Alpine Linux: Minimal Install, Minimal Attack Surface

· Installing Alpine Linux as a lightweight, secure server OS.

#linux

1. Alpine Linux

Alpine Linux is a very small distribution, built on musl libc and busybox. It uses ash as the default shell, OpenRC as the init system, and apk as the package manager. According to their website, an Alpine container "requires no more than 8 megabytes (MB) and a minimal installation to disk requires around 130 MB of storage." An actual bare metal machine is recommended to have 100 MB of RAM and 0-700 MB of storage space.

Historically, I've used Ubuntu's minimal installation image as my server OS for the last five years. Ubuntu worked well and helped as my original server contained an nVidia GPU (graphics processing unit) and no onboard graphics, so quite a few distributions won't boot or install without a lot of tinkering.

Alpine has given me a huge increase in performance across my Docker apps and Nginx websites. CPU (central processing unit) load for the new server I'm using to test Alpine hovers around 0-5% on average with an Intel(R) Core(TM) i3-6100 CPU @ 3.70 Gigahertz (GHz).

The only services I haven't moved over to Alpine are Plex Media Server and Syncthing, which may increase CPU load quite a bit depending on how many streams are running.

1.1. Installation

In terms of installation, Alpine has an incredibly useful wiki that will guide a user throughout the installation and post-installation processes, as well as various other articles and guides.

To install Alpine, find an appropriate image to download and flash it to a USB (Universal Serial Bus) using software such as Rufus or Etcher. I opted to use the Standard image for my x8664 architecture.

Once the USB is ready, plug it into the machine and reboot. Note that you may have to use a key such as Esc or F1-12 to access the boot menu. The Alpine Linux terminal will load quickly and for a login.

To log in to the installation image, use the root account; there is no password. Once logged-in, execute the setup command:

setup-alpine

The setup script will ask a series of questions to configure the system. Be sure to answer carefully or else you may have to re-configure the system after boot.

  • Keyboard Layout: Local keyboard language and usage mode, e.g., us and variant of us-nodeadkeys.
  • Hostname: The name for the computer.
  • Network: For example, automatic internet protocol (IP) address discovery with the Dynamic Host Configuration Protocol (DHCP) protocol.
  • DNS (Domain Name System): DNS servers to query. For privacy reasons, it is NOT recommended to route every local request to servers like Google's 8.8.8.8.
  • *Timezone
  • Proxy: Proxy server to use for accessing the web. Use none for direct connections to the internet.
  • Mirror: From where to download packages. Choose the organization you trust giving your usage patterns to.
  • SSH (Secure Shell Protocol): Remote access server. Openssh is part of the default install image. Use none to disable remote login, e.g. on laptops.
  • NTP (Network Time Protocol): Client used for keeping the system clock in sync with a time-server. Package chrony is part of the default install image.
  • Disk Mode: Select between diskless (none), data or sys, as described above.

Once the setup script is finished, be sure to reboot the machine and remove the USB device.

reboot

1.2. Post-Installation

There are many things you can do once your Alpine Linux system is up and running, and it largely depends on what you'll use the machine for. I'm going to walk through my personal post-installation setup for my web server.

  1. Upgrade the System

    First, login as root in order to update and upgrade the system:

    apk -U upgrade

  2. Adding a User

    I needed to add a user so that I don't need to log in as root. Note that if you're used to using the sudo command, you will now need to use the doas command on Alpine Linux.

    apk add doas
adduser <username>
adduser <username> wheel

    You can now log out and log back in using the newly-created user:

    exit

  3. Enable Community Packages

    In order to install more common packages that aren't found in the main repository, you will need to enable the community repository:

    doas nano /etc/apk/repositories

    Uncomment the community line for whichever version of Alpine you're running:

    /media/usb/apks
http://dl-cdn.alpinelinux.org/alpine/v3.16/main
http://dl-cdn.alpinelinux.org/alpine/v3.16/community
#http://dl-cdn.alpinelinux.org/alpine/edge/main
#http://dl-cdn.alpinelinux.org/alpine/edge/community
#http://dl-cdn.alpinelinux.org/alpine/edge/testing

  4. Install Required Packages

    Now that the community packages are available, you can install any packages you need. In my case, I installed the web server packages I need for my services:

    doas apk add nano nginx docker docker-compose ufw

  5. SSH

    If you didn't install OpenSSH as part of the installation, you can do so now:

    doas apk add openssh

    Next, either create a new key or copy your SSH key to the server from your current machines:

    # Create a new key
ssh-keygen

    If you need to copy an existing SSH key from a current machine:

    # Copy key from existing machines
ssh-copy-id <username>@<ip_address>

  6. Firewall

    Lastly, I installed ufw above as my firewall. To set up, default to deny incoming and allow outgoing connections. Then selectively allow other ports or apps as needed.

    doas ufw default deny incoming
doas ufw default allow outgoing
doas ufw allow SSH
doas ufw allow "WWW Full"
doas ufw allow 9418 # Git server port

  7. Change Hostname

    If you don't like the hostname set during installation, you just need to edit two files. First, edit the simple hostname file:

    doas nano /etc/hostname
    <hostname>

    Next, edit the hosts file:

    doas nano /etc/hosts
    127.0.0.1   <hostname>.local <hostname> localhost.local localhost
::1         <hostname> <hostname>.local

2. Nginx Web Server

To set up my web server, I simply created the www user and created the necessary files.

doas adduser -D -g 'www' www
mkdir /www
doas mkdir /www
doas chown -R www:www /var/lib/nginx/
doas chown -R www:www /www

If you're running a simple webroot, you can alter the main nginx.conf file. Otherwise, you can drop configuration files in the following directory. You don't need to enable or symlink the configuration file like you do in other systems.

doas nano /etc/nginx/http.d/example_website.conf

Once the configuration is set and pointed at the /www directory to serve files, enable the Nginx service:

# Note that 'default' must be included or Nginx will not start on boot
doas rc-update add nginx default

3. Docker Containers

Docker works exactly the same as other systems. Either execute a docker run command or create a docker-compose.yml file and do docker-compose up -d.

4. Git Server

I went in-depth on how to self-host a git server in another post: Self-Hosting Guide: Git & cGit.

However, there are a few differences with Alpine. First note that in order to change the git user's shell, you must do a few things a little different:

doas apk add libuser
doas touch /etc/login.defs
doas mkdir /etc/default
doas touch /etc/default/useradd
doas lchsh git

5. Thoughts on Alpine

So far, I love Alpine Linux. I have no complaints about anything at this point, but I'm not completely finished with the migration yet. Once I'm able to upgrade my hardware to a rack-mounted server, I will migrate Plex and Syncthing over to Alpine as well - possibly putting Plex into a container or VM.

The performance is stellar, the apk package manager is seamless, and system administration tasks are effortless. My only regret is that I didn't install Alpine sooner.

Reply via email →
...or, comment on this post on Bubbles!