I've read the wiki on replacing a cache drive but am not sure I understand what is being asked and why it is necessary.

 

I have a 2TB spinning disk for cache formatted with BTRFS. It is in the system and running fine.  I also installed a new 2TB SSD M.2 drive that I want to move everything from cache over to.  I have VM, docker, mysql, etc on the cache drive.

 

I was planning on stopping VM, dockers, and the array.  

format the new drive

copy everything from the old drive to the new drive

assigning the new drive as cache

starting everything back up.

 

The wiki talks about using mover to move things, or adding the new drive to the cache pool and having balancer do things.  I assume the default setting for a cache pool is to mirror the data across both disks?

 

I'd prefer doing this the "recommended way" but have concerns.  I really don't want to screw it up and have to rebuild everything from scratch.

 

thanks,

david

Hi,

 

I've used this methodology 2x without issue.

 

Good luck,

cesman

You can also do an online replacement, but note the bold parts about single device replacement.

5 hours ago, JorgeB said:

You can also do an online replacement, but note the bold parts about single device replacement.

I backed up the drive using the CA plugin, then started this last night.  It is now about 93% done.

 

thanks

the balancing finished.  I then stopped all dockers, stopped the array.  Changed the cache to only have the new drive and restarted the array.

 

It is now doing a balancing again.  Is this expected?

 

I can see the data mounted.

Edited January 5, 20215 yr by lovingHDTV

4 minutes ago, lovingHDTV said:

Changed the cache to only have the new drive and restarted the array.

That's not the procedure I linked.

But you can post the diags if you want to check current status.

OK I guess I really screwed things up.

 

exactly what I didn't want to do.

 

Here is what cache is saying now.  The second balance happened, but btrfs is still running keeping me from stopping the array.

 

 

It also now says single instead of Raid1.  I can't show a picture of the main page because I can't stop the array.  But I have cache pool set to 2 slots, and the first slot is this drive, and the second is set to no device.

 

How do I recover?  The system is running, but I see a lot of these in the log file:

 

For clarity sake, the old cache drive is sdf.

Jan 5 15:23:58 tower kernel: BTRFS info (device sdf1): relocating block group 4975807692800 flags data
Jan 5 15:24:14 tower kernel: BTRFS info (device sdf1): found 8 extents
Jan 5 15:24:15 tower kernel: BTRFS info (device sdf1): found 8 extents
Jan 5 15:24:15 tower kernel: BTRFS info (device sdf1): relocating block group 4973660209152 flags data
Jan 5 15:24:31 tower kernel: BTRFS info (device sdf1): found 8 extents
Jan 5 15:24:32 tower kernel: BTRFS info (device sdf1): found 8 extents
Jan 5 15:24:32 tower kernel: BTRFS info (device sdf1): relocating block group 4971512725504 flags data
Jan 5 15:24:48 tower kernel: BTRFS info (device sdf1): found 8 extents
Jan 5 15:24:48 tower kernel: BTRFS info (device sdf1): found 8 extents
Jan 5 15:24:49 tower kernel: BTRFS info (device sdf1): relocating block group 4969365241856 flags data

 

thanks

david

tower-diagnostics-20210105-1525.zip

Edited January 5, 20215 yr by lovingHDTV

Convert the pool to raid1 then remove the old device.