Complete Noob

Further to my 1st post in the 'Complete Noob' forum section, the Acer Aspire notebook I'm using to learn is set up with ZSF whereas the Sony Vaio is USF. TBH honest that seemed to be the default selections.

Question: my goal is to get to a Desktop PC[notebook for now] kind-of 'daily driver', so is ZSF really the way to go?

I ask as, if I can become proficient enough with FreeBSD I'd set my sights on converting one of my Desktop machines (HP Prodesk)...maybe even my Mac that currently runs Manjaro Linux... So given sufficient capacity disks I might not need to trouble over the FileSystem again!
 
Question: my goal is to get to a Desktop PC[notebook for now] kind-of 'daily driver', so is ZSF really the way to go?
To be honest, I really do think that ZFS is the way to go - the nice part is that if you give the whole disk to ZFS, the FreeBSD installer will give you some sensible defaults, but you're not stuck with them - you can adjust anything you can think of AFTER the installation. No partitioning required, either. ZFS does have a ton of features, but it's OK if you don't know about most of them - you can leave those advanced features alone, and go with a set of very simple detaults that the installer offers. And you're not stuck with the decision you make.

WIth UFS, you do have to make some decisions about partitioning the disk BEFORE you start the installation, and to accept the idea that if you want to tweak those details later, that does mean a complete reinstall of the whole machine.

Well, it's your metal, you ultimately decide what works for you.

I've been daily-driving FreeBSD with ZFS since it first came out in 2017, and have no complaints about ZFS...
 
To be honest, I really do think that ZFS is the way to go - the nice part is that if you give the whole disk to ZFS, the FreeBSD installer will give you some sensible defaults, but you're not stuck with them - you can adjust anything you can think of AFTER the installation. No partitioning required, either. ZFS does have a ton of features, but it's OK if you don't know about most of them - you can leave those advanced features alone, and go with a set of very simple detaults that the installer offers. And you're not stuck with the decision you make.

WIth UFS, you do have to make some decisions about partitioning the disk BEFORE you start the installation, and to accept the idea that if you want to tweak those details later, that does mean a complete reinstall of the whole machine.

Well, it's your metal, you ultimately decide what works for you.

I've been daily-driving FreeBSD with ZFS since it first came out in 2017, and have no complaints about ZFS...
Hi, thank you. Yes, I had a look at how ZFS has sliced things up and it does look comprehensive and sensible, even has a useful 'swap' allocation. I'll read up on the subject when I've got further with setting my notebook up. Thank you again. 🙂
 
Question: my goal is to get to a Desktop PC[notebook for now] kind-of 'daily driver', so is ZSF really the way to go?
You seem to be beyond the very basics about how to deal with the system, so I don't see no reason, why not.
So given sufficient capacity disks I might not need to trouble over the FileSystem again!
With ZFS you don't need to bother about max. drive size, max. file size, max. number of files - that was one of the core points when ZFS was developed in the first place: practically infinite storage addressable.
And it brings a lot many other benefits.
The core difference to other FS is, it's not working with partitions, but with pools, consisting of partitions, or drives. You can add more drives, or partitions to a pool any time (while there are some restrictions: Having a mirror of two 1TB drives for example, then trying to add a 512G drive to it will be rejected, of course, because shrinking poolsizes is not possible, while growing pools is no big deal - just add drives [depending on the kind of pool, of course, but I don't want to go into details.])
But you can exchange drives - defected, or larger ones, when you are using pools of more than one drive, any time, so in theory no need to move a filesystem to other drives anymore, while it can stay forever on the same pool.

It's recommended to use lables, which make additions and exchanges a lot easier.

But you may consider of doing raid-pools (except stripe, I very discourage to use that kind, except maybe for very special temporary exception purposes, if even), and of course adapt your backup plan.
Even while ZFS is a very stable and reliable FS (by now I never had any data losses by power outages; but my small systems, and data volumes I deal with are no reference) it does not protect you from hardware failures, and stupidity - the #1 reason for to recover from backups.😁
 
And what makes ZFS "looking difficult to use" may be because it's consisted with different concept.
This synonyms made me understand ZFS easier:
  • BSD slice in MBR drives = ZFS pool
  • BSD partitions in BSD slice = ZFS dataset
  • filesystem = dataset with required options (including inherited ones)
When installing 15.0-BETA4, I get the "STABLE" output instead of "BETA4."
$ uname -apKU
Code:
FreeBSD hk 15.0-STABLE FreeBSD 15.0-STABLE stable/15-n281026-e2a87c08171a GENERIC amd64 amd64 1500501 1500501
Also, to install 15.0-BETA4 WITHOUT partitioning, should I, as far as I understand,
select manual partitioning instead of automatic partitioning, or go to the shell and configure something there?
I still don't understand how I can start the process WITHOUT partitioning the disk,
and completely abandon the partitioning paradigm. I need a detailed description
of ZFS (I have the basic knowledge). But how can I do an initial installation directly on ZFS?
I don't see the trigger: it's hard for me to grasp the idea itself.
If anyone knows, could you please provide a roadmap—where should I go when I enter the disk partitioning menu? Should I discard all the "auto" options and go straight to the shell?
Am I understanding the process correctly? Where should I turn?
Thanks.
 
