Once you have containers running the next step is to allow them to be accessed from the Internet. Convox automatically helps you set up and configure load balancers appropriately to route traffic to your containers.
Load balancers will be automatically created for any ports listed in your
web: build: . command: bin/web ports: - 80:5000 worker: build: . command: bin/worker
In this example, Convox will create a load balancer in front of the
web process. This load balancer will accept traffic from the internet on port 80 and forward it to the
web containers on port
You can find the load balancer hostname(s) for your application using
convox apps info:
$ convox apps info Name docs Status running Release RHUFNNNVEAP Processes web Endpoints docs-web-R72RMTP-326048479.us-east-1.elb.amazonaws.com:80 (web)
Internal Load Balancers
You can create a load balancer that is only accessible inside your Rack by specifying a single port:
web: ports: - 5000
Because Convox creates only one load balancer per service if you specify both internal and external ports an internal load balancer will be created.
You can specify one of four protocol types for a load balancer port in your
web: labels: - convox.port.443.protocol=https ports: - 443:5000
||Unencrypted HTTP (includes common HTTP headers but does not support websockets)|
||Encrypted HTTP (includes common HTTP headers but does not support websockets)|
||Unencrypted TCP (arbitrary TCP including websockets, no HTTP header injection)|
||Encrypted TCP (arbitrary TCP including websockets, no HTTP header injection)|
If no protocol label is specified the default of
tcp will be used.
Health Check Options
By default Convox will set up a
tcp health check to your application. You can customize the health check by adding the following labels to your
web: labels: - convox.health.path=/_health - convox.health.port=5000 - convox.health.timeout=3 ports: - 443:5000
||The endpoint the load balancer will use to determine the application's health.|
||This is the port that your container is set up to listen on, not the load balancer port.|
||The time in seconds after which no response means a failed health check. If the process fails 2 consecutive health checks it will be restarted. By default, the interval between health checks is this value plus 2.|
By default, HTTPS/TLS is terminated at the load balancer, and the resulting data is transmitted unencrypted to your application. This is OK, because traffic between your load balancer and your application happens entirely on your Rack’s internal network. However, for extra security you can encrypt the traffic between your load balancer and application by setting the `convox.port.
web: labels: - convox.port.443.secure=true - convox.port.443.protocol=https ports: - 443:5001
When you use this option you will need to terminate HTTPS or TLS directly inside your application or with a reverse proxy like nginx or haproxy.
When using the
tls protocols, standard proxy HTTP headers like
X-Forwarded-For are not injected. You can get access to information about the remote endpoint using the PROXY protocol. Once you configure your application to accept this extra header you can configure your load balancer to send it in your
web: labels: - convox.port.443.protocol=tls - convox.port.443.proxy=true ports: - 443:5000
Limited Application Access
For security reasons, access to an application might need to be limited. To achieve this, an existing security group can be applied to an application’s load balancer. For example, within said security group, access can be granted only to an office VPN.
This is done via an application parameter with a known security group ID:
convox apps params --app <name> set SecurityGroup=sg-123456
Validate the setting has been applied by running:
convox apps params --app <name>