Cache drive showing full but it's not on identical drives

[djvj]

I have 2 identical SSD drives in my cache pool in Raid-1. GUI showing they are full, but over CLI, they are not.

Using UnRaid v6.8.3

 

Why is this happening and how can I go about fixing it? Tried searching and I find threads on this issue with drives are not identical.

 

root@Storinator:/mnt/cache# du -h -d 1 /mnt/cache
60G     /mnt/cache/docker
84G     /mnt/cache/domains
6.1G    /mnt/cache/isos
1.0G    /mnt/cache/system
96G     /mnt/cache/appdata
0       /mnt/cache/downloads
171M    /mnt/cache/.PhAzE-Common
246G    /mnt/cache

Edited July 19, 2020 by djvj

[JorgeB]

du is not reliable with btrfs, GUI will show the correct used/free space (with equal size devices), if you have Windows VMs this should help to recover some space.

[djvj]

Thank you.

 

I have a win7 VM, but that still shows 50GB vdisk size and has never changed. This isn't what's been taking up more and more space gradually then. Your guide is more for win8/10. I tried for win7 but defrag found 0% defragmentation.

 

If the Unraid GUI is correct, then what tool can I use to show the correct space usage on each folder so I can find the culprit?

 

 

[JonathanM]

28 minutes ago, djvj said:

I have a win7 VM, but that still shows 50GB vdisk size and has never changed. This isn't what's been taking up more and more space gradually then.

Depending on how you are looking at it, a raw vdisk CAN and will gradually take up more space, all while showing the final allocated size. VM raw vdisk files are created as sparse, which show the full size, but only occupy what is actually used. Then as more and more writes occur, there will be less and less difference between the two sizes until finally the file will occupy the full allocation. That will happen regardless of actually how much data is IN the VM, unless you are using an OS that understands how to manage that type of situation and it's configured correctly. Windows 7 doesn't know how to properly manage its data automatically, so you have to do things like defragment and fill the empty space with zeroes using a third party utility, then resparsify the vdisk file.

 

To see the actual usage vs total allowed, you can use the du command. At the console or SSH session, cd into the folder that holds your image file, probably /mnt/user/domains/Windows \7 or something like that, and use du -h followed by du -h --apparent-size.

That will give you what's actually in use compared to the final allocated size.

[djvj]

Thanks for explaining the differences.

 

root@Storinator:/mnt/user/domains/Windows 7# du -h
49G     .
root@Storinator:/mnt/user/domains/Windows 7# du -h --apparent-size
51G     .
 

My concern isn't the actual usage of the VMs, it's the discrepancy of my cache disk size vs free space and figuring out where that free space is being used up. In this case, it isn't the Windows 7 VM alone using up that much space. I have 100GB worth of VM disk usage among a few VMs. This still leaves 380 of 480GB left. DU only shows 246GB used which 84 of that says its from domains. 480 - 246 (from DU) = 234GB is somewhere and I don't know where!

 

For instance on Windows, I could just look at the properties of a folder and find it's exact usage, amount of files/folders/etc. Then trace out where my space is being used up. Then decide is I need to move it to another drive, delete, etc. If I can't trust DU command because cache is btrfs, what can I trust to figuring this out?