I'm migrating off a self-managed store to unRAID because, well... I'm tired of maintaining it. However, I am totally lost as to how the storage management works--and yes, I have [read through the documentation](https://docs.unraid.net/unraid-os/manual/storage-management/).

 

Basically, what do I do when I have vastly different storage disks? For example, right now I have:

 

* 128G flash drive that unRAID lives on (Samsung Bar Plus)
* 1TB m2 NVME which has historically been where the operating system resides
* 1TB 2.5" SSD (store thumbnails and other stuff for Immich for fast access)
* 2x 2TB 5400RPM , zfs mirrored (photos using Immich)

* 2TB 5400RPM (frigate clips)
* 2x 14TB EXOs X16s, zfs mirrored (movies, TV shows, etc.)

 

When I initially set this up, I basically modeled it like I would a normal system. However, I'm realizing that this is the "wrong" way of doing this as I have no "array" disks specified so everything is basically a cache.

 

What's the right way to model storage when there are significantly differences in capacity?

Edited May 15, 20251 yr by redmumba

The normal unraid way would be to put all HDD's as array with (one of) the biggest ones as parity for protection. Since that is 14TB, you'd have the leftover drives which is 14TB+3x2TB=20TB as storage. You could set up the media as a folder, or as a share, and the photos as well.

 

Of course if you want to change, you'll need to empty one of the 14TB drives and convert the other because you can't/shouldn't use ZFS as a file-system in the array, so you'd need to find space for the files.

What about the "cache" drives? Based on all the documentation and videos, in 6.x it seems like these were "first class citizens." What does that look like in 7.x? Just a pool named "cache"?

3 hours ago, redmumba said:

What about the "cache" drives? Based on all the documentation and videos, in 6.x it seems like these were "first class citizens." What does that look like in 7.x? Just a pool named "cache"?

Even in the 6.x series the ‘cache’ drive was just a pool that happened to have that name (because for historical reasons that is the default name for the first pool).   The ‘caching’ functionality is configured at the share level and does not care what is the name of the pool being used.   The only difference is that in Unraid v7 it is no longer a requirement to have the Unraid style main array - you can just have pools.   The functionality is the same otherwise.

This has been extremely helpful; as far as the shares go, if I point my docker container at my `media` folder on the array, and I configure the share to use a cache drive, will docker itself read through the cache, or will it always read directly from the underlying disk?

5 hours ago, redmumba said:

if I point my docker container at my `media` folder on the array, and I configure the share to use a cache drive, will docker itself read through the cache, or will it always read directly from the underlying disk?

At any point in time a file is either on the pool set to be used for caching or on the array (never both).   As long as the docker container is configured to use the User Share path ( /mnt/user/media ) then the current location is transparent to a docker container.