just fired up Plex to watch a few TV shows and noticed that large chunks of TV episodes were missing. When I had a look at my Unraid server I noticed that Disk 6 is acting really strange. I've had a few disks die on me, but I've never had this happen. In the Unraid UI the file contents of disk 6 - rather than listing my shares - seems to be listing the contents of the underlying OS root!

This can't be good.

 

When I bring up a terminal window and use mc tolist the contents of /mnt everything appears AOK except for disk6 which is listed in red and has a question mark in front of it! When I double-click on "disk6" mc states "Error: Cannot read directory contents"

This can't be good.

 

I am crossing fingers, hoping that if I do a reboot my Disk 6 will magically re-appear...

 

My syslog looks like this:

 

I have dumped the diagnostics file. Looking at syslog1.txt I can see lines like this:

May 28 18:24:37 Tower kernel: XFS (md6p1): Corruption detected. Unmount and run xfs_repair

 

I just watched SpaceInvader One's video "How to fix XFS File System Corruption on an Unraid Server" and I am starting to feel a little bit better.

 

Before I commit to the XFS repair I was hoping the Unraid brains trust could chime in and let me know if I am heading in the right direction?

 

Cheers,

Jaso 

tower-diagnostics-20240529-1811.zip

Edited May 29 by jaso
edit 1: added Unraid version to topic. Edit 2: marked as solved

[JorgeB]

Before checking filesystem replace the cables on disk6, since there are constant ATA errors, and probably what caused the issues, then run xfs_repair.

[jaso]

8 minutes ago, JorgeB said:

Before checking filesystem replace the cables on disk6, since there are constant ATA errors, and probably what caused the issues, then run xfs_repair.

 

Thanks JorgeB - will do. I did a parity check the other day and it makes my components run HOT and I always get a lot of ATA errors. I usually prop up a fan to the server when that happens but forgot to do it this time.

 

(The irony is that I ordered new kit lat week (mobo, CPU, ram, case, PSU, HBA, etc.) to migrate my drives into a new Unraid host. They are sitting in the courier depots and will probably arrive tomorrow.)

[jaso]

Update:

• Replaced data cables for disk6
• executed a xfs repair on Disk6 from Unraid UI, which took about 5 minutes 
• repair went pretty good - there were only a handful of files that were placed in lost+found folder, which took about 20 minutes of renaming and moving.

Question: Can I trust the parity status or do I need to redo parity?

[JorgeB]

Parity is maintained during filesystem repair, assuming it was correctly done, e.g., using the WebGUI.