HAProxy docker image

HAProxy docker image with Lets Encrypt SSL auto renewal using certbot with built in support for wildcard certificates using AWS Route53.

Paths

• /deployment/letsencrypt - Certbot config directory where generated certificates are stored
• /etc/haproxy/haproxy.cfg - Default location of haproxy configuration file
• /etc/haproxy/certs - Static (non certbot) certificates includes self-signed and any other static certificates should be volume mapped into this folder
• /var/log/* - Location of log files (all are symlinked to stdout)

Environment variables

• DOMAINNAME - IANA TLD subdomain for which a Lets Encrypt certificate should be requested
• DOMAINNAMES - Comma separated list of IANA TLD subdomain names for which Lets Encrypt certificates should be requested (this is a multi-value alternative to DOMAINNAME)
• HAPROXY_USER_PARAMS - Additional arguments that should be passed to the haproxy process during startup
• HAPROXY_CONFIG - Location of HAProxy config file (default: /etc/haproxy/haproxy.cfg)
• PROXY_LOGLEVEL - Log level for HAProxy (default: notice)
• HTTP_PORT - The container binds to this port for handling HTTP requests (default: 80)
• HTTPS_PORT - The container binds to this port for handling HTTPS requests (default: 443)
• HTTPS_FORWARDED_PORT - The port set in the X-Forwarded-Port header of requests sent to the Manager/Keycloak (default: %[dst_port] this is the HAProxy port)
• NAMESERVER - The nameserver hostname and port used for resolving the Manager/Keycloak hosts (default: 127.0.0.11:53)
• MANAGER_HOST - Hostname of OpenRemote Manager (default: manager)
• MANAGER_WEB_PORT - Web server port of OpenRemote Manager (default 8080)
• MANAGER_MQTT_PORT - MQTT broker port of OpenRemote Manager (default 1883)
• MANAGER_PATH_PREFIX - The path prefix used for OpenRemote Manager HTTP requests (default not set, example: /openremote)
• KEYCLOAK_HOST - Hostname of the Keycloak server (default: keycloak)
• KEYCLOAK_PORT - Web server port of Keycloak server (default 8080)
• KEYCLOAK_PATH_PREFIX - The path prefix used for Keycloak HTTP requests (default not set, example: /keycloak)
• LOGFILE - Location of log file for entrypoint script to write to in addition to stdout (default none)
• AWS_ROUTE53_ROLE - AWS Route53 Role ARN to be assumed when trying to generate wildcard certificates using Route53 DNS zone, specifically for cross account updates (default none)
• LE_EXTRA_ARGS - Can be used to add additional arguments to the certbot command (default none)
• DISABLE_ACME - Disable certbot/ACME initialization and renewal logic in the entrypoint; useful when TLS is terminated externally such as with ACM on an AWS load balancer (accepted true values: 1, true, yes, on)
• SISH_HOST - Defines the destination hostname for forwarding requests that begin with gw- used in combination with SISH_PORT
• SISH_PORT - Defined the destination port for forwarding requests tha begin with gw- used in combination with SISH_HOST
• MQTT_RATE_LIMIT - Enable rate limiting for MQTT connections (connections/s)

Custom certificate format

Any custom certificate volume mapped into /etc/haproxy/certs should be in PEM format and must include the full certificate chain and the private key, i.e.:

cat privkey.pem cert.pem chain.pem > ssl-certs.pem

See haproxy SSL cert documentation.

Edge gateway tunnelling using SISH

The built in haproxy.cfg has support for forwarding requsts beginning with gw- to https://SISH_HOST:SISH_PORT just define these environment variables to enable this.

Kubernetes

When running the proxy in Kubernetes make sure to set the HTTP_PORT and HTTPS_PORT environment variables to a non-privileged port (> 1024). If you use an Ingress, reconfigure the HTTPS_FORWARDED_PORT to the HTTPS port of your Ingress (443).

You will also need to set the NAMESERVER environment variable to the cluster DNS (usually 10.96.0.10:53). The cluster DNS typically only resolves fully qualified hostnames, so make sure to set these using the MANAGER_HOST and KEYCLOAK_HOST environment variables (e.g. manager.default.svc.cluster.local).

Edge-Terminated TLS

If TLS is terminated upstream before traffic reaches this pod, for example by an AWS NLB with ACM, an ALB, an ingress controller, or another reverse proxy, then:

• Set DISABLE_ACME=true to disable certbot initialization and renewal in the container
• Use HAPROXY_CONFIG=/etc/haproxy/haproxy-edge-terminated-tls.cfg
• Set HTTP_PORT to a non-privileged container port such as 8080
• Set HTTPS_FORWARDED_PORT=443 so upstream services see the original external HTTPS port
• Configure the upstream load balancer or proxy to forward decrypted HTTP traffic to the pod HTTP_PORT

For MQTT in the same setup, if MQTT TLS is also terminated upstream:

• Terminate TLS on the upstream listener (for example external port 8883)
• Forward plaintext TCP traffic from that listener to the pod's MQTT port
• The provided haproxy-edge-terminated-tls.cfg listens for MQTT on MANAGER_MQTT_PORT and forwards it to the configured manager MQTT backend

The haproxy-edge-terminated-tls.cfg file removes local TLS certificate usage from the pod and preserves the usual X-Forwarded-* HTTP headers for upstream applications. Do not use this config if HTTPS or MQTT TLS is still passed through to the pod.