I have a strange issue (to me...)

 

Drive 3 repeatedly disables based on error count. It is only ever drive 3. All array drives (except Parity) are SAS drives running off a 2 port LSI SAS controller, using the miniSAS 8087 to 4x SAS/SATA power combined connector (SATA power plugs into the back)

Steps taken so far:

Replaced the drive with a known good unit

Changed power connection for SAS drive

Changed SAS connection for SAS drive

Changed port on controller for the 8087 end

Changed cable to SAS drives from controller (bought brand new one)

Replaced power supply for system (was using a Gamemax now using Corsair RM Modular - all drives are directly connected with no splitters)

 

Even with all those changes it's still only ever drive 3 that disables.

 

 

I'm stumped and going to resort moving the emulated content to the other drives and cutting down to 4 array drives in the mean time.

Diagnostic data added

 

Thanks

serenity-diagnostics-20250211-1407.zip

It's not logged as a disk problem, but looks like you already replaced everything, I would try once more to swap cables/slot with a different disk, and see where the issue follows.

I've now tested with another known good drive with the same result. All 3 drives have been tested again under Windows and Linux with no issues and as above not showing as a disk problem.

It's always disk 3 regardless of which physical SAS connection or SATA power connection is used.

Currently it's looking like I may have to move to one of the above alternatives and lose the array functionality which isn't something I want to do.

On 2/11/2025 at 9:20 AM, Cactus42 said:

Drive 3 repeatedly disables based on error count

A drive is disabled when a write to it fails, error count isn't even considered.

8 minutes ago, trurl said:

A drive is disabled when a write to it fails, error count isn't even considered.

OK, I was under the impression it was error count based as that's the point that this one is being disabled, but that doesn't explain why Unraid is disabling good drives only when on a specific chosen point in the array. When not in the array they all pass preclear, they have been tested in other machines. This happens on all 8 SAS connectors which work fine for other drives, same with SATA power connectors.

Does Unraid get upset at other reasons for a write fail even if a drive is fine? like bad data on a good drive being copied (not that I've done that while this has been going on).

Could a bad parity write/update cause this (Parity was fine and tested every month until the first failure showed up)?

Edited February 13, 20251 yr by Cactus42

Any reason the write fails is a reason to disable the drive because the failed write makes it out-of-sync with parity. Parity was updated to reflect the write, so that write and any subsequent writes are emulated by parity and can be recovered by rebuilding.

 

If a write to parity failed, then parity would be the disabled disk.

13 minutes ago, trurl said:

Any reason the write fails is a reason to disable the drive because the failed write makes it out-of-sync with parity. Parity was updated to reflect the write, so that write and any subsequent writes are emulated by parity and can be recovered by rebuilding.

 

If a write to parity failed, then parity would be the disabled disk.

So does that mean if I run the array without a parity drive (I mean for testing - I can back up the important contents elsewhere while I do so) i can see if i still get errors but it won't disable the drive? I want to see if it's something to do with my data, the way it's being synced etc as I've pretty much ruled out hardware.

46 minutes ago, Cactus42 said:

i can see if i still get errors but it won't disable the drive?

Correct.