We have an existing production Daemon on Vercel Pro, configured as one instance in one region with 1 GiB memory.
New production source deployments finish READY and successfully build a different container image. The production web alias points to the new deployment, but vercel daemons inspect continues to show the old daemon deployment ID, old image digest, and unchanged instance timestamps.
Supported steps already attempted:
- A normal source deployment using the official Vercel CLI.
- A real promotion to another READY deployment containing the same reviewed application and worker source (promotion succeeded).
- Resetting temporary scaling overrides to the original one-instance configuration.
- One documented instance restart. It was accepted, and application health evidence confirmed graceful release and reconnection within a few seconds. The daemon still did not adopt the new image.
The application is available, but the optimized worker release cannot be verified as active. Runtime behavior still shows an older code path, so this is more than an old timestamp in the UI.
Is there a supported way to reconcile an existing production Daemon to the image of the current production deployment while keeping the same instance count and memory? Could the Daemons team investigate a deployment binding/control-plane issue?
Private project/deployment identifiers and detailed evidence are available through a private channel. This public report deliberately excludes account names, domains, IDs, credentials, financial information, and raw logs. Please do not suggest increasing capacity or changing billing settings as a workaround.