1.17.0
,latest
(1.17.0/Dockerfile)1.16.3
, (1.16.3/Dockerfile)1.16.2
, (1.16.2/Dockerfile)1.16.1
, (1.16.1/Dockerfile)1.16.0
, (1.16.0/Dockerfile)1.15.0
, (1.15.0/Dockerfile)1.14.0
, (1.14.0/Dockerfile)
Unbound is a validating, recursive, and caching DNS resolver.
Run this container with the following command:
docker run \
--name=my-unbound \
--detach=true \
--publish=53:53/tcp \
--publish=53:53/udp \
--restart=unless-stopped \
mvance/unbound:latest
By default, this image forwards queries Cloudflare DNS server over TLS. In other words, it does not act as a recursive server. The unbound.sh file provides the configuration unless it is overriden as described below.
Note: The example unbound.conf file is different from the one set by unbound.sh file. The example is provided to help you re-configure this as a recursive server.
By default, forwarders are configured to use Cloudflare DNS. You can retrieve the configuration in the forward-records.conf file.
You can create your own configuration file and override the one placed in /opt/unbound/etc/unbound/forward-records.conf
in the container. This is useful if you prefer to use something other than Cloudflare DNS but do not want to provide a custom unbound.conf file.
Example forward-records.conf
:
forward-zone:
# Forward all queries (except those in cache and local zone) to
# upstream recursive servers
name: "."
# my DNS
forward-addr: 192.168.0.1@53#home.local
Another example forward-records.conf
:
forward-zone:
# Forward all queries (except those in cache and local zone) to
# upstream recursive servers
name: "."
# Queries to this forward zone use TLS
forward-tls-upstream: yes
## CleanBrowsing Family Filter
forward-addr: 185.228.168.168@853#family-filter-dns.cleanbrowsing.org
forward-addr: 185.228.169.168@853#family-filter-dns.cleanbrowsing.org
Once the file has your entries in it, mount your version of the file as a volume when starting the container:
docker run \
--name my-unbound \
--detach=true \
--publish=53:53/tcp \
--publish=53:53/udp \
--restart=unless-stopped \
---volume $(pwd)/forward-records.conf:/opt/unbound/etc/unbound/forward-records.conf:ro \
mvance/unbound:latest
Instead of using this image's default configuration for Unbound, you may supply your own configuration. If your customized configuration is located at /my-directory/unbound/unbound.conf
, pass /my-directory/unbound
as a volume when creating your container:
docker run --name=my-unbound \
--detach=true \
--publish=53:53/tcp \
--publish=53:53/udp \
--restart=unless-stopped \
--volume=/my-directory/unbound:/opt/unbound/etc/unbound/ \
mvance/unbound:latest
This will expose all files in /my-directory/unbound/
to the container. As an alternate way to serve custom DNS records for any local zones, either place them directly in your unbound.conf
, or place the local zones in a separate file and use Unbound's include directive within your unbound.conf
:
include: /opt/unbound/etc/unbound/local-zone-unbound.conf
Your volume's contents might eventually look something like this:
/my-directory/unbound/
-- unbound.conf
-- local-zone-unbound.conf
-- secret-zone.conf
-- some-other.conf
Overall, this approach is very similar to the a-records.conf
approach described below. However, by passing your unbound directory rather than a single file, you have more options for customizing and segmenting your Unbound configuration.
Note: Care has been taken in the image's default configuration to enable security options so it is recommended to use it as a guide.
If you want to run Unbound on a different port such as 5353, modify the publish flags:
sudo docker run \
--name=my-unbound \
--publish=5353:53/tcp \
--publish=5353:53/udp \
--detach=true \
--restart=unless-stopped \
--volume=$(pwd)/my-directory/forward-records.conf:/opt/unbound/etc/unbound/forward-records.conf:ro \
--volume=$(pwd)/my-directory/a-records.conf:/opt/unbound/etc/unbound/a-records.conf:ro \
mvance/unbound:latest
While Unbound is not a full authoritative name server, it supports resolving custom entries on a small, private LAN. In other words, you can use Unbound to resolve fake names such as your-computer.local within your LAN.
To support such custom entries using this image, you need to provide an
a-records.conf
or srv-records.conf
file.
This configuration file is where you will define your custom
entries for forward and reverse resolution.
The a-records.conf
file should use the following format:
# A Record
#local-data: "somecomputer.local. A 192.168.1.1"
local-data: "laptop.local. A 192.168.1.2"
# PTR Record
#local-data-ptr: "192.168.1.1 somecomputer.local."
local-data-ptr: "192.168.1.2 laptop.local."
Once the file has your entries in it, mount your version of the file as a volume when starting the container:
docker run \
--name my-unbound \
--detach=true \
--publish=53:53/tcp \
--publish=53:53/udp \
--restart=unless-stopped \
--volume $(pwd)/a-records.conf:/opt/unbound/etc/unbound/a-records.conf:ro \
mvance/unbound:latest
The srv-records.conf
file should use the following format:
# SRV records
# _service._proto.name. | TTL | class | SRV | priority | weight | port | target.
_etcd-server-ssl._tcp.domain.local. 86400 IN SRV 0 10 2380 etcd-0.domain.local.
