Disk read errors following a filesystem error

[veri745]

A few weeks ago my array went into an error state, and one of the disks went red due to some filesystem corruption

 

It's a relatively new disk, so I ran SMART diagnostics, ran the xfs_repair tool, ran memtest for several hours, and rebuilt the disk back onto itself

 

Since then, I've had several occasions where disk read errors have cropped back up.  It's always drives 2 and 4, they always have the same number of read errors, and so far it has resolved itself after a reboot.

 

I am wondering if I experienced some data corruption as part of the initial issue, and the parity was someone incorrect and rebuilt the corrupt data back onto the drive.

 

Attaching my diagnostics from before rebooting (ending 2045, no SMART for the two disks with read errors), and after (ending 2050)

 

 

tower-diagnostics-20230322-2050.zip tower-diagnostics-20230322-2045.zip

[JorgeB]

6 hours ago, veri745 said:

and one of the disks went red due to some filesystem corruption

Just to be clear disk goes red when it gets disabled, it doesn't detect filesystem corruption.

 

Disks are dropping offline, this can be a power/connection problem, but since both are on a Marvell controller and these are known to sometimes drop disks without a reason first recommendation would be to replace that with a recommended controller.

 

[veri745]

7 hours ago, JorgeB said:

Just to be clear disk goes red when it gets disabled, it doesn't detect filesystem corruption.

 

Disks are dropping offline, this can be a power/connection problem, but since both are on a Marvell controller and these are known to sometimes drop disks without a reason first recommendation would be to replace that with a recommended controller.

 

 

Right, I meant that xfs_repair detected a filesystem error after the disk got disabled

 

I had extra ports free on the motherboard, so I moved the two suspect disks to different ports.  I will see if that resolves the issue over the next month or two and consider replacing my SATA controller card

 

Thanks

Edited March 23, 2023 by veri745
typo

[veri745]

This card has an "approved" chipset, but seems really inexpensive.  Decent option?
 

https://www.amazon.com/Expansion-Controller-Adapter-Profile-Non-Raid/dp/B09KY3HB8Z/ref=sr_1_6?crid=D0AKVB0XY7UT

[JorgeB]

Should be, it's always difficult to say for sure with these no name brands, sometimes build quality is not good, resulting in CRC errors or similar.

[veri745]

No issues so far after moving all drives off of the Marvell controller