The SSL certificates are obtained, and renewed from Let's Encrypt automatically.
Docker Hub page: https://hub.docker.com/r/steveltn/https-portal/
Table of Contents
- See It Work
- Quick Start
- Advanced Usage
- How It Works
- About Rate Limits of Let's Encrypt
HTTPS-PORTAL is shipped as a Docker image. To use it, you need a Linux machine (either local or remote host) which:
- Has 80 and 443 port available and exposed.
- Has Docker Engine installed. In addition, Docker Compose is highly recommended, for it makes your life easier. Examples in our documents are mainly in Docker Compose format.
- Has all domains you're going to use in the following examples resolving to it.
Though it is good to have, knowledge about Docker is not required to use HTTPS-PORTAL.
See It Work
docker-compose.yml file with the following content in any directory
of your choice:
https-portal: image: steveltn/https-portal:1 ports: - '80:80' - '443:443' environment: DOMAINS: 'example.com' # STAGE: 'production'
docker-compose up command in the same directory. A moment later you'll
have a welcome page running in
Here is a more real-world example: Create the file
docker-compose.yml in another
https-portal: image: steveltn/https-portal:1 ports: - '80:80' - '443:443' links: - wordpress restart: always environment: DOMAINS: 'wordpress.example.com -> http://wordpress' # STAGE: 'production' # FORCE_RENEW: 'true' wordpress: image: wordpress links: - db:mysql db: image: mariadb environment: MYSQL_ROOT_PASSWORD: '>'
docker-compose up -d command. A moment later you'll get a WordPress
running on https://wordpress.example.com.
In the example above, only the environment variables under the
section are HTTPS-PORTAL specific configurations. This time we added an extra
-d, which will tell Docker Compose to run the apps defined in
docker-compose.yml in the background.
staging by default, which results in a test
(untrusted) certificate from Let's Encrypt.
You can test HTTPS-PORTAL with your application stack locally.
https-portal: # ... environment: STAGE: local DOMAINS: 'example.com'
By doing this, HTTPS-PORTAL will create a self-signed certificate. This certificate is not likely to be trusted by your browser, but you can use it to test your docker-compose file. Make sure it works with your application stack.
Note that HTTPS-PORTAL only listens to
example.com, as you specified in the compose file.
In order to make HTTPS-PORTAL respond to your connection, you need to either:
- modify your
hostsfile to have
example.comresolving to your docker host,
- set up DNSMasq on your computer/router. This method provides more flexibility.
Once you are done testing, you can deploy your application stack to the server.
HTTPS-PORTAL support quick setup for redirections. It is deliberately made to support https targets only because otherwise it'd be against the idea of this project.
https-portal: # ... environment: STAGE: local DOMAINS: 'example.com => target.example.com/foo/bar' # Notice it's "=>" instead of the normal "->"
Automatic Container Discovery
HTTPS-PORTAL is capable of discovering other Docker containers running on the same host, as long as the Docker API socket is accessible within the container.
In order to make it so, launch HTTPS-PORTAL using the following
version: '2' services: https-portal: # ... volumes: - /var/run/docker.sock:/var/run/docker.sock:ro
and launch one or more web applications with:
version: '2' services: a-web-application: # ... ports: - '8080:80' environment: # tell HTTPS-PORTAL to set up "example.com" VIRTUAL_HOST: example.com
Caveat: Your web application must be created in the same network as HTTPS-PORTAL.
Note that here is
no need to link your web service to HTTPS-PORTAL, and you
example.com in environment variable
DOMAINS of HTTP-PORTAL.
This feature allows you to deploy multiple web applications on the same host without restarting HTTPS-PORTAL itself or interrupting any other application while adding/removing web applications.
If your web service has more than one port exposed (mind that ports can be exposed in your web service Dockerfile),
use the environment variable
VIRTUAL_PORT to specify which port accepts HTTP requests:
a-multi-port-web-application: # ... ports: - '8080:80' - '2222:22' environment: VIRTUAL_HOST: example.com VIRTUAL_PORT: '8080'
Of course container discovery works in combination with ENV specified domains:
https-portal: # ... volumes: - /var/run/docker.sock:/var/run/docker.sock:ro environment: DOMAINS: 'example.com -> http://upstream'
Hybrid Setup with Non-Dockerized Apps
Web applications that run directly on the host machine instead of in Docker
containers are available at
For instance, if an application accepts HTTP requests on port 8080 of the host machine, you can start HTTPS-PORTAL by:
https-portal: # ... environment: DOMAINS: 'example.com -> http://dockerhost:8080'
You can specify multiple domains by splitting them with commas:
https-portal: # ... environment: DOMAINS: 'wordpress.example.com -> http://wordpress, gitlab.example.com -> http://gitlab'
You can also specify the stage (
production) for each individual site, note that stages of individual sites overrides the global stage:
DOMAINS: 'wordpress.example.com -> http://wordpress #local, gitlab.example.com #production'
Serving Static Sites
Instead of forwarding requests to web applications, HTTPS-PORTAL can also serve (multiple) static sites directly:
https-portal: # ... environment: DOMAINS: 'hexo.example.com, octopress.example.com' volumes: - /data/https-portal/vhosts:/var/www/vhosts
After HTTPS-PORTAL is started, it will create corresponding sub-directories for
each virtual host in the
/data/https-portal/vhosts directory on the host machine:
/data/https-portal/vhosts ├── hexo.example.com │ └── index.html └── octopress.example.com └── index.html
You can place your own static files in this directory hierarchy, they will not
be overwritten. You need an
index.html to be served as the homepage.
