Few newbie questions I haven't found an answer for

Status
Not open for further replies.

Montecrysto

Cadet
Joined
Nov 22, 2018
Messages
3
Hi folks!
I just started to learn how to use FreeNAS, the best practices of the many things you can do with it and, while toying with my new creature, I asked myself (and tried to find the answers on the web) these three things:

  1. It's better to share the "root" dataset or all of its children? E.g. pool/users vs pool/users/user1, pool/users/user2
  2. Can I have a dataset with UNIX/Mac as share type inside a dataset with Windows share type which is being shared with SMB? If yes, are there any problems I may occur in?
  3. What are the differences between creating a home share via the webUI during the setup process of a SMB share versus manually creating a dataset (allowing me to manage their quotas) with all its users subfolders and then linking each one of them to the specific user? The only difference I can think of is that with this last one method you can see other users folders, but you can still avoid it disabling the "Browsable to Network Clients" checkbox, can't you?
---

Don't know if that may help you personalising your responses, but this is the hierarchy I'm planning for my pool:
  • p0
  • p0/private
    This dataset will contain many other datasets with different permissions, as pointed below. Question n° 1 is about sharing this folder or its subfolders.

  • p0/private/backups
    This dataset will contain at least one time machine backup, so an AFP share with MAC share type, which lead me to ask question n° 2.

  • p0/private/downloads
    This dataset will be used to store downloads from a VM (outside FreeNAS), so it's the only one which will have read/write permissions, while all the others will just have the read one.

  • p0/private/shares
    Only the family users will be able to write/read on this dataset.

  • p0/private/users
    This dataset will contains all the users private folders; question n° 3 is about how to manage the home directories.
  • p0/public
    This dataset will be shared among all the users and will be used as a way to exchange data.
Keep in mind that I have Windows, macOS and Ubuntu machines in my home, so my go-to choice was a SMB share with a Windows machine dedicated for setting all the permissions through properties/security.
If all of this is not recommended or not possible my backup plan is to scrap all the private/public thing and simply have the datasets under the root (p0) and share them acconrdingly as needed (so SMB every single of them but /backups, which will be manually shared as per share type requirements).

---


Thanks for your time, hope all of this actually made sense and wasn't answered elsewhere.
Also sorry for my english, I'm not a native speaker.
 

kdragon75

Wizard
Joined
Aug 7, 2016
Messages
2,457
It's better to share the "root" dataset or all of its children? E.g. pool/users vs pool/users/user1, pool/users/user2
Idealy each share will be its own dataset. This allow much greater flexibility is setting ZFS options where needed. For example, you may have a share that's mostly office files that compress well and another for video archiving that does not compress. Here you could selectively enable compression. Another use case would be setting quotas. There are several other use cases.

Can I have a dataset with UNIX/Mac as share type inside a dataset with Windows share type which is being shared with SMB? If yes, are there any problems I may occur in?
I would have to play around and see how this works but I imagine that it would be just fine. Windows mode uses Access control lists (Like NTFS permissions) and Unix uses simple owner:group:everyone permissions.
What are the differences between creating a home share via the webUI during the setup process of a SMB share versus manually creating a dataset (allowing me to manage their quotas) with all its users subfolders and then linking each one of them to the specific user? The only difference I can think of is that with this last one method you can see other users folders, but you can still avoid it disabling the "Browsable to Network Clients" checkbox, can't you?
I haven't played with this but seems like an area where FreeNAS could be vastly improved. I would like to see adding new users auto create new dataset based home folders with definable default quotas.
p0/private/backups
This dataset will contain at least one time machine backup, so an AFP share with MAC share type, which lead me to ask question n° 2.
I would advise having a dataset just for the time machine. Time machine will fill the disk and this can be limited by ZFS quota.
 

Montecrysto

Cadet
Joined
Nov 22, 2018
Messages
3
Idealy each share will be its own dataset. This allow much greater flexibility is setting ZFS options where needed. For example, you may have a share that's mostly office files that compress well and another for video archiving that does not compress. Here you could selectively enable compression. Another use case would be setting quotas. There are several other use cases.
Shouldn't I still retain the ability to customise the compression (and other ZFS options) sharing the entire "root" of the dataset instead of the single datasets on their own?
Maybe I didn't explain myself that well in the first place. On my attached screenshot ( Screenshot 2018-11-22 at 21.30.11.png ) you can see d0, e0 and e1 and everyone of them has a different compression (I'm just toying); what I was asking is: creating a SMB share of d0 is any different from creating two different SMB shares of e0 and e1?

I would have to play around and see how this works but I imagine that it would be just fine. Windows mode uses Access control lists (Like NTFS permissions) and Unix uses simple owner:group:everyone permissions.
Okay, cool. From what I've seen in these days it plays nice (or at least I haven't encountered an error yet :D).

I haven't played with this but seems like an area where FreeNAS could be vastly improved. I would like to see adding new users auto create new dataset based home folders with definable default quotas.
I like your idea a lot! Also I agree with you: at the moment it seems like one of the most improvable features. I think I'll go with the manual approach because being able to set the user's quota is something I wanna be able to.

I would advise having a dataset just for the time machine. Time machine will fill the disk and this can be limited by ZFS quota.
Yeah, that's my plan. For simplicity reasons I just listed the dataset as p0/private/backups, but on my papers it contains other datasets, like the time machine one (e.g. p0/private/backups/macbook).
 
Last edited:

Montecrysto

Cadet
Joined
Nov 22, 2018
Messages
3
Is anyone else able to provide some insight on the above queries? Specifically in the regards of the first question.

As always, thanks your time!
 
Status
Not open for further replies.
Top