Hello ,

We have been using Railway to host several backend microservices, and we’ve recently encountered a recurring issue that has caused production downtime on multiple occasions.

Twice now, one of our custom subdomains—already correctly connected to a Railway service—has unexpectedly required a new CNAME record to be configured. There was no prior alert, notification, or warning from Railway. Each time, we only discovered the issue after the service became unavailable, which led to extended downtime while we investigated and debugged the root cause.

Our domain is managed externally, and we rely on stable CNAME records to route traffic to our backend services. From our observations, this CNAME requirement appears to change periodically (approximately every three months), which makes it extremely difficult to operate reliably. Even a minute of downtime is unacceptable for us, and this behavior results in significantly more disruption than that.

Could you please clarify:

• Why these CNAME changes are required after a period of time?

• Whether this behavior is expected or indicative of a misconfiguration?

• If there is a way to configure a permanent or stable DNS target to avoid these recurring changes?

• Whether Railway can provide alerts or notifications before such changes impact live services?

This issue is becoming a serious operational risk for us, and we need a reliable way to prevent DNS-related outages going forward.

Thank you in advance for your support. We appreciate any guidance or solution you can provide.