Under Linux, all partitions used by the system must be listed in /etc/fstab. This file contains the mount points of those partitions (where they are seen in the file system structure), how they should be mounted and with what special options (automatically or not, whether users can mount them or not, etc.)
Creating the fstab file
The /etc/fstab file uses a table-like syntax. Every line consists of six fields, separated by whitespace (space(s), tabs or a mixture). Each field has its own meaning:
- The first field shows the block special device or remote filesystem to be mounted. Several kinds of device identifiers are available for block special device nodes, including paths to device files, filesystem labels and UUIDs, and partition labels and UUIDs.
- The second field shows the mount point at which the partition should be mounted.
- The third field shows the filesystem used by the partition.
- The fourth field shows the mount options used by mount when it wants to mount the partition. As every filesystem has its own mount options, users are encouraged to read the mount man page (man mount) for a full listing. Multiple mount options are comma-separated.
- The fifth field is used by dump to determine if the partition needs to be dumped or not. This can generally be left as 0 (zero).
- The sixth field is used by fsck to determine the order in which filesystems should be checked if the system wasn't shut down properly. The root filesystem should have 1 while the rest should have 2 (or 0 if a filesystem check isn't necessary).
The default /etc/fstab file provided by Gentoo is not a valid fstab file but instead more of a template.
nano -w /etc/fstab
In the remainder of the text, we use the default /dev/sd* block device files as partition.
Filesystem labels and UUIDs
Both MBR (BIOS) and GPT include support for filesystem labels and filesystem UUIDs. These attributes can be defined in /etc/fstab as alternatives for the mount command to use when attempting to find and mount block devices. Filesystem labels and UUIDs are identified by the LABEL and UUID prefix and can be viewed with the blkid command:
If the filesystem inside a partition is wiped, then the filesystem label and the UUID values will be subsequently altered or removed.
Because of uniqueness, readers that are using an MBR-style partition table are recommended to use UUIDs over labels to define mountable volumes in /etc/fstab.
Partition labels and UUIDs
Users who have gone the GPT route have a couple more 'robust' options available to define partitions in /etc/fstab. Partition labels and partition UUIDs can be used to identify the block device's individual partition(s), regardless of what filesystem has been chosen for the partition itself. Partition labels and UUIDs are identified by the PARTLABEL and PARTUUID prefixes respectively and can be viewed nicely in the terminal by running the blkid command:
While not always true for partition labels, using a UUID to identify a partition in fstab provides a guarantee that the bootloader will not be confused when looking for a certain volume, even if the filesystem would be changed in the future. Using the older default block device files (/dev/sd*N) for defining the partitions in fstab is risky for systems that are restarted often and have SATA block devices added and removed regularly.
The naming for block device files depends on a number of factors, including how and in what order the disks are attached to the system. They also could show up in a different order depending on which of the devices are detected by the kernel first during the early boot process. With this being stated, unless one intends to constantly fiddle with the disk ordering, using default block device files is a simple and straightforward approach.
Let us take a look at how to write down the options for the /boot/ partition. This is just an example, and should be modified according to the partitioning decisions made earlier in the installation. In our mips partitioning example, /boot/ is usually the /dev/sda1 partition, with ext4 as filesystem. It needs to be checked during boot, so we would write down:
/dev/sda1 /boot ext4 defaults 0 2
Some users don't want their /boot/ partition to be mounted automatically to improve their system's security. Those people should substitute defaults with noauto. This does mean that those users will need to manually mount this partition every time they want to use it.
Add the rules that match the previously decided partitioning scheme and append rules for devices such as CD-ROM drive(s), and of course, if other partitions or drives are used, for those too.
Below is a more elaborate example of an /etc/fstab file:
/dev/sda1 /boot ext4 defaults,noatime 0 2 /dev/sda10 none swap sw 0 0 /dev/sda5 / ext4 noatime 0 1 /dev/cdrom /mnt/cdrom auto noauto,user 0 0
auto is used in the third field, it makes the mount command guess what the filesystem would be. This is recommended for removable media as they can be created with one of many filesystems. The
user option in the fourth field makes it possible for non-root users to mount the CD.
To improve performance, most users would want to add the
noatime mount option, which results in a faster system since access times aren't registered (those are not needed generally anyway). This is also recommended for solid state drive (SSD) users, who should also enable the
discard mount option (ext4 and btrfs only for now) which makes the
TRIM command work.
Double-check the /etc/fstab file, save and quit to continue.
Host and domain information
One of the choices the user has to make is name his/her PC. This seems to be quite easy, but lots of users are having difficulties finding the appropriate name for their Linux PC. To speed things up, know that the decision is not final - it can be changed afterwards. In the examples below, the hostname tux is used within the domain homenetwork.
nano -w /etc/conf.d/hostname
# Set the hostname variable to the selected host name hostname="tux"
Second, if a domain name is needed, set it in /etc/conf.d/net. This is only necessary if the ISP or network administrator says so, or if the network has a DNS server but not a DHCP server. Don't worry about DNS or domain names if the system uses DHCP for dynamic IP address allocation and network configuration.
