Issue: Replacing Cache Drive (1TB SSD → 2TB NVMe)

I’m trying to upgrade my 1TB SSD cache drive to a 2TB NVMe. Here’s what I did:

 

• Followed forum guides: moved data off, stopped the array, set the cache slot to "no device".
• Attempted to start the array in maintenance mode, but I got an error:
  "cache - too many wrong or missing devices"
• Tried adding the new NVMe temporarily to then remove it (as per official docs), but no success.

 

I also tried:

• Reducing cache slots from 2 to 1 — unRAID wouldn’t allow it. Is that normal?
• Ran

  wipefs -af /dev/sdX

  on both drives, rebooted — but they still show up as "missing" in the cache pool.
• Renamed super.dat to super.dat.bak in /boot/config to trigger a rebuild — this unassigned my data and parity drives, but NOT the cache drives... So I had to reassign them manually.

 

Now I’m getting this error for the parity drive:

"All existing data on this device will be OVERWRITTEN when array is Started"

I triple-checked serial numbers — they match exactly. My question:

Can I safely start the array? Or will this force a parity rebuild? Should I be creating a new configuration instead?

 

Even after all that, the cache pool is still not functional. I’m kind of stuck. Any ideas on how to fix this?I

Edited April 30, 20251 yr by 2TC
Mark as solved

Which filesystem was the original cache device?

@JorgeB It was/is using BTRFS

Post the output from

btrfs fi show

 

btrfs fi show

The response is empty. The array is not started (if that is required). I am still worried about the part where "All existing data on this device will be OVERWRITTEN when array is Started". If I can start it without it triggering a new parity sync or if I can rebuild the config without having to triggering it. 

 

Essentially afraid I will lose the content on the data drives.

13 hours ago, 2TC said:

I triple-checked serial numbers — they match exactly. My question:

Can I safely start the array? Or will this force a parity rebuild? Should I be creating a new configuration instead?

 

If the devices are correctly assigned, you can check "parity is already valid"and start the array.

 

I missed this part before:

 

13 hours ago, 2TC said:

Ran

wipefs -af /dev/sdX

on both drives

 

 

So I assume there wasn't any data there? You can now just create a new pool.

 

Alright. That is great!

 

yes, they both were completely empty and all the data was moved over to the array before I tried working on the cache drives.

 

I created a new pool. But the old pool os still there. Indicating missing devices. I will not be able to start the array unless they are assigned. Now I have 2 pools giving me the same error.

 

I would like to delete both pools, add one new pool and put the new 2TB nvme drive into it.

 

I figured it out after reading how to delete a pool.. I did not realize I could click the pool name and then remove the pool.. That really feels like a very rookie mistake.

I did the following

 

* Deleted both pools

* Created a new pool

* Assigned the new drive and started the array.

* formatted the newly assigned drive.