NoOne wrote:
>
> What's the sence to set up different mount points for / , /usr , /home ,
> /anything_else if all of them are located on a single harddrive. I can
> understand this steps for /boot 'cause it must be located in first 1023
> cyls, but what about / , /usr do you really need the separate diskspaces?
>
> It looks like disk partitions under DOS/Windows, but the ideo of splitting
> disks there is separating and more comfortable storage of information. At
> the same time all the mount points you make are connected do the same
> directory tree... I do not understand the sence of it...
>
/ (root) should be it's own partition so that you can keep critical
system libraries and binaries in their own little spot. This speeds up
fsck time if you system crashes and also makes it likely that you'll be
able to do more recovery if other partitions are badly damaged.
/var should really be it's own partition so that system logs don't fill
your system and make system recovery difficult. e2fs guarantees 5% of
the drive will be available for root's use, but if your logs are owned
by root, that's of no help.
/usr as it's own partition allows you to mount that partition
read-only. Not only does that mean that system crashes won't do any
damage if they occur, it also means that cracking the system becomes a
_bit_ more difficult for a would-be attacker. Going further with that
idea, using chattr to set all of those files as immutable means that
they're going to have to work even harder to screw up the system.
/home as a separate partition means you get to keep that partition when
you upgrade or reinstall your operating system. The same should
probably be done for /usr/local, since those are locally installed
packages that are not under the control of the package manager.
Separate partitions mean that you get to make extremely intelligent
choices about the amount of work you'll have to do to upgrade or protect
your system.
--
Steve Philp
Network Administrator
Advance Packaging Corp.
[EMAIL PROTECTED]