Share Certificates with Other Apps
You can mount an arbitrary host directory to
/var/lib/https-portal as a
https-portal: # ... volumes: - /data/ssl_certs:/var/lib/https-portal
Now your certificates are available in
/data/ssl_certs on your host.
HTTP Basic Auth
You can set up an HTTP Basic Auth easily. It is useful when you put the website online but don't want to open it to public until ready.
In your docker-compose file:
https-portal: # ... environment: DOMAINS: 'username:firstname.lastname@example.org ->
Configure Nginx through Environment Variables
There are several additional environment variables that you can use to config Nginx.
They correspond to the configuration options that you would normally supply in
The following are the config keys with default values:
WORKER_PROCESSES=1 WORKER_CONNECTIONS=1024 KEEPALIVE_TIMEOUT=65 GZIP=on SERVER_TOKENS=off SERVER_NAMES_HASH_MAX_SIZE=512 SERVER_NAMES_HASH_BUCKET_SIZE=32 # defaults to 32 or 64 based on your CPU CLIENT_MAX_BODY_SIZE=1M # 0 disables checking request body size PROXY_BUFFERS="8 4k" # Either 4k or 8k depending on the platform PROXY_BUFFER_SIZE="4k" # Either 4k or 8k depending on the platform RESOLVER="Your custom solver string" PROXY_CONNECT_TIMEOUT=60; PROXY_SEND_TIMEOUT=60; PROXY_READ_TIMEOUT=60; ACCESS_LOG=off;
You can also add
to make HTTPS-PORTAL proxy WEBSOCKET connections.
To avoid nginx DNS caching activate dynamic upstream
RESOLVER="127.0.0.11 ipv6=off valid=30s" DYNAMIC_UPSTREAM=true
Override Nginx Configuration Files
You can override default nginx settings by providing a config segment of
nginx.conf containing a valid
server block. The custom nginx configurations
are ERB templates and will be
rendered before usage.
For instance, to override both HTTPS and HTTP settings for
you can launch HTTPS-PORTAL by:
https-portal: # ... volumes: - /path/to/http_config:/var/lib/nginx-conf/my.example.com.conf.erb:ro - /path/to/https_config:/var/lib/nginx-conf/my.example.com.ssl.conf.erb:ro
Another example can be found here.
If you want to make an Nginx configuration that will be used by all sites, you can overwrite
/var/lib/nginx-conf/default.ssl.conf.erb. These two files will be propagated to each site if the site-specific configuration files are not provided.
Manually Set RSA Private Key Length
By default, HTTPS-PORTAL generate
2048 bits long RSA private key.
However, you can manually set RSA private key length (
openssl genrsa command) through
NUMBITS environment variable.
https-portal: # ... environment: NUMBITS: '4096'
How It Works
- obtains an SSL certificate for each of your subdomains from Let's Encrypt.
- configures Nginx to use HTTPS (and force HTTPS by redirecting HTTP to HTTPS)
- sets up a cron job that checks your certificates every week, and renew them. if they expire in 30 days.
About Rate Limits of Let's Encrypt
Let's Encrypt's service is rate limited to ensure fair usage. Please familiarize yourself with the various rate limits. This documentation page is the authoritative source for the current rate limit values.
For most people the most important rate limits are:
- 5 failed validation attempts per hour
- 50 certificates per registered domain per week
If you want to use HTTPS for multiple sub-domains with a single certificate Let's Encrypt supports putting up to 100 domains in one certificate, however it can require careful planning and is hard to automate. So in HTTPS-PORTAL we only deal with single domain name certificates.
HTTPS-PORTAL stores your certificates in a data volume and will not re-sign
certificates until 30 days before expiration if a valid certificate is found
(you can force renew certificates by using
FORCE_RENEW: 'true' environment
variable). However if you play around with the image a lot, you can hit the
limit. That's why
staging by default, and thus we use the
Let's Encrypt staging server. When you have finished your experiments and feel
everything is good, you can switch to production mode with
Source : https://github.com/SteveLTN/https-portal