Introduced in 0.5.0, Enhanced in 0.6.0
proxy:ports <app> # List proxy port mappings for app proxy:ports-set <app> <scheme>:<host-port>:<container-port> [<scheme>:<host-port>:<container-port>...] # Set proxy port mappings for app proxy:ports-add <app> <scheme>:<host-port>:<container-port> [<scheme>:<host-port>:<container-port>...] # Add proxy port mappings for app proxy:ports-clear <app> # Clear all proxy port mappings for app proxy:ports-remove <app> <host-port> [<host-port>|<scheme>:<host-port>:<container-port>...] # Unset proxy port mappings for app
In Dokku 0.5.0, port proxying was decoupled from the
nginx-vhosts plugin into the proxy plugin. Dokku 0.6.0 introduced the ability to map host ports to specific container ports. In the future this will allow other proxy software - such as HAProxy or Caddy - to be used in place of nginx.
Proxy port mapping
Introduced in 0.6.0
You can now configure
host -> container port mappings with the
proxy:ports-* commands. This mapping is currently supported by the built-in nginx-vhosts plugin.
To inspect the port mapping for a given application, use the
The above application is listening on the host's port
80, which we can test via curl:
There are cases where we may wish for the service to be listening on more than one port, such as port 8080. Normally, this would not be possible:
However, we can use the
proxy:ports-add command to add a second external port mapping -
8080 - to our application's port
We can now test that port 80 still responds properly:
And our new listening port of
8080 also works:
Port mappings can also be force set using the
A port mapping can be removed using the
proxy:ports-remove command if it no longer necessary:
By default, buildpack apps and dockerfile apps without explicitly exposed ports (i.e. using the
EXPOSE directive) will be configured with a listener on port
80 (and additionally a listener on 443 if ssl is enabled) that will proxy to the application container on port
5000. Dockerfile apps with explicitly exposed ports will be configured with a listener on each exposed port and will proxy to that same port of the deployed application container.
This default behavior will not be automatically changed on subsequent pushes and must be manipulated with the
proxy:ports-* syntax detailed above.
Port management by Deployment Method
If you set a proxy port map but do not have a global domain set, Dokku will reset that map upon first deployment.
For buildpack deployments, your application must respect the
PORT environment variable. We will typically set this to port
5000, but this is not guaranteed. If you do not respect the
PORT environment variable, your containers may start but your services will not be accessible outside of that container.
Changed as of 0.5.0
Dokku's default proxy implementation - nginx - only supports HTTP request proxying. At this time, we do not support proxying plain TCP or UDP ports. UDP ports can be exposed by disabling the nginx proxy with
dokku proxy:disable myapp. If you would like to investigate alternative proxy methods, please refer to our proxy management documentation.
Applications using EXPOSE
Dokku will extract all tcp ports exposed using the
EXPOSE directive (one port per line) and setup nginx to proxy the same port numbers to listen publicly. If you would like to change the exposed port, you should do so within your
For example, if the Dokku installation is configured with the domain
dokku.me and an application named
node-js-app is deployed with following Dockerfile:
The application would be exposed to the user at
node-js-app.dokku.me:1234. If this is not desired, the following application configuration may be applied:
Applications not using EXPOSE
Any application that does not use an
EXPOSE directive will result in Dokku defaulting to port
5000. This behavior mimics the behavior of a Buildpack deploy. If your application does not support the
PORT environment variable, then you will either need to:
- modify your application to support the
- switch to using an
EXPOSEdirective in your Dockerfile.
EXPOSE usage modes
When switching between
EXPOSE usage modes, it is important to reset your port management. The following two commands can be used to reset your state and redeploy your application.
When deploying an image, we will use
docker inspect to extract the
ExposedPorts configuration and if defined, use that to populate port mapping. If this behavior is not desired, you can override that configuration variable with the following commands.
All other port-related behavior is the same as when deploying via Dockerfile.