a month ago
Hello Railway Support Team,
We are conducting a security assessment of our production infrastructure hosted on Railway.
Our application provides IT diagnostic tools, including an SSL/TLS certificate checker that establishes outbound connections to user-specified public hosts.
Before enabling this functionality publicly, we need to validate the network egress architecture to ensure that outbound connections cannot unintentionally reach private, loopback, link-local, or otherwise restricted destinations.
Our current deployment:
- Platform: Railway
- Service: Web
- Region: US West (California, USA)
- Replicas: 1
We would appreciate clarification on the following points:
- Outbound IPv4 and IPv6 connectivity
Does Railway provide native outbound IPv4 and IPv6 connectivity for services deployed in this region?
Are there any differences in outbound routing behavior between IPv4 and IPv6?
- NAT64 / 464XLAT
Does Railway use NAT64, 464XLAT, DNS64, or any other IPv6-to-IPv4 translation mechanism for outbound connections in this environment?
If so:
- Which translation mechanisms are used?
- What IPv6 translation prefixes are configured?
- Are these prefixes fixed or subject to change?
- Does the configuration vary by region, service, or deployment?
- Outbound destination security
Does Railway provide network-level controls that can restrict outbound connections to public IP addresses only?
Specifically, can outbound traffic to private, loopback, link-local, metadata service addresses, and other non-public destinations be blocked at the network egress layer?
If such controls are available, how can they be configured and verified?
- Configuration evidence
For our security assessment, we need evidence applicable to the actual production egress path.
Can Railway provide documentation or confirmation covering:
- The outbound network architecture for our deployment region.
- Whether NAT64/464XLAT is enabled or disabled.
- Any configured translation prefixes.
- Available network-level egress restrictions.
- Whether these characteristics are guaranteed or may change without notice.
We understand that some internal infrastructure details may not be publicly disclosed. A written technical confirmation or reference to authoritative documentation would be sufficient for our initial assessment.
Our objective is to ensure that our application can safely establish outbound TLS connections without creating unintended access paths to restricted network destinations.
Thank you for your assistance.
1 Reply
a month ago
Services get outbound IPv4 by default. Outbound IPv6 is opt-in per service under the service's Settings, in the Networking section (Enable Outbound IPv6). While it is off, IPv6 connection attempts fail with ENETUNREACH. When it is on, IPv4 and IPv6 both work at the same time. See Outbound Networking.
Outbound IPv6 is routed natively. It does not use NAT64, 464XLAT or DNS64 translation, so there are no IPv6 translation prefixes to account for.
The outbound settings available for a service are Outbound IPv6 and, on Pro, Static Outbound IPs. Neither one restricts which destinations a service can reach. Outbound connections to private, link-local or metadata-style addresses are not blocked by default, so for a certificate checker that connects to hosts users supply, the destination filtering needs to happen in your application. Filter on the resolved IP address, both IPv4 and IPv6, and check it at connect time.
We don't publish a written guarantee that this egress architecture won't change.
Status changed to Awaiting User Response Railway • about 1 month ago
24 days ago
This thread has been marked as solved automatically due to a lack of recent activity. Please re-open this thread or create a new one if you require further assistance. Thank you!
Status changed to Solved Railway • 24 days ago