consensus-entrypoint always overrides BASE_NODE_P2P_ADVERTISE_IP and exits if public IP discovery fails

[Gercekkriptocu]

Summary

consensus-entrypoint always calls get_public_ip and assigns its result to BASE_NODE_P2P_ADVERTISE_IP, overriding any value the operator set in their env file. If none of the four upstream IP services are reachable, the script exits with code 8 and the node fails to start.

What I found

In consensus-entrypoint:

• The script unconditionally invokes get_public_ip against four external endpoints (ifconfig.me, api.ipify.org, ipecho.net, v4.ident.me).
• It then runs export BASE_NODE_P2P_ADVERTISE_IP=$PUBLIC_IP, which silently replaces any value already provided via .env.
• If all four endpoints are unreachable (firewalled host, restricted egress, transient network failure), the script exits with code 8 and the consensus container will not start.

Expected behavior

• If the operator has already set BASE_NODE_P2P_ADVERTISE_IP, the script should respect that value and skip discovery.
• Discovery failure should not be a hard startup failure when the operator has provided their own value.

Suggested fix

Skip discovery when BASE_NODE_P2P_ADVERTISE_IP is already set:

if [[ -z "${BASE_NODE_P2P_ADVERTISE_IP:-}" ]]; then ... fi

Impact

• Operators on hosts with restricted outbound HTTP cannot start the consensus node even when they know their public IP.
• Operators behind NAT or with a static advertised IP cannot pin it through env config.