The /etc/conf.d/net file does not exist by default, so needs to be created.
nano -w /etc/conf.d/net
# Set the dns_domain_lo variable to the selected domain name dns_domain_lo="homenetwork"
If no domain name is configured, then users will notice they get "This is hostname.(none)" messages at their login screen. This should then be fixed by editing /etc/issue and deleting the string
.\Ofrom that file.
If a NIS domain is needed (users that do not know this will not need one), define that one too:
nano -w /etc/conf.d/net
# Set the nis_domain_lo variable to the selected NIS domain name nis_domain_lo="my-nisdomain"
For more information on configuring DNS and NIS, please read the examples provided in /usr/share/doc/netifrc-*/net.example.bz2 which can be read using bzless. Also, it might be interesting to install net-dns/openresolv to help manage the DNS/NIS setup.
Configuring the network
During the Gentoo Linux installation, networking was already configured. However, that was for the installation CD itself and not for the installed environment. Right now, the network configuration is made for the installed Gentoo Linux system.
More detailed information about networking, including advanced topics like bonding, bridging, 802.1Q VLANs or wireless networking is covered in the Gentoo Network Configuration section.
All networking information is gathered in /etc/conf.d/net. It uses a straightforward yet perhaps not intuitive syntax. But don't fear, everything is explained below. A fully commented example that covers many different configurations is available in /usr/share/doc/netifrc-*/net.example.bz2.
First install net-misc/netifrc:
emerge --ask --noreplace net-misc/netifrc
DHCP is used by default. For DHCP to work, a DHCP client needs to be installed. This is described later in Installing Necessary System Tools.
If the network connection needs to be configured because of specific DHCP options or because DHCP is not used at all, then open /etc/conf.d/net:
nano -w /etc/conf.d/net
Set both config_eth0 and routes_eth0 to enter IP address information and routing information:
This assumes that the network interface will be called eth0. This is, however, very system dependent. It is recommended to assume that the interface is named the same as the interface name when booted from the installation media if the installation media is sufficiently recent. More information can be found in Network Interface Naming.
config_eth0="192.168.0.2 netmask 255.255.255.0 brd 192.168.0.255" routes_eth0="default via 192.168.0.1"
To use DHCP, define config_eth0:
Please read /usr/share/doc/netifrc-*/net.example.bz2 for a list of all available options. Be sure to also read up on the DHCP client man page if specific DHCP options need to be set.
If the system has several network interfaces, then repeat the above steps for config_eth1, config_eth2, etc.
Now save the configuration and exit to continue.
Automatically start networking at boot
To have the network interfaces activated at boot, they need to be added to the default runlevel.
ln -s net.lo net.eth0
rc-update add net.eth0 default
If the system has several network interfaces, then the appropriate net.* files need to be created just like we did with net.eth0.
If after booting the system we find out that the assumption about the network interface name (which is currently documented as
eth0) was wrong, then execute the following steps to rectify this:
- Update the /etc/conf.d/net file with the correct interface name (like
- Create new symbolic link (like /etc/init.d/net.enp3s0).
- Remove the old symbolic link (rm /etc/init.d/net.eth0).
- Add the new one to the default runlevel.
- Remove the old one using rc-update del net.eth0 default.
The hosts file
Next inform Linux about the network environment. This is defined in /etc/hosts and helps in resolving host names to IP addresses for hosts that aren't resolved by the nameserver.
nano -w /etc/hosts
# This defines the current system and must be set 127.0.0.1 tux.homenetwork tux localhost # Optional definition of extra systems on the network 192.168.0.5 jenny.homenetwork jenny 192.168.0.6 benny.homenetwork benny
Save and exit the editor to continue.
Optional: Get PCMCIA working
PCMCIA users should now install the sys-apps/pcmciautils package.
emerge --ask sys-apps/pcmciautils
Set the root password using the passwd command.
The root Linux account is an all-powerful account, so pick a strong password. Later an additional regular user account will be created for daily operations.
Init and boot configuration
Gentoo (at least when using OpenRC) uses /etc/rc.conf to configure the services, startup, and shutdown of a system. Open up /etc/rc.conf and enjoy all the comments in the file. Review the settings and change where needed.
nano -w /etc/rc.conf
Next, open /etc/conf.d/keymaps to handle keyboard configuration. Edit it to configure and select the right keyboard.
nano -w /etc/conf.d/keymaps
Take special care with the keymap variable. If the wrong keymap is selected, then weird results will come up when typing on the keyboard.
Finally, edit /etc/conf.d/hwclock to set the clock options. Edit it according to personal preference.
nano -w /etc/conf.d/hwclock
If the hardware clock is not using UTC, then it is necessary to set
clock="local" in the file. Otherwise the system might show clock skew behavior.