Please see this docs section - https://docs.railway.app/guides/private-networking#initialization-time

I’m not sure this is the solution. The application retries connecting to the service every second. The container crashes after a minute. That should be enough time for the network to come up.

Railway restarting the container does not solve this, please try adding a 3-second delay to the start of your app.

Another thing to note is that the service you are trying to connect to needs to be in the same project.

Adding it via the start command (sleep 3 && …) results in the container changing to "completed" state with no logs.
Adding it via the Dockerfile also doesn't help, because of the start command override.

The app is checking, as you can see over multiple seconds:

time="2024-03-25T19:15:03Z" level=info msg="Using S3 at bucket:9000"
time="2024-03-25T19:15:03Z" level=fatal msg="Get \"[http://bucket:9000/modules/?location=](http://bucket:9000/modules/?location=)": dial tcp: lookup bucket on [fd12::10]:53: dial udp [fd12::10]:53: connect: cannot assign requested address" op=storage

time="2024-03-25T19:15:04Z" level=info msg="Using S3 at bucket:9000"
time="2024-03-25T19:15:04Z" level=fatal msg="Get \"[http://bucket:9000/modules/?location=](http://bucket:9000/modules/?location=)": dial tcp: lookup bucket on [fd12::10]:53: dial udp [fd12::10]:53: connect: cannot assign requested address" op=storage

time="2024-03-25T19:15:05Z" level=info msg="Using S3 at bucket:9000"
time="2024-03-25T19:15:05Z" level=fatal msg="Get \"[http://bucket:9000/modules/?location=](http://bucket:9000/modules/?location=)": dial tcp: lookup bucket on [fd12::10]:53: dial udp [fd12::10]:53: connect: cannot assign requested address" op=storage

time="2024-03-25T19:15:07Z" level=info msg="Using S3 at bucket:9000"
time="2024-03-25T19:15:07Z" level=fatal msg="Get \"[http://bucket:9000/modules/?location=](http://bucket:9000/modules/?location=)": dial tcp: lookup bucket on [fd12::10]:53: dial udp [fd12::10]:53: connect: cannot assign requested address" op=storage

You are using a Dockerfile, thus the start command in the service settings overwrites the ENTRYPOINT in exec format meaning shell features like && don't work, you would need to wrap your start command in a shell - /bin/sh -c "sleep 3 && [start command]"

The app is checking, as you can see over multiple seconds

From my perspective, this looks like the container crashing and Railway restarting it, and as previously mentioned, this behaviour does not solve the issue.

Can you confirm the bucket service is located within the same project as the service that's trying to connect to it?

Yes, I can confirm, bucket and the service are running in the same project.
Sorry for dragging this out, wrapping the start command in a shell worked perfectly. 🙏