Tried to add 2nd cache drive - can't mount cache now

[fysmd]

Unraid pro 6.0.1

I have een running this release for some time and I have just tried to add a 2nd drive to cache.

I reduced the number of slots in my main array to 23, added one to cache pool (making it two).

As soon as I did this, the FS type for my existing drive reported at BTRFS - it's not, it's still RFS.

 

I backed out, removed the 2nd drive and now when I start the array, the cache is reported as not mountable?

can I force the FS type back to RFS?

 

 

[itimpi]

Unraid pro 6.0.1

I have een running this release for some time and I have just tried to add a 2nd drive to cache.

I reduced the number of slots in my main array to 23, added one to cache pool (making it two).

As soon as I did this, the FS type for my existing drive reported at BTRFS - it's not, it's still RFS.

 

I backed out, removed the 2nd drive and now when I start the array, the cache is reported as not mountable?

can I force the FS type back to RFS?

As long as the array is stopped then you should be able to click on the cache drive in the Main tab and change its format.

 

Note that as soon as you want a cache pool (i.e. more than one drive in the cache) then the only supported format is BTRFS which will explain what happened. 

 

To change the cache drive from RFS to BTRFS, you would first need to back up all the files; then stop the array and change the format to BTRFS; start the array which will give you an option to format the drive as BTRFS; copy back the files that should be on the cache.

[fysmd]

That sorted it  - phew!

Thanks.

 

So to switch to a cahce pool is the following a valid approach:

1.  Adjust all shares to NOT use cache

2. Run Mover

3. do as you describe above..

 

I have (probably stupidly) a number of docker images / mounts which address /mnt/cache/blahh directly (I didn't see the point in protecting in the array).  I guess I should change them all to /mnt/user/blahh and set them to cache only if thats still what I want?

[HrtAttK]

I am running through the same process at the moment and since you're in the middle of this does your cache drive (pool?) appear the same way mine does?  The attached is a screen shot of what the system is reporting through the GUI.

[fysmd]

No, I've backed out of cache pool now - I need to make some systems changes before I can re-attempt a pool.

 

So I just ave a single cache drive, using RFS.

[HrtAttK]

Well, if that's the case I am wondering if anyone else viewing would be able to verify if my cache pool is implemented correctly?  Thanks for the replies by the way.

[trurl]

That sorted it  - phew!

Thanks.

 

So to switch to a cahce pool is the following a valid approach:

1.  Adjust all shares to NOT use cache

2. Run Mover

3. do as you describe above..

 

I have (probably stupidly) a number of docker images / mounts which address /mnt/cache/blahh directly (I didn't see the point in protecting in the array).  I guess I should change them all to /mnt/user/blahh and set them to cache only if thats still what I want?

Do you not already have them set to cache-only? I would think so else mover would have already moved them.

 

No need to use /mnt/user/blahh instead of /mnt/cache/blahh. I always use /mnt/cache/blahh when referring to a cache-only share. Note that you still must set the share to cache-only however you refer to it.

[trurl]

I am running through the same process at the moment and since you're in the middle of this does your cache drive (pool?) appear the same way mine does?  The attached is a screen shot of what the system is reporting through the GUI.

Looks normal to me. Very similar to mine.

[itimpi]

1.  Adjust all shares to NOT use cache

2. Run Mover

This would not matter if you are going to put back any files you have copied off.

 

I have (probably stupidly) a number of docker images / mounts which address /mnt/cache/blahh directly (I didn't see the point in protecting in the array).  I guess I should change them all to /mnt/user/blahh and set them to cache only if thats still what I want?

I think it works better if you address the cache drive directly.

 

The thing to do is to turn off docker (and VM's if necessary) while making this change.  Once you have all files back on the redone cache drive you can turn them back on.

[HrtAttK]

Thanks for the clarification, I would have expected the drive labeled as "Cache 2" to show the "Used" and "Free" indicators as well.  I guess the pool reads like a parity drive?  Where as the largest drive would be the only one present?  I installed two 120GB SSDs so if I were to install say another drive that was 256GB the screen would display 256GB as my cache but the total would be 496GB I believe.

[trurl]

Thanks for the clarification, I would have expected the drive labeled as "Cache 2" to show the "Used" and "Free" indicators as well.  I guess the pool reads like a parity drive?  Where as the largest drive would be the only one present?  I installed two 120GB SSDs so if I were to install say another drive that was 256GB the screen would display 256GB as my cache but the total would be 496GB I believe.

It is actually a btrfs raid1 (mirror) rather than parity, though the two can be equivalent in the case of only 2 drives. In some ways the numbers displayed can be confusing and maybe the GUI could be reworked to clear this up a little. Even though you have 240 total, 120 is all you can use because the other 120 is a mirror of the first.

 

With a 256 added to the 2x120, you would have 240 usable. See here

[HrtAttK]

Ah, I see.  Thanks for the link to the website and information!