I currently have a 13+yr running home server setup with OMV on an i3-2100T system with a 240gb m.2 drive and 2x 12tb & 2x 8tb hdds. In the past year I've setup a separate proxmox hypervisor (12th gen i5 with 128gb RAM) to run all compute tasks and have simply used the older OMV system for storage (without the parity insurance). I'm looking to ditch OMV and use unraid, NAS only, and would appreciate any input from the community.

After a bit of research my plan is to use xfs for both the cache and the hdd array, use one of the 12tb as parity and allocate the remaining 3 disks (12+8+8tb) as three distinct drives in the array (not a pool?). The three disks will be personal photos/videos/owncloud storage (12tb), media/arr (8tb), and personal business/document backup (8tb). I have a few questions assuming this is the best starting point:

1. Is the 240gb NVME enough for simple initial cache storage before transferring to the array? Is there a real time schedular that monitors folders on the cache drive to immediately send to the array, or is this a function that happens nightly/weekly/etc?

2. If the cache drive is scheduled for a later transfer to the array, how do I access the data between initial cache and transfer? Is the data tagged somehow with the ultimate destination drive or is there a waiting period for accessing until it has transferred to the array?

3. In proxmox is it better to use NFS or samba shares?

4. What are my blindspots to using unraid for this solution and is there a better initial setup for both simplicity and reliability?

I appreciate any insight here as I'm either over or underthinking this 😜

5 hours ago, tstack77 said:

Is the 240gb NVME enough for simple initial cache storage before transferring to the array? Is there a real time schedular that monitors folders on the cache drive to immediately send to the array, or is this a function that happens nightly/weekly/etc?

If you set up Unraid shares to use the cache drive for writes, any new files will be written first to the cache drive. Then the 'Mover' will move the files to the array on predetermined interval. See below:

As to whether a 250GB drive is large enough is really a question that only you can answer. Normally, the transfer fro the cache to the array is scheduled to the array takes place at night. (This is when most servers are generally idle.) IF you normally upload less then about 150GB per day, you should be fine. BTW, upgrading a cache drive is not that difficult. So don't feel you have get a monster one to begin with.

IF you have a lot of data to transfer, you would not use the cache until that is completed. Second, you might also consider not assigning a parity drive until you have done the initial transfer as it will go faster.

5 hours ago, tstack77 said:

If the cache drive is scheduled for a later transfer to the array, how do I access the data between initial cache and transfer? Is the data tagged somehow with the ultimate destination drive or is there a waiting period for accessing until it has transferred to the array?

If you are using the setup suggested above,any files on the cache drive are automatically a part of the share. So you have access to them the second that the write operation is completed.

Thank you, Frank! I started the process a few hours ago and already canceled the first parity sync as it was going to take 13 hours to confirm nothing was on two drives lol...great idea to have simply not assigned a parity drive to begin with. In the meantime I've discovered that it's important to also first install plugins and learned of the great preinstalled tool of mc (midnightcommander). The documentation of unraid is missing a comprehensive tutorial (that is easy to find) for someone looking to migrate over from another service... even their getting started page has broken links (Learn more about storage management) 🤪

I look forward to discovering how the cache files are automatically part of the share as I have yet to come to that part of the process, thank you again for the insight!

Have a look through these sections of the Docs:

https://docs.unraid.net/unraid-os/using-unraid-to/manage-storage/shares/#primary-and-secondary-storage

and

https://docs.unraid.net/unraid-os/using-unraid-to/manage-storage/array-configuration/

The Docs are not the strongest feature of Unraid. I have been around Unraid for about 14 years now and I find them hard to use when want to find material for folks (like yourself)who have questions.

"Cache to array

Use case: Moving files from your high-speed cache to the array for long-term storage.

1. In the share settings, set Primary Storage to Cache and Secondary Storage to Array.

2. Set the Mover action to Cache -> Array.

3. The Mover runs automatically based on your schedule to transfer files. Manual Mover runs are possible but not necessary. "

Ok, this is new and worth exploring while I wait the 15 hours for my 8tb drive to transfer to an xfs disk. I'm halfway understanding this and trying to document my situation for future readers. I clearly need to read up on "share settings" and will provide future edits below as I come across questions (might be a bit of stream of consciousness) and provide clarification once I have a better understanding of the machinations here 👍

Edit 1: I am still a bit lost on the difference between array and pool...not that I don't get the idea behind what they offer, but why is a "pool" different than simply having an array cache?

