I'm building a new "proper" home server and rather than slap shit together just enough that it works I want to do it right, so naturally I'm looking at ZFS.
The plan is 4 HDDs (of some kind, depends on what I'll find second hand) that will make up one pool in RAIDZ1 and actually store the data I care about, and a boot m.2 SSD running ext4 dedicated for boot and programs installed on the machine.

I have a few ZFS questions I want to clear up.
>datasets
In ZFS datasets are essentially a sort-of partition that sits in a pool, is more flexible than an actual partition, can have its own settings (which can be changed easily later) and from a daily-use perspective just look like directories, right?
So if my data can be divided into media (videos and images) and text-based files (source code, markdown-based notebooks, office documents) I could create a 'media' and a 'docs' datasets, have each have their own settings, and if I want to access a specific movie the path will be something like /media/movies/die_hard_7_the_hardening/, right?
>Compression
From what I've read compression has virtually no drawbacks when a decent CPU is involved and I wouldn't really notice a speed difference if I were to use it.
BUT if I have a dataset full of almost-exclusively non-compressible files like media, then I'd have the CPU work for no compression benefit, right? so I might as well turn compression off for a dataset like that
>Resilvering
If a drive dies on me, do I just pull it out, replace it with a new one and it starts the resilvering process automagically? or do I need to tell it to start.
>failure during upgrade
If hypothetically I wanted to increase the size of my pool I'd need to pull out a drive, replace it with a bigger capacity one and resilver it ofor each drive, right?
Say I was in the process of doing that and a drive died.
Would putting the old drive back into the same spot recover my data? (assuming RAIDz1, so 1 drive redundancy)

thanks dudes

>>713447291
nice crops