When installing 15.0-BETA4, I get the "CURRENT" output instead of "BETA4."
$ uname -apKU
Code:
FreeBSD hk 15.0-STABLE FreeBSD 15.0-STABLE stable/15-n281026-e2a87c08171a GENERIC amd64 amd64 1500501 1500501
Also, to install 15.0-BETA4 WITHOUT partitioning, should I, as far as I understand,
select manual partitioning instead of automatic partitioning, or go to the shell and configure something there?
I still don't understand how I can start the process WITHOUT partitioning the disk,
and completely abandon the partitioning paradigm. I need a detailed description
of ZFS (I have the basic knowledge). But how can I do an initial installation directly on ZFS?
I don't see the trigger: it's hard for me to grasp the idea itself.
If anyone knows, could you please provide a roadmap—where should I go when I enter the disk partitioning menu? Should I discard all the "auto" options and go straight to the shell?
Am I understanding the process correctly? Where should I turn?
Thanks.
What I can tell is, ugh, "completely forget about the idea!".

To boot, on legacy BIOS boots with GPT (as MBR scheme would too strictly limit the usable size of disk drives),
  • freebsd-boot partition
  • freebsd-zfs or freebsd-ufs partition
are mandatory.
On UEFI boots,
  • ESP (EFI System Partition)
  • freebsd-zfs or freebsd-ufs partition
are mandatory. Without freebsd-boot (for legacy BIOS) or ESP (for UEFI), the BIOS / UEFI firmware cannot kick the installed FreeBSD to boot.

And in both cases, if you want swap, you additionally need freebsd-swap partition at least for Root on ZFS. Possibly swap file in the freebsd-ufs partiton could work (not at all tried), but using swap file on ZFS, which is a memory-guzzler, would be dangerous.
This is because, on quite high memory pressures, ZFS could want memory to write to swap, could cause panic with double faults.
 
You seem to be beyond the very basics about how to deal with the system, so I don't see no reason, why not.

With ZFS you don't need to bother about max. drive size, max. file size, max. number of files - that was one of the core points when ZFS was developed in the first place: practically infinite storage addressable.
And it brings a lot many other benefits.
The core difference to other FS is, it's not working with partitions, but with pools, consisting of partitions, or drives. You can add more drives, or partitions to a pool any time (while there are some restrictions: Having a mirror of two 1TB drives for example, then trying to add a 512G drive to it will be rejected, of course, because shrinking poolsizes is not possible, while growing pools is no big deal - just add drives [depending on the kind of pool, of course, but I don't want to go into details.])
But you can exchange drives - defected, or larger ones, when you are using pools of more than one drive, any time, so in theory no need to move a filesystem to other drives anymore, while it can stay forever on the same pool.

It's recommended to use lables, which make additions and exchanges a lot easier.

But you may consider of doing raid-pools (except stripe, I very discourage to use that kind, except maybe for very special temporary exception purposes, if even), and of course adapt your backup plan.
Even while ZFS is a very stable and reliable FS (by now I never had any data losses by power outages; but my small systems, and data volumes I deal with are no reference) it does not protect you from hardware failures, and stupidity - the #1 reason for to recover from backups.😁
Thank you for this insight. Unfortunately, the installer when I selected ZFS highlights stripe and not knowing any different I took that to be default or 1st choice. So that is what this Acer Aspire notebook is using.
Moreover I'll read more about ZFS before I go for a desktop PC install.
Again, many thanks.
 
Moreover I'll read more about ZFS before I go for a desktop PC install.
Again, many thanks.
No prob. You're welcome.
When I guess right, you're in a test, experimenting, and learning phase. So, no valuable data to be stored there, yet.
If you use one drive for a pool, the kind of pool doesn't matter - it can only be 'stripe', of course.
When I think of ZFS I automatically think of more than one drive used for a pool.:cool:
For that case it's not recommened to chose stripe consist of single drives, since it adds all drives to one large volume. If a single drive fails, the whole pool is killed - no redundancy at all. With the capacity today's drive's provide I don't see no reasonable need to do that.

type of pool(*):min amount partitions(=drives) needed:max. drives fail from n in pool:
stripe10
mirror1n-1
raid104(1/2)n
raidz131
raidz242
raidz353
(*) once it's created the type of a pool cannot be changed.

If I'm allowed to make a suggestion just as an idea for how to set up your desktop, I would recommend to chose a small 2 drives mirror for the system. I have two 256G NVMEs for that, less efi, boot, and swap I have a 217G mirror for the system and userland, except /home/, which still has 173G free.
Then one additional pool for /home, and everything else ("data") For that you need to see yourself how much storage you want for it, how many drives you can use, and what kind of pool you chose for it, etc.
But it's a good idea to keep the system separated from the data, and the system itself doesn't need much space.

You may also find some ZFS calculators on the net, that give you an idea how much storage a pool will provide on a certain type, and drives at which capacity.

And it's very recommendable to use GPT labels.

You will manage it! :cool:
 
but using swap file on ZFS, which is a memory-guzzler, would be dangerous.
This is because, on quite high memory pressures, ZFS could want memory to write to swap, could cause panic with double faults.
Yeah, you're thinking of what you see when you run gpart show. Thing is, at installation, zfs-on-root saves you the thinking efforts, precisely because virtually everything about ZFS can be adjusted later.

ZFS does require 8 GB RAM as a minimum. I have done that, and I have compiled my way into a usable KDE desktop on such a machine. Yeah, that takes awhile. zfs-swap does default to 2 GB, and it's normally enough to actually run a usable system. You really put pressure on RAM when you start complex stuff like compiling stuff from the Ports Collection, esp. big items like Firefox.

itdm5j21 : striping is a fine option for a default. I'd recommend figuring out how it's different from a RAID setup. To be honest, I never figured out how to set up a RAID-based system. It's an interesting rabbit-hole to spend your time on, but for someone who's just starting out with FreeBSD, I'd think of prioritizing the idea of having a running system first, and exploring the features later. And - don't be afraid of doing a complete reinstall of the system. Time-consuming, I know, but there are benefits to that, you learn what works for your hardware and keeps your options open down the road.
 
Back
Top