Cache upgrade did not automatically invoke btrfs replace?

[Kyle W]

I have a 500gb SATA SSD that was being utilized as a cache drive when I noticed an error stating that TRIM was not enabled. After some digging it sounds like it is because TRIM is not supported by the firmware version of my Dell H310 SAS card.

 

Rather than bother flashing/downgrading the firmware, I ordered and installed a 1tb NVMe drive to replace my cache pool and followed the procedure here:

 

After doing so, it doesn't appear that anything has been transferred from the old cache pool device to the new one. My Docker/VMs are all missing.

 

Any advice on how to transfer these? Attached diagnostics. Currently the old cache drive is unassigned.

 

I have not touched anything yet because I'm pretty new to working with Unraid, only having had this machine up and running for about two weeks. I found the Unassigned Devices plugin and thought that could be helpful for mounting the drive and copying the contents of the old cache, but please correct me if there's a better/more correct way.

 

Thanks!

 

Edit: doing some more digging, looks like my answer could be to use the command:

 

btrfs replace start -f /dev/sdX1 /dev/sdY1 /mnt/cache

 

Sourced from here:

Can anyone confirm before I execute? I think at this point I just need to figure out the device IDs (edit: I'm blind). Will I need the UD plugin for that command to work properly?

 

 

Looking at my devices it looks like I should be running:

btrfs replace start -f /dev/sdb /dev/nvme0n1p1 /mnt/cache

 

Are there any other steps I'm missing, like disabling Docker and the VM manager? Should I leave the NVMe in the cache pool while running this command?

 

server-diagnostics-20221127-1852.zip

 

Edited November 28, 2022 by Kyle W

[JorgeB]

Replace only works for multi device pools, please post the output of

btrfs fi show

 

[Kyle W]

Dang, I thought I was close!

 

Results of btrfs fi show:

 

[Kyle W]

I tried downloading Unassigned Devices to explore the old cache drive but the mount button is grayed out so I'm not sure what to do.

 

[JorgeB]

No btrfs filesystem exists on old cache, possibly it's been wiped, try this:

 

btrfs-select-super -s 1 /dev/sdb1

 

If the command is successful post again the output of

btrfs fi show

 

 

[Kyle W]

Fortunately it doesn't appear to have been wiped:

 

 

The mount button is no longer grayed out with Unassigned Devices but the partition size says 0B. Typing this from my phone but I'll take a closer look soon.

Edited November 29, 2022 by Kyle W

[JorgeB]

7 minutes ago, Kyle W said:

Fortunately it doesn't appear to have been wiped:

It was wiped, but the select-super command restored a backup superblock, if you assign it to a pool it should mount, you can then add the new device to the pool and once that's done, remove the old one.

[Kyle W]

So, shut down the array and add it to a new pool?

 

When both devices are in the pool what happens at that point, will the data be restored to the new device?

[Kyle W]

Update, I added it to a new pool named testrestore and I can see my old data, thank you!

 

Now can I just transfer this to the new drive in the other cache pool to restore my Docker/VMs?

[JorgeB]

You can, or like mentioned you can add the new device to the same pool, by default Unraid will create a raid1 pool, then remove the old device.

[Kyle W]

Alright, I'll play around with it some more and see if I can get things back to where they were. I downloaded Dynamix File Manager to make a backup of the old cache on my array and will copy it to the new cache from there as well.

 

Thanks for all of your help!

 

Edit: it worked! All Docker/VMs restored and working!

Edited November 29, 2022 by Kyle W