8 days ago
During isolated nonproduction application qualification, we recorded two separate failures using the official Railway CLI 5.71.0 base GraphQL API command:
- A read-only project lookup returned a GraphQL error categorized INTERNAL_SERVER_ERROR on the project path.
- A separate read-only environment lookup, with configuration decryption explicitly disabled, returned INTERNAL_SERVER_ERROR on the environment path before we accepted an identity or configuration result.
Each final diagnostic dispatched once with no retry. We retained only content-free error categories, paths, process status, byte counts and hashes. We have not established a shared cause or an authorization failure. Our earlier receipt review was completed on November 10, 2026; precise server request timestamps and correlation identifiers were not captured.
These blocked diagnostic paths performed no SSH-key registration, configuration change, source upload, deployment creation or AI-provider request. A later browser visit can open the project dashboard, but that does not establish that these exact API lookups recovered.
Could Railway confirm whether these resolver failures correspond to a known control-plane issue, or provide the supported remediation and the request-scoped correlation information needed to diagnose them? Please also clarify which control-plane logs are available and their retention window. If target identifiers are required, please provide a private channel for them; we have intentionally omitted resource IDs, hostnames, credentials, raw responses, configuration and customer data from this public thread.
This concerns the Railway control-plane lookups. We are not requesting application-code debugging. We will verify the exact resolver paths after recovery before resuming qualification.
1 Reply
8 days ago
On our GraphQL API, an INTERNAL_SERVER_ERROR category does not by itself mean a server-side crash or a control-plane fault. It is the default GraphQL code for any error that does not carry a code of its own, and our authorization refusals carry none, so those responses typically come back with HTTP 200.
The most common cause of a project or environment lookup failing this way is the credential. A token that cannot access that project or environment, such as a workspace token for a different workspace or a project token scoped to another project, gets refused. So does a token that no longer exists. Every API token, project token and CLI session created before the October 21 incident was revoked, and those credentials cannot be reactivated.
You can check which tokens exist and their scope in the Scope column at railway.com/account/tokens. For the CLI, railway logout followed by railway login issues a fresh session.
Status changed to Awaiting User Response Railway • 8 days ago
19 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 19 hours ago