3.1.11 Creating filesystems
3.1.11.1 Hard disk partition
I prefer to use different partitions for different directory trees to limit
damage upon system crash. E.g.,
/ == (/ + /boot + /bin + /sbin)
== 50MB+
/tmp == 100MB+
/var == 100MB+
/home == 100MB+
/usr == 700MB+ with X
/usr/local == 100MB
The size of the /usr directory is very dependent on X Window
applications and documentation. /usr/ can be 300MB if one runs a
console terminal only, whereas 2GB–3GB is not an unusual size if one has
installed many Gnome applications. When /usr/ grows too big,
moving out /usr/share/ to a different partition is the most
effective cure. With the new large prepackaged Linux 2.4 kernels,
/ may need more than 200MB.
For example, the current status of my Internet gateway machine is as follows
(output of the df -h command):
Filesystem Size Used Avail Use% Mounted on
/dev/hda3 300M 106M 179M 38% /
/dev/hda7 100M 12M 82M 13% /home
/dev/hda8 596M 53M 513M 10% /var
/dev/hda6 100M 834k 94M 1% /var/lib/cvs
/dev/hda9 596M 222M 343M 40% /usr
/dev/hda10 596M 130M 436M 23% /var/cache/apt/archives
/dev/hda11 1.5G 204M 1.2G 14% /var/spool/squid
(The large area reserved for /var/spool/squid/ is for a proxy
cache for package downloading.)
Following is fdisk -l output to provide an idea of partition
structure:
# fdisk -l /dev/hda # comment
/dev/hda1 1 41 309928+ 6 FAT16 # DOS
/dev/hda2 42 84 325080 83 Linux # (not used)
/dev/hda3 * 85 126 317520 83 Linux # Main
/dev/hda4 127 629 3802680 5 Extended
/dev/hda5 127 143 128488+ 82 Linux swap
/dev/hda6 144 157 105808+ 83 Linux
/dev/hda7 158 171 105808+ 83 Linux
/dev/hda8 172 253 619888+ 83 Linux
/dev/hda9 254 335 619888+ 83 Linux
/dev/hda10 336 417 619888+ 83 Linux
/dev/hda11 418 629 1602688+ 83 Linux
A few unused partitions exist. These are for installing a second Linux
distribution or as expansion space for growing directory trees.
3.1.11.2 Mount filesystems
Mounting the above filesystems properly is accomplished with the following
/etc/fstab:
# /etc/fstab: static filesystem information.
#
# filesystem mount point type options dump pass
/dev/hda3 / ext2 defaults,errors=remount-ro 0 1
/dev/hda5 none swap sw 0 0
proc /proc proc defaults 0 0
/dev/fd0 /floppy auto defaults,user,noauto 0 0
/dev/cdrom /cdrom iso9660 defaults,ro,user,noauto 0 0
#
# keep partitions separate
/dev/hda7 /home ext2 defaults 0 2
/dev/hda8 /var ext2 defaults 0 2
/dev/hda6 /var/lib/cvs ext2 defaults 0 2
# noatime will speed up file access for read access
/dev/hda9 /usr ext2 defaults,noatime 0 2
/dev/hda10 /var/cache/apt/archives ext2 defaults 0 2
# very big partition for proxy cache
/dev/hda11 /var/spool/squid ext2 rw 0 2
# backup bootable DOS
/dev/hda1 /mnt/dos vfat rw,noauto 0 0
# backup bootable Linux system (not done)
/dev/hda2 /mnt/linux ext2 rw,noauto 0 0
#
# nfs mounts
mickey:/ /mnt/mickey nfs ro,noauto,intr 0 0
goofy:/ /mnt/goofy nfs ro,noauto,intr 0 0
# minnie:/ /mnt/minnie smbfs ro,soft,intr,credentials={filename} 0 2
For NFS, I use noauto,intr combined with the default
hard option. This way, it is possible to recover from a hung
process due to a dead connection using Ctrl-C.
For a Windows machine connected with Samba (smbfs),
rw,auto,soft,intr may be good idea. See
Samba
configuration, Section 3.5.
For a floppy drive, using noauto,rw,sync,user,exec instead
prevents file corruption after accidental disk eject before unmount, but this
slows the write process.
3.1.11.3 Autofs mount
Key points to auto mount:
-
Load the vfat module to allow /etc/auto.misc to
contain -fstype=auto:
# modprobe vfat # prior to the floppy access attempt
... or to automate this setting,
# echo "vfat" >> /etc/modules
... and reboot the system.
-
Set /etc/auto.misc as follows:
floppy -fstype=auto,sync,nodev,nosuid,gid=100,umask=000 :/dev/fd0
... where gid=100 is "users".
-
Create cdrom and floppy links in
/home/user, that point to
/var/autofs/misc/cdrom and /var/autofs/misc/floppy
respectively.
-
Add user to the "users" group.
3.1.11.4 NFS mount
The external Linux NFS server (goofy) resides behind a firewall (gateway). I
have a very relaxed security policy on my LAN since I am the only user. To
enable NFS access, the NFS server side needs to add /etc/exports
as follows:
# /etc/exports: the access control list for filesystems which may be
# exported to NFS clients. See exports(5).
/ (rw,no_root_squash)
This is needed to activate the NFS server in addition to installing and
activating the NFS server and client packages.
For simplicity, I usually create a single partition of 2GB for an experimental
or secondary lazy Linux install. I optionally share swap and /tmp
partitions for these installs. A multipartition scheme is too involved for
these usages. If only a simple console system is needed, 500MB may be more
than sufficient.
