Cache drive failed - replaced with smaller drive - can't start array - CANNOT START ARRAY - "WRONG POOL STATE ; CACHE - TOO MANY MISSING/WRONG DEVICES"

[The Transplant]

I definitely don't do this every day so haven't been able to figure it out yet.  My first cache drive failed.  It was a 240GB.  I had a 180GB SSD on hand so I replaced it.  Now I can't bring the array back online so obviously I am missing something?  Diagnostics attached.  Thanks.

odin-diagnostics-20240224-1106.zip

[JorgeB]

Note that since it's a single device pool there's no option to replace the device and keep the data, if you want to use the new smaller device as a new pool, don't assign any device to that pool, start the array, stop the array, assign new pool device, start the array, format new pool device.

 

[The Transplant]

1 hour ago, JorgeB said:

Note that since it's a single device pool there's no option to replace the device and keep the data, if you want to use the new smaller device as a new pool, don't assign any device to that pool, start the array, stop the array, assign new pool device, start the array, format new pool device.

 

To be honest I wasn't aware that my cache drives were a single point of failure.  Seems obvious now, but I learn as I go.  

 

I had 2 240GB SSDs and had dedicated one to VMs - the one that failed, and one to the other stuff like appdata, system, etc.  I had hoped that I could do something with the spare SSD I have but it is too small to accommodate my VMs.  So I ordered two 1TB SSDs that are arriving today.

 

Here is my plan:

 

The array is running right now and I have backups of my VMs.

 

Install the first 1TB in place of the failed drive.

Restore the VMs to this drive.

Move all data from the existing 240GB SSD to this drive.

Take our the 240GB and replace it with the second 1TB.

Mirror the second 1TB to the first 1TB.

 

Is my plan a good approach?  This then should remove my single point of failure?  Thanks.

[itimpi]

A lot of people run with a single drive in the cache pool that holds the appdata share and then install the Appdata Backup plugin to make regular scheduled backups to the main array.

[The Transplant]

6 minutes ago, itimpi said:

A lot of people run with a single drive in the cache pool that holds the appdata share and then install the Appdata Backup plugin to make regular scheduled backups to the main array.

Which is essentially what I was doing.  But if I implement my approach then I can (I think) prevent downtime from a cache drive failure.  Not that my system is mission critical but it did take out my Home Assistant which is a little annoying.