DNSMasq, the Pint-Sized Super Dæmon!
I've always been a fan of putting aftermarket firmware on consumer-grade routers. Whether it's DD-WRT, Tomato, OpenWRT or whatever your favorite flavor of "better than stock" firmware might be, it just makes economic sense. Unfortunately, my routing needs have surpassed my trusty Linksys router. Although I could certainly buy a several-hundred-dollar, business-class router, I really don't like spending money like that. Thankfully, I found an incredible little router (the EdgeRouter Lite) that can route a million packets per second and has three gigabit Ethernet ports. So far, it's an incredible router, but that's all it does—route. Which brings me to the point of this article.
I've always used the DHCP and DNS server built in to DD-WRT to serve my network. I like having those two services tied to the router, because if every other server on my network fails, I still can get on-line. I figure the next best thing is to have a Raspberry Pi dedicated to those services. Because all my RPi devices currently are attached to televisions around the house (running XBMC), I decided to enlist the Cubox computer I reviewed in November 2013 (Figure 1). It's been sitting on my shelf collecting dust, and I'd rather have it do something useful.
Figure 1. The Cubox is more powerful than a Raspberry Pi, but even an RPi is more power than DNSMasq requires!
Although the Cubox certainly is powerful enough to run BIND and the ISC DHCP server, that's really overkill for my network. Plus, BIND really annoys me with its serial-number incrementation and such whenever an update is made. It wasn't until I started to research alternate DNS servers that I realized just how powerful DNSMasq can be. Plus, the way it works is simplicity at its finest. First, let's look at its features:
-
Extremely small memory and CPU footprint: I knew this was the case, because it's the program that runs on Linux-based consumer routers where memory and CPU are at a premium.
-
DNS server: DNSMasq approaches DNS in a different way from the traditional BIND dæmon. It doesn't offer the complexity of domain transfers, master/slave relationships and so on. It does offer extremely simple and highly configurable options that are, in my opinion, far more useful in a small- to medium-size network. It even does reverse DNS (PTR records) automatically! (More on those details later.)
-
DHCP server: where the DNS portion of DNSMasq lacks in certain advanced features, the DHCP services offered actually are extremely robust. Most routers running firmware like DD-WRT don't offer a Web interface to the advanced features DNSMasq provides, but it rivals and even surpasses some of the standalone DHCP servers.
-
TFTP server: working in perfect tandem with the advanced features of DHCP, DNSMasq even offers a built-in TFTP server for things like booting thin clients or sending configuration files.
-
A single configuration file: it's possible to use multiple configuration files, and I even recommend it for clarity's sake. In the end, however, DNSMasq requires you to edit only a single configuration file to manage all of its powerful services. That configuration file also is very well commented, which makes using it much nicer.
Installation
DNSMasq has been around for a very long time. Installing it on any Linux operating system should be as simple as searching for it in your distribution's package management system. On Debian-based systems that would mean something like:
sudo apt-get install dnsmasq
Or, on a Red Hat/CentOS system:
yum install dnsmasq (as root)
The configuration file (there's just one!) is usually stored at /etc/dnsmasq.conf, and like I mentioned earlier, it is very well commented. Figuring out even the most advanced features is usually as easy as reading the configuration file and un-commenting those directives you want to enable. There are even examples for those directives that require you to enter information specific to your environment.
After the dnsmasq package is installed, it most likely will get started automatically. From that point on, any time you make changes to the configuration (or make changes to the /etc/hosts file), you'll need to restart the service or send an HUP signal to the dæmon. I recommend using the init script to do that:
sudo service dnsmasq restart
But, if your system doesn't have the init service set up for dnsmasq, you can issue an HUP signal by typing something like this:
sudo kill -HUP $(pidof dnsmasq)
This will find the PID (process ID) and send the signal to reload its configuration files. Either way should work, but the init script will give you more feedback if there are errors.
First Up: DNS
Of all the features DNSMasq offers, I find its DNS services to be the most useful and awesome. You get the full functionality of your upstream DNS server (usually provided by your ISP), while seamlessly integrating DNS records for you own network. To accomplish that "split DNS"-type setup with BIND, you need to create a fake DNS master file, and even then you run into problems if you are missing a DNS name in your local master file, because BIND won't query another server by default for records it thinks it's in charge of serving. DNSMasq, on the other hand, follows a very simple procedure when it receives a request. Figure 2 shows that process.
Figure 2. DNSMasq makes DNS queries simple, flexible and highly configurable.
For my purposes, this means I can put a single entry into my server's /etc/hosts file for something like "server.brainofshawn.com", and DNSMasq will return the IP address in the /etc/hosts file. If a host queries DNSMask for an entry not in the server's /etc/hosts file, www.brainofshawn.com for instance, it will query the upstream DNS server and return the live IP for my Web host. DNSMasq makes a split DNS scenario extremely easy to maintain, and because it uses the server's /etc/hosts file, it's simple to modify entries.
Shawn Powers is a Linux Journal Associate Editor. You might find him on IRC, Twitter, or training IT pros at CBT Nuggets.
Trending Topics
| Ansible: the Automation Framework That Thinks Like a Sysadmin | Jan 05, 2018 |
| Now What? | Jan 03, 2018 |
| Happy New Year- Welcome to Linux Journal 2.0! | Jan 01, 2018 |
| Linux Journal Ceases Publication | Dec 01, 2017 |
| So Long, and Thanks for All the Bash | Dec 01, 2017 |
| Banana Backups | Nov 21, 2017 |
- Ansible: the Automation Framework That Thinks Like a Sysadmin
- Happy New Year- Welcome to Linux Journal 2.0!
- Understanding Firewalld in Multi-Zone Configurations
- Simple Server Hardening
- The Weather Outside Is Frightful (Or Is It?)
- Linux Journal February 2017
- Now What?
- Buddy Platform Limited's Parse on Buddy Service
- Non-Linux FOSS: Control Web-Based Music!
- From vs. to + for Microsoft and Linux
Geek Guides
Pick up any e-commerce web or mobile app today, and you’ll be holding a mashup of interconnected applications and services from a variety of different providers. For instance, when you connect to Amazon’s e-commerce app, cookies, tags and pixels that are monitored by solutions like Exact Target, BazaarVoice, Bing, Shopzilla, Liveramp and Google Tag Manager track every action you take. You’re presented with special offers and coupons based on your viewing and buying patterns. If you find something you want for your birthday, a third party manages your wish list, which you can share through multiple social- media outlets or email to a friend. When you select something to buy, you find yourself presented with similar items as kind suggestions. And when you finally check out, you’re offered the ability to pay with promo codes, gifts cards, PayPal or a variety of credit cards.
Get the Guide