"Allocation method

When you create a new user share or add any file or folder inside it, the system needs to decide where to store this data. Usually, it will choose the data disk that has the most free space available. However, you can adjust how this decision is made using certain settings.

There are three straightforward options you can select for how the system allocates storage for your user share:

• High-Water (Recommended)

• Most-Free

• Fill-Up

How it works

The High-water method progressively fills disks using "switch points" based on half the largest drive's capacity. This balances disk usage while keeping related files together and minimizing drive spin-ups.

Example with 8TB, 3TB, and 2TB drives

1. First pass: Fill 8TB drive until 4TB free remains (half of 8TB).

2. Second pass: Fill 8TB/3TB drives until 2TB free remains.

3. Third pass: Fill all drives until 1TB free remains."

Edit 2: Ok now thoroughly confused as I'm ultimately going for 3 separate disks and not a single conglomeration (pool?). Am I missing the point of unraid if I want to have an easy rebuild point of each disk in case of failure?

Edited September 12, 2025Sep 12 by tstack77

50 minutes ago, tstack77 said:

Edit 1: I am still a bit lost on the difference between array and pool...not that I don't get the idea behind what they offer, but why is a "pool" different than simply having an array cache?

The array refers to the traditional way of storing files in Unraid. Before cache and pools, there was only the array. Each drive in the array (except parity) has its own individual file system and operates independently with real-time parity updates and no striping. Then cache was introduced as a way to leverage the higher write speed, so the files would first go to the cache drive, then be offloaded to the array, typically during nighttime when the server is idle.

Then the ability to add cache drives to a pool became available. The cache is now essentially a pool of either single or multiple drives. If multiple drives, it has to be in either the BTRFS or ZFS file system, which then forms a RAID.

So the two ways of storing files are now boiling down to a couple of things:

1. Array is for slow writes and a different parity protection system. If more than 2 (max) drives fail at the same time, you lose only data on those drives and not the rest of the array. And of course, mixing of drives with different sizes.

2. The pool(s) are for faster write speed and a parity protection system, where if more than 3 (max) drives fail at the same time, you lose not only the data on those drives but ALL your data.

So it's essentially 2 different use cases.

53 minutes ago, tstack77 said:

Edit 2: Ok now thoroughly confused as I'm ultimately going for 3 separate disks and not a single conglomeration (pool?). Am I missing the point of unraid if I want to have an easy rebuild point of each disk in case of failure

The allocation method is just a set of rules for the way you want to fill up your drives in the ARRAY. As they are individual drives with an independent file system, they can be written to in different ways. Either balanced (high water), write to the drive with the most free space, or simply just fill up one drive at a time until your most free space setting is reached before it moves to the next drive.

In a pool, all files are striped over all drives and will fill up the same, so the allocation method doesn't apply; it only applies to drives in the array.

Hope this makes sense :)

Edited September 12, 2025Sep 12 by strike

Thank you Strike, that explanation helps a lot. Still one thing I'm trying to fully understand.

On 9/11/2025 at 11:26 PM, strike said:

The allocation method is just a set of rules for the way you want to fill up your drives in the ARRAY. As they are individual drives with an independent file system, they can be written to in different ways. Either balanced (high water), write to the drive with the most free space, or simply just fill up one drive at a time until your most free space setting is reached before it moves to the next drive.

In a pool, all files are striped over all drives and will fill up the same, so the allocation method doesn't apply; it only applies to drives in the array.

Is there another option to where I simply want to fill up the array drives based on a folder structure in cache? Let's say I have three drives where #1 is pictures, #2 videos, and #3 apps. Can I write to a picture folder in cache where it will then always write to the picture drive in the array? I'm coming from a long history of muscle memory of manually organizing my data this way and I still like the idea in case I ever need to pull a drive for some reason and have it still work in a different system.

I'm sure this will all make complete sense once I'm actually up and running (wasted 2 days copying files to a new 12tb drive only to find out that midnight commander (mc) didn't actually complete a copy as I didn't stay logged into the session (all good now after reading up on tmux and terminal sessions 😆)

4 hours ago, tstack77 said:

Can I write to a picture folder in cache where it will then always write to the picture drive in the array?

You can if you use separate shares for them; they can then have specific settings, including which disks they can use.