Skip to content
SSL certificates for custom domains before DNS cutover
neowceng
OP

8 days ago

Hi, we are preparing to migrate uniquehome.my and www.uniquehome.my to Railway. Both domains show ownership verified, but their certificates show ISSUE_FAILED with an internal error. DNS traffic still points to the current provider because the existing website must remain available during migration.

Is there a supported way to issue certificates for both domains before changing the live DNS records? If not, what is the safest cutover and rollback process to minimise HTTPS downtime? We also need to preserve the existing DNS records for email and other services, including the current _acme-challenge record.

Thanks.

Solved

3 Replies

Railway
BOT

8 days ago

We can't issue certificates for these domains before the cutover. They are validated by the certificate authority fetching a token over plain HTTP from the domain itself, so a certificate can only issue once each domain's traffic record points at Railway. Both domains already have ownership verified, so the remaining step is changing each traffic record to the CNAME target shown for that domain in your service's Settings → Networking.

For the cutover, change only those two traffic records and leave your email records and the existing _acme-challenge record alone. These domains don't use _acme-challenge for validation, so it doesn't need to change. Once the new records have propagated, use the "Try Again" button on the "Failed to issue TLS certificate" banner for each domain to start a fresh certificate order. The current failed attempts won't recover by themselves, and you don't need to remove and re-add the domains. Expect a short window where some visitors still reach the old site or see HTTPS errors while DNS caches expire. How long that lasts depends on the TTL of the records you're replacing.

To roll back, change the two traffic records back to their previous values. Rolling back is subject to the same TTL-based propagation, and your other records stay untouched either way.


Status changed to Awaiting User Response Railway • 8 days ago


neowceng
OP

7 days ago

Thanks. We still need clarification before risking the live site. Could a Railway team member confirm these points?

  1. For ordinary apex and www domains, is HTTP validation after traffic cutover the only supported option? Is there any supported DNS-only prevalidation, or a documented way for the current host to serve only the ACME HTTP challenge while keeping normal traffic there? We will not replace the existing _acme-challenge record or weaken TLS validation.

  2. If traffic is moved and issuance still returns the internal error, what exact checks distinguish DNS propagation from a certificate-order/platform failure? What retry and rate-limit guidance applies, and when should we roll back rather than retry? Please clarify whether the current failed orders require a manual retry as the automated reply states.

  3. After rolling DNS back, should the Railway domain bindings remain in place for a later attempt? Are there certificate-order or renewal consequences we should account for? We understand DNS rollback is not instantaneous and does not guarantee uninterrupted HTTPS.

We are asking for a supported procedure and its limitations, not a guarantee of zero downtime. Please do not make any changes on our behalf. Thanks.


Status changed to Awaiting Railway Response Railway • 7 days ago


Railway
BOT

7 days ago

For ordinary apex and www domains, we validate certificates over HTTP after traffic cutover. We can't prevalidate either domain for you, and we can't issue their certificates from a DNS-only record such as your _acme-challenge record. Both traffic records don't point at us yet, so the current failures are expected. The internal error is a generic label and doesn't mean anything is wrong on our side. The failed orders won't recover by themselves. After cutover, use "Try Again" on the "Failed to issue TLS certificate" banner in the service's Settings → Networking to start a fresh order.

After cutover, check that each traffic record in Networking shows as propagated and that plain HTTP to the domain no longer reaches your old host. The certificate authority checks from several networks, so one cached answer that still points at the old host can fail an order. If a fresh attempt fails while the records are correct, wait out the previous TTL before retrying, and avoid rapid repeated retries. If the error becomes a rate limit, the "Try Again" button isn't shown, and waiting is the only remedy.

If you roll back, leave both domain entries in place. Removing and re-adding a domain gives it a new CNAME target and verification token, which means updating those DNS records again. Lowering the TTL on the two traffic records well before cutover shortens both the switch and any rollback.


Status changed to Awaiting User Response Railway • 7 days ago


Railway
BOT

11 hours 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 • about 11 hours ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...