Re: changing 15.0 default zroot zfs datasets and partitions?

There was time when differences between beadm(8) and bectl(8) were quite large ... but after some bugs filled and solved they do (fortunately) mostly the same.

The main differences right now are:

beadm(8):
- written in POSIX sh(1) and awk(1)
- is NOT in the Base System (its in the FreeBSD Ports and/or packages)
- provides 'chroot' option
- provides 'reroot' option

bectl(8):
- written in C
- is in the Base System
- provides 'jail' option

Hope that helps :)

Regards,
vermaden






Temat: Re: changing 15.0 default zroot zfs datasets and partitions?
Data: 2025-12-31 2:55
Nadawca: "Karl Denninger" <karl@denninger.net>
Adresat: freebsd-hackers@freebsd.org; 



> This probably explains why I don't have issues with this stuff...
> I use beadm. :-)
> 
> If its under the root it is part of the boot environment.
> If not, then not.
> 
> 
> On 12/30/2025 8:44 PM, vermaden wrote:
> 
>> 
>>> 
>>> Following up to myself after reading bectl(8) more thoroughly:
it seems like reorganizing the zroot into a 'deep boot
environment' (vs. default 'shallow') and maybe using
canmount=noauto (instead of canmount=off) could be a way
to declare the members of the BE, and affect what shows
up in various traditional Unix command outputs.

>>> 
>> 
>> Funny thing.

The way I wrote/designed beadm(8) - the bectl(8) precursor - was that no matter how many additional ZFS datasets (and how many levels) were all treated the same. I just created a recursive ZFS snapshot/clone.

When bectl(8) was implemented someone assumed that only one 'dataset' is there to cover it.

After some bugs filled that beadm(8) works well and bectl(8) does not someone figured out what is the problem and they came with 'deep' ZFS BE concept naming ... something that would not exist if they would implement bectl(8) 1:1 to beadm(8) in the first place :)

Regards,
vermaden


>> 
> 
> -- 
> Karl Denninger
> karl@denninger.net
> The Market Ticker
> [S/MIME encrypted email preferred]
>