7 hours ago
Subject: Deployments stuck in QUEUED indefinitely across all services — not a billing or code issue
Body:
Project: creative-unity (675339c4-516c-4d33-9ab1-af70417697e2)
Environment: production (694a9a59-a374-424d-a37e-dd8594bcf796)
Affected services: karaoke-backend (9c5b0fa6-1482-4d40-9c95-cd08bfb4852a), karaoke-guest-app (39a7862c-2f1a-40d8-998d-f7801ec886bf), karaoke-admin-app (1fd5f41b-e4ff-474b-a217-bd3c65be911b), karaoke-kj-panel (0e12840e-3c0e-4a4b-9d9a-e17634e1ddb0)
All 4 services in this environment have deployments stuck in QUEUED status indefinitely, on every single deploy attempt over the past several hours (repro timestamps: 2026-09-16 00:20, 01:00, 01:12, 01:25, 01:30, 01:49 UTC).
Consistent symptom pattern (every attempt, every service):
SNAPSHOT_CODE, BUILD_IMAGE and PUBLISH_IMAGE all complete successfully — image builds and pushes with no errors.
The deployment then never transitions to DEPLOYING. It just sits in QUEUED forever. No error, no failure reason, no event after PUBLISH_IMAGE.
Ruled out already:
Not a code/build issue — builds complete cleanly every time, and we've fixed the one real code bug we found (a missing module causing an ImportError) days ago; the QUEUED-forever symptom persists unchanged after that fix.
Not billing/usage — checked the Usage page directly: current bill is $0.00, $15 of credits available, well under the included $5 Hobby allowance.
Not a plan/resource limit — plan limits (replicas, CPU, memory) are nowhere near being hit.
Not a platform-wide incident — nothing on the Railway status page for this window.
Example stuck deployment IDs for your reference:
karaoke-backend: fab6a59b-3c1b-4540-b226-78d5e15a8962, cda6b324-d789-4428-9ac2-b3d156379050, 88f14428-18a8-46fd-837b-aa2ec567f38a
karaoke-admin-app: 04b9caea-69dd-42a3-9a6d-361080abd5e5
karaoke-guest-app: 95a99bf2-d5ba-42fc-962a-44ca7f8e1708
Request: Please check the deployment scheduler/queue for this project's production environment and find out why builds complete but deployments never get picked up for the deploy stage. Manually unstick the queue if that's what's needed on your end.
1 Replies
6 hours ago
All 3 of your workspace's Hobby deploy slots are held by earlier karaoke-backend deployments that are stuck on the pre-deploy command (flask db upgrade && python manage.py add-club ... && python manage.py add-admin ...), which has been running for over 14 minutes without completing or exiting. Because every slot is occupied, all subsequent deployments across your four services are waiting behind them. The builds finish, but the deploy stage never starts because there is no free slot. To unblock: cancel those three stuck deployments from the karaoke-backend service's deployment list, then fix the pre-deploy command so it does not hang. A common cause is flask db upgrade waiting on a database lock, or the seed commands prompting for input or failing silently without exiting non-zero. Those seed commands also re-run on every deploy, so consider guarding them with an idempotent check or moving them out of the pre-deploy step entirely.
Status changed to Awaiting User Response Railway • about 7 hours ago