Generally I would expect somebody who is new to OpenBSD to do a scratch
install on a spare machine to try out the system, but then re-install at
least once rather than continue using that scratch system as their 'final'
installation. For nothing else, because generally people fiddle with all
sorts of settings that have effects beyond what they realise, (including
security related effects), and it's not always obvious how to twiddle the
knobs back. Also, some people install without FDE and later want it. A
re-install is not a bad thing.
(In fact, part of my job is to re-install our servers from scratch for every
release - we never use the upgrade tools.)
The issue at hand is the limitation of time. Personally, I have made
irreversible installation errors on operating systems that I currently
use. However, I am unable to afford the time required to reinstall and
reconfigure everything to its present state. I think this situation is
common for many individuals.
But more specifically to the issue of disk partitioning, there are three other
points here:
Firstly, the physically last partition that the installer automatically creates
is mounted on /home, and this is usually the one that the average home user
is going to want to expand in to free space that follows.
Secondly, having an enormous /home or any other standard system partition is
not necessarily desirable for various reasons, some of which have already been
pointed out. So adding a separate partition using the free space and mounting
it on, say, /storage, might also be a better idea.
Thirdly, unless you are on a machine with hardware constraints, (laptop, SBC,
etc), why would you want a boot/system disk larger than 512 Gb?
Keeping the boot/system disk small and installing a separate drive for bulk
data storage has the benefits of easy replacement in case of failure, (you're
only replacing and re-installing the system drive, not all of your data), and
it's also convenient if you want to swap out the drive for another to install
a new release, but keep the option of returning to the old system without the
hassle of restoring.
I acknowledge that it is always preferable to use separate storage units
for the operating system and for data. The issue with fsck(8) that Åke
Nordin mentioned at the outset is a compelling reason to adopt this
practice. However, it is not always feasible due to hardware limitations
or budget constraints. My specific situation is a combination of both. I
have a home server with a 4TB hard disk, and given its usage, I should
have only enlarged the partitions mounted on /var and /home. Instead, I
expanded all the partitions proportionally.
I hope to never have to use the fsck(8) utility. For this reason, and to
prevent data corruption, I use an Uninterruptible Power Supply (UPS) and
run the stable branch. I think a UPS is essential for any desktop or
server running OpenBSD.
