Proxies

Configuring Replicated to work with an HTTP proxy

The Replicated install script provides several configuration options for working with an HTTP proxy. Passing the no-proxy flag to the install script will disable all proxy discovery and configuration.

Proxy Discovery

Replicated attempts to discover an HTTP proxy by checking the following environment variables in sequence:

  • HTTP_PROXY
  • http_proxy
  • HTTPS_PROXY
  • https_proxy

If none of these is set, Replicated will prompt the user to enter a proxy address or confirm that no proxy is required.

Proxy Configuration

If a proxy address is discovered or set, Replicated will assemble a default list of NO_PROXY addresses including localhost, the address of the docker0 interface, and the address of the server running the Replicated daemon. Additional addresses can be added with the additional-no-proxy parameter, which can be specified multiple times. Note that this parameter does not accept an IP range in CIDR notation.

Clusters

For clustered installations where remote nodes should be accessed without the proxy, each node’s IP must be specified separately with the additional-no-proxy flag.

cat kubernetes-init.sh | sudo bash -s additional-no-proxy=corporate.internal additional-no-proxy=10.128.0.9 additional-no-proxy=10.128.0.10

To prevent errors during image pulls, Replicated will add the HTTP_PROXY and NO_PROXY environment variables to docker and restart it.

The installer will configure the Replicated deployment to run with the HTTP_PROXY and NO_PROXY variables set in the environment.

Replicated will not add the HTTP_PROXY or NO_PROXY environment variables to any application pods automatically. Use the ConsoleSetting template function to add HTTP_PROXY to the environment of any of your application pods.

env:
  - name: HTTP_PROXY
    value: {{repl ConsoleSetting "http.proxy"}}