Convert Cache from RFS to XFS - Worked I Think??

[spuddo]

Hi,

 

I recently sought advice on changing my 60GB SSD Cache from RFS to XFS. Many thanks to bjp999 and ronj for their advice.

 

The conversion is complete and it seems to be working but some quirky things happened during the process.

 

I thought I might describe the process so that wiser people than I may be able to help others learn from this, or maybe I have created and underlying problem not yet obvious.

 

I have Docker running with one Container for Plex

 

On the 60GB cache drive I had a 10GB Docker image and a cache only share "apps" to store the Plex library.

 

I should say that I was doing this to avoid having to recreate the Plex library.

 

This was the process employed:

 

Create ‘t’ folder on disk 1. (This was used to store the cache data whilst the Cache disk is reformatted)

 

Stop Plex. Navigate to Docker tab. Click on Plex Media Server icon, select Stop.

 

Stop Docker. Navigate to Settings tab. Click on System Settings>Docker. Set Enable Docker to No.

 

Delete Docker image file. Click on the checkbox next to Docker image that says Delete Image File. Click the Delete button after clicking the checkbox to delete the image.

 

Logon to the sever. I use Telnet

 

Open a screen session

 

Copy the existing cache only share which contains the Plex library to the previously created “t” folder on disk1. (I use rsync to copy files)

 

Test copy first  rsync –navPX /mnt/cache/ /mnt/disk1/t

 

Copy (remove –n option)  rsync –avPX /mnt/cache/ /mnt/disk1/t

 

Check copy worked rsync -nrcvPX /mnt/cache /mnt/disk1/t

 

At this point I got the first unexpected message. Many, many instances of 'skipping non-regular" file

 

Googling revealed that this was likely due to the files being symbolic links and that these could be moved by adding a -l option to the rysnc command.

 

Undaunted or perhaps unwisely I continued and added the - l option to correct the errored files.

 

Corrected with this rsync -lrcvPX /mnt/cache /mnt/disk1/t

 

That seemed to work and moved the files.

 

STOP the ARRAY

 

Second unexpected problem. unRAID just hung.  After a very long wait I decided to press and hold the Power Button to power off the server. (I know but what else?)

 

Powered up the server again and of course it came up as stopped but otherwise looked fine.

 

Still undaunted or perhaps even less wisely I decided to continue.

 

With array stopped, changed the filesystem on cache to XFS

 

Started the array, the Cache disk showed unformatted, all the other disks were formatted.

 

The server initiated a Parity Check - no doubt due to the improper shutdown.

 

I elected to Format the unformatted cache disk with XFS

 

Restore the stored [disk1/t] cache only share which contains the Plex library to the Cache disk. This time I included the -l option

 

Test copy first  rsync –lnavPX /mnt/ disk1/t /mnt/ cache

 

No errors

 

Copy (remove –n option)  rsync –lavPX /mnt/ disk1/t /mnt/ cache

 

Check copy worked rsync -lnrcvPX /mnt/dik1/t /mnt/cache

 

No errors

 

Delete the disk1/t folder

 

Re enable Docker. Navigate to Settings tab. Click on System Settings>Docker Set Enable Docker to Yes

 

Image size 10Gb

 

Docker image /mnt/cache/docker.img

 

With Docker restarted I re-downloaded Plex (needo)

 

Volume Mappings

/config  /mnt/cache/apps/Plex/ (as before)

 

unraid_media  /mnt/user/media (as before)

 

All seems to work fine. Parity Check still running.

 

Have I created any unknown problems? Who knows.

 

Would I do this again? Nope. I would just format the Cache disk and resinstall Docker, Plex and rebuild the library.  Might take a few hours but the process I used scares me.

 

Very interested in your views.