I've recently purchased a Pro license and have installed a trial on my existing NAS hardware (previously running fine with Proxmox) using 4 brand new WD Red Pro 16TB drives.

The parity read-check won't complete without the drive being marked as disabled (I've tried twice). Unraid is also reporting 832 errors on disk 2, but there is nothing loggedmarvin-diagnostics-20231220-1300.zip on a smart test.

Confused why the new drives are failing under Unraid.

 

RAID controllers are not recommended with Unraid for many reasons, but doesn't look like any of your array disks are using it.

 

You have a lot of Unassigned disks attached, and since you are having multiple disk problems, I suspect power. Any splitters?

 

 

You are correct - hardware RAID is not being used anywhere in this system.

The machine is built in a case with a backplane supporting 24 3.5" bays. These are connected to SAS RAID controllers, but I've never used hardware RAID - even in the previous NAS (non-Unraid) build running in Proxmox, the disks on the controllers were passed through to the NAS OS and the array was managed via the NAS OS. All 16 drives were working fine in that setup.

In fact, the 4 new 16TB drives replaced 16 3.5" HHDs, so the machine is using much less power than previously, so with the spread across the rails, the 1000w PSU shouldn't be having any issue.

 

Would it be worthwhile trying the two drives reporting errors in different bays? I assume Unraid will have no issue with them being moved?

Edited December 20, 20232 yr by DrRed

2 hours ago, DrRed said:

Unraid will have no issue with them being moved?

Unraid tracks disk assignments by serial number, so no problem moving them (one of the reasons RAID controllers not recommended is because they don't always pass serial number).

I did a test by swapping the bay with the Parity disk with the drive for disk 1 (via the drive bay caddies).

 

Result: Still no errors with disk 1, but the parity drive still got read errors and was disabled after around 3 hours of the Read-Check.

 

Run an extended SMART self-test

16TB drive, so that took a while.

"Completed without error"

Post new diagnostics

Here's the latest...

 

marvin-diagnostics-20231223-1352.zip

On 12/21/2023 at 7:06 PM, DrRed said:

I did a test by swapping the bay with the Parity disk with the drive for disk 1 (via the drive bay caddies).

 

Result: Still no errors with disk 1, but the parity drive still got read errors and was disabled after around 3 hours of the Read-Check.

 

Unraid disables a disk because a write to it fails. Syslog is showing write errors to parity and read errors on disk2. Looks like you were doing a parity rebuild instead of a read check.

 

Some messages in syslog about the RAID controller. Are you sure no drives are using it? Can it be removed?

8 minutes ago, trurl said:

Unraid disables a disk because a write to it fails. Syslog is showing write errors to parity and read errors on disk2. Looks like you were doing a parity rebuild instead of a read check.

 

Okay - "Read-Check" wasn't the correct terminology in my previous post - I've never managed to get a working/built parity drive so far, so yes it's doing a rebuild.

8 minutes ago, trurl said:

Some messages in syslog about the RAID controller. Are you sure no drives are using it? Can it be removed?

The system has the following Adaptec raid cards installed:

• ASR-71605 (16 port)
• ASR-72405 (24 port)

The main array drives (currently 4x 16TB) are plugged into the SAS back plane in the server case, which is connected to the 24 port RAID controller (again, no hardware RAID is running).

The other drives will eventually (if I can get the main array working correctly) be used to as drive pools for OS images, dockers, virtual machines (or passed directly to the VMs) etc.

 

I think the next steps need to be along the lines of:

1) Connect the parity drive on a seperate/isolated power cable and PSU rail - then retry the parity rebuild

2) Temporarily remove the other drives - then retry the parity rebuild.

 

Any other sugesstions in addition to the above?

7 minutes ago, DrRed said:

• ASR-71605 (16 port)
• ASR-72405 (24 port)

You should downgrade to v6.12.4 now, there's a known kernel bug affecting those controllers, see the release notes, it will cause timeouts resulting in read/write errors.

1 hour ago, JorgeB said:

You should downgrade to v6.12.4 now, there's a known kernel bug affecting those controllers, see the release notes, it will cause timeouts resulting in read/write errors.

Based on my setup, would it make more sense to run Unraid on Proxmox? My thinking is that Proxmox will handle the raid controllers and pass the virtual disk into Unraid.

Expect the kernel bug to be dealt with by new kernels on a future Unraid version. If that is the only reason you have for virtualizing Unraid, then I wouldn't do it. I'm never quite sure why anyone chooses to virtualize Unraid anyway since Unraid can host VMs itself.

 

And virtualizing Unraid is not officially supported, though there is a subforum for that where other users help each other. Of course, "peer" support is mostly what happens on the forum anyway.

12 hours ago, DrRed said:

Based on my setup, would it make more sense to run Unraid on Proxmox?

Latest Proxmox 8.1 is affected by the same kernel bug.

 

12 hours ago, DrRed said:

My thinking is that Proxmox will handle the raid controllers and pass the virtual disk into Unraid.

That's usually not recommend, though virtualizing Unraid is not officially supported, if you are going to do it you should pass-through the controller, don't use virtual disks.

I've downgraded to v6.12.4 and no longer have any issues. Thank you so much for pointing me in the right direction.

Are there any indicators when a fix will be released?

First needs to be fixed in the kernel, after that a new Unraid release including that or a newer kernel will have the fix, you can keep an eye on this bug report:

 

https://bugzilla.kernel.org/show_bug.cgi?id=217599

Having downgraded to v6.12.4 my system has been working fine on 4 main array disks.

However, I've just added 2 brand new disks to the array and am getting errors again 7.7% (2 hours 10 mins) through the parity check.

 

marvin-diagnostics-20240116-1613.zip

New disks 4 and 5 disconnected.

3 hours ago, trurl said:

New disks 4 and 5 disconnected.

I saw that, and at exactly the same time too. Looking likely to be the sata cables or power cables.