I have a single array and a cache pool that consists of 2 mirrored ZFS m.2 drives. Recently, a few days into my uptime after updating for the security patches, I see one of the two drives in my cache dropped offline. I thought that maybe my motherboard was acting weird, so I tried to restart the whole system to see if the drive would come back online. However, the array fails to start because of the messed up cache drive. I restarted the server again, unassigned the cache drive, clicked yes I want to do this, and tried to hit start. After doing this, I've tried starting the array in normal and maintenance mode, but I get endless syslog errors regarding the unassigned drive being unreadable, and the array never starts. Do I need to change the slots from 2 back to 1? I don't want to lose my data on the cache drive that is still good. It's mirrored so I want to think I'll be okay to do this?

I rely on my docker containers pretty heavily so this is a huge bummer. I'm not sure why unraid handled this more gracefully while the array was already started. diags attached. Thank you much

thewired-diagnostics-20260516-2301.zip

One of the NVMe devices appears to be failing cauisng very slow i/o:

May 16 23:00:01 TheWired kernel: critical medium error, dev nvme0n1, sector 0 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 2

May 16 23:00:01 TheWired kernel: nvme0n1: I/O Cmd(0x2) @ LBA 0, 8 blocks, I/O Error (sct 0x2 / sc 0x81)

May 16 23:00:01 TheWired kernel: critical medium error, dev nvme0n1, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2

May 16 23:00:01 TheWired kernel: Buffer I/O error on dev nvme0n1, logical block 0, async page read

May 16 23:01:01 TheWired kernel: nvme0n1: I/O Cmd(0x2) @ LBA 0, 8 blocks, I/O Error (sct 0x2 / sc 0x81)

May 16 23:01:01 TheWired kernel: critical medium error, dev nvme0n1, sector 0 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 2

May 16 23:01:01 TheWired kernel: nvme0n1: I/O Cmd(0x2) @ LBA 0, 8 blocks, I/O Error (sct 0x2 / sc 0x81)

May 16 23:01:01 TheWired kernel: critical medium error, dev nvme0n1, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2

Physically remove that device and try starting the pool with only the other one.

On 5/17/2026 at 4:25 AM, JorgeB said:

One of the NVMe devices appears to be failing cauisng very slow i/o:

Physically remove that device and try starting the pool with only the other one.

Legend, worked perfectly. I do wonder why the software was able to handle this while the array was already started, but not after a reboot. Bad time to buy an m.2 right now, but at least my data was protected.

Question though, when my replacement m.2 arrives, can I simply add it back into the cache pool to fill in the missing spot without losing data on the good device?

4 minutes ago, Friarsgate_Customs said:

when my replacement m.2 arrives, can I simply add it back into the cache pool to fill in the missing spot without losing data on the good device?

Should be as simple as that.