Put browser-valid TLS termination in front of any Dockerized HTTP service with one command.
docker run --detach \
--name lets-nginx \
--link backend:backend \
--env [email protected] \
--env DOMAIN=mydomain.horse \
--env UPSTREAM=backend:8080 \
--publish 80:80 \
--publish 443:443 \
smashwilson/lets-nginx
Issues certificates from letsencrypt, installs them in nginx, and schedules a cron job to reissue them monthly.
⚡ To run unattended, this container accepts the letsencrypt terms of service on your behalf. Make sure that the subscriber agreement is acceptable to you before using this container. ⚡
Before you begin, you'll need:
- A place to run Docker containers with a public IP.
- A domain name with an A record pointing to your cluster.
Launch your backend container and note its name, then launch smashwilson/lets-nginx
with the following parameters:
--link backend:backend
to link your backend service's container to this one. (This may be unnecessary depending on Docker's networking configuration.)-e EMAIL=
your email address, used to register with letsencrypt.-e DOMAIN=
the domain name.-e UPSTREAM=
the name of your backend container and the port on which the service is listening.-p 80:80
and-p 443:443
so that the letsencrypt client and nginx can bind to those ports on your public interface.-e STAGING=1
uses the Let's Encrypt staging server instead of the production one. I highly recommend using this option to double check your infrastructure before you launch a real service. Let's Encrypt rate-limits the production server to issuing five certificates per domain per seven days, which (as I discovered the hard way) you can quickly exhaust by debugging unrelated problems!
You can distribute traffic to multiple upstream proxy destinations, chosen by the Host header. This is useful if you have more than one container you want to access with https.
To do so, separate multiple corresponding values in the DOMAIN and UPSTREAM variables separated by a ;
:
-e DOMAIN="domain1.com;sub.domain1.com;another.domain.net"
-e UPSTREAM="backend:8080;172.17.0.5:60;container:5000"
Since --link
s don't survive the re-creation of the target container, you'll need to coordinate re-creating
the proxy container. In this case, you can cache the certificates and Diffie-Hellman parameters with the following procedure:
Do this once:
docker volume create --name letsencrypt
docker volume create --name letsencrypt-backups
docker volume create --name dhparam-cache
Then start the container, attaching the volumes you just created:
docker run --detach \
--name lets-nginx \
--link backend:backend \
--env [email protected] \
--env DOMAIN=mydomain.horse \
--env UPSTREAM=backend:8080 \
--publish 80:80 \
--publish 443:443 \
--volume letsencrypt:/etc/letsencrypt \
--volume letsencrypt-backups:/var/lib/letsencrypt \
--volume dhparam-cache:/cache \
smashwilson/lets-nginx
The entry point of this image processes the nginx.conf
file in /templates
and places the result in /etc/nginx/nginx.conf
. Additionally, the file /templates/vhost.sample.conf
will be processed once for each ;
-delimited pair of values in $DOMAIN
and $UPSTREAM
. The result of each will be placed at /etc/nginx/vhosts/${DOMAINVALUE}.conf
.
The following variable substitutions are made while processing all of these files:
${DOMAIN}
${UPSTREAM}
For example, to adjust nginx.conf
, create that file in your new image directory with the baseline content and desired modifications. Within your Dockerfile
ADD this file and it will be used to create the nginx configuration instead.
FROM smashwilson/lets-nginx
ADD nginx.conf /templates/nginx.conf