Incorrect free space on btfs catch pool

[RIzzaz]

Hi there, 

Not finding any answers to this one

 

I have a nvme drive in a PCI express adapter as a btfs cache pool. All appears ok and I'm mainly running a debian VM on the pool, however the free space on the cach keeps going down, it currently reports around 150gb free on a 2 TB drive when the total partitions on the debian os adds up to around 600gb..

 

I've tried balance, scrub and manual trim command, which all seemed to work, however no change on the amount of free space,

 

Any ideas?? 

 

see attached diagnostics, screen grab of share, and the VM which you can see is only using 727G

 

 

 

 

 

diagnostics-20221023-1316.zip

Edited October 23, 2022 by RIzzaz

[itimpi]

Are you sure none of your shares are using it as a cache?   it might be worth going to the Shares tab, pressing the Compute All button and posting a screenshot of the result.
 

You are also likely to get better informed feedback if you attach your system’s diagnostics zip file to your next post in this thread. 

[RIzzaz]

I have edited the original post to include screen shots and diagnostics, any help appreciated 

 

Thanks

[JorgeB]

Vdisks can grow with time if not trimmed/unmapped, see here, it's for a Windows VMs but same principle applies, there are also reports that defragmenting the filesystem also helps, but don't do that if you use snapshots, another option is to move the vdisk elsewhere then move back with cp --sparse=always.

[RIzzaz]

I have manually run the trim command which did ssem to work, although made little difference to the free space reported,

 

It's a debian VM, just installed. 

 

I added the discard='unmap'  option as per that link, booted the VM, that didn't seem to make any difference

I ran the trim command from within unraid, which worked however trimmed very little

 

I then thought to run trim from directly within the OS and that has cut back to about 1.1tb which is good, still more than the 700 or so but good, I'm running a balance as it now said 60% ? 99% before, I'll try trim again in the os after that, 

 

does this make sense, should i have to set up trim manually in the VM OS, I had the impression unraid now handles this ?

 

Thanks

[JorgeB]

19 hours ago, RIzzaz said:

I then thought to run trim from directly within the OS and that has cut back to about 1.1tb which is good

Yes, this is normal, you should schedule the trim inside the VM OS, or run it manually when needed.