_etcd-server-ssl._tcp.domain.local. 86400 IN SRV 0 10 2380 etcd-1.domain.local.
_etcd-server-ssl._tcp.domain.local. 86400 IN SRV 0 10 2380 etcd-2.domain.local.
Run a container that use this SRV config file:
docker run \
--name my-unbound \
--detach=true \
--publish=53:53/tcp \
--publish=53:53/udp \
--restart=unless-stopped \
--volume $(pwd)/srv-records.conf:/opt/unbound/etc/unbound/srv-records.conf:ro \
mvance/unbound:latest
The following docker-compose.yml
file is a starting point. The provided example shows how to override default forward and serve custom DNS records for your LAN. It requires forward-records.conf
and a-records.conf
files be provided at the ./my_conf/
.
version: '3'
services:
unbound:
container_name: unbound
image: "mvance/unbound:latest"
expose:
- "53"
networks:
- dns
network_mode: bridge
ports:
- target: 53
published: 53
protocol: tcp
mode: host
- target: 53
published: 53
protocol: udp
mode: host
volumes:
- type: bind
read_only: true
source: ./my_conf/forward-records.conf
target: /opt/unbound/etc/unbound/forward-records.conf
- type: bind
read_only: true
source: ./my_conf/a-records.conf
target: /opt/unbound/etc/unbound/a-records.conf
restart: unless-stopped
networks:
dns:
volumes:
mydata:
If you would rather provide a fully custom unbound.conf
file, you will need to provide an unbound.conf
file and mount it as a volume:
volumes:
- type: bind
read_only: true
source: ./my_conf/unbound.conf
target: /opt/unbound/etc/unbound/unbound.conf
The method described here is basic and not recommended for larger environments. While this example is provided, support for Kubernetes related issues is outside the scope of this project.
To spin the deployment up use:
kubectl apply -f unbound-main-conf.yml -f other-files.yml ...
When taking it down, remember to use the reverse order in which you spun the deployment up.
Restarting:
kubectl rollout restart deployment dns
An example deployment can be viewed here. It is not ready since you need to fill it with your records and the main unbound configuration file.
A fair warning: The example is not using a Service but a hostPort, thus this is only a mock-up. One should not use hostPort in a production cluster.
Additional warning: As per this document the default secrets configuration is unencrypted per default. You are responsible to harden this yourself and should do so!
The default config forwards forwards DNS queries to another DNS server over TLS. If you would rather this work as a recursive DNS server, you must use a customized Unbound configuration. An example unbound.conf file to configure unbound as a recursive server is available as a guide.
For a DNS server with lots of short-lived connections, you may wish to consider
adding --net=host
to the run command for performance reasons. However, it is
not required and some shared container hosting services may not allow it. You
should also be aware that using --net=host
can be a security risk in some situations. The
Center for Internet Security Docker 1.6
Benchmark
recommends against this mode since it essentially tells Docker to not
containerize the container's networking, thereby giving it full access to the
host machine's network interfaces. It also mentions this option could cause the
container to do unexpected things such as shutting down the Docker host as
referenced in Docker Issue #6401
. For the most secure deployment, unrelated services with confidential data
should not be run on the same host or VPS. In such cases, using --net=host
should have limited impact on security.
Logging is very limited in the default config created by unbound.sh. If using the default config as an example starting point, a placeholder for a logfile (unbound.log
) has been provided with the correct file ownership at the path /opt/unbound/etc/unbound/
in case you want to increase logging and send to a file.
By default, this image includes a healthcheck that performs a query for cloudflare.com on localhost at a regular interval.
To disable the healthcheck, add the --no-healthcheck
flag to your Dockerfile. If using docker-compose, you can configure the healthcheck differently as explained in the Docker docs.
The following message may appears in the logs about IPv6 Address Assignment:
[1644625926] libunbound[24:0] error: udp connect failed: Cannot assign requested address for 2001:xxx:xx::x port 53
While annoying, the container works despite the error. Search this issues in this repo for "udp connect" to see more discussion.
Documentation for this image is stored right here in the README.md
.
Documentation for Unbound is available on the project's website.
If you have any problems with or questions about this image, please contact me through a GitHub issue.
You are invited to contribute new features, fixes, or updates, large or small. I imagine the upstream projects would be equally pleased to receive your contributions.
Please familiarize yourself with the repository's README.md
file
before attempting a pull request.
Before you start to code, I recommend discussing your plans through a GitHub issue, especially for more ambitious contributions. This gives other contributors a chance to point you in the right direction, give you feedback on your design, and help you find out if someone else is working on the same thing.
The code in this image is heavily influenced by DNSCrypt server Docker image, though the upstream projects most certainly also deserve credit for making this all possible.
Unless otherwise specified, all code is released under the MIT License (MIT).
See the repository's LICENSE
file for
details.
- Docker: Apache 2.0
- DNSCrypt server Docker image: ISC License
- LibreSSL: Various
- OpenSSL: Apache-style license
- Unbound: BSD License