Summary
Both railway scale (top-level) and railway service scale (subcommand) panic
with a GraphQL field error, making CLI-based replica scaling impossible. The
Railway dashboard works fine; only the CLI surface is broken.
Environment
- CLI version: 4.33.0 (output of
railway --version)
- OS: macOS (darwin)
- Project type: standard Railway project with a service that supports
multi-replica scaling (queue worker)
Reproduction
railway loginrailway link to a project that has a service supporting multi-replica- Try either of:
railway scale --service "Queue Worker" --replicas 5
railway service scale --service "Queue Worker" --replicas 5
Expected behavior
The command should adjust the replica count for the targeted service in the active environment and return either a success confirmation or the new replica state.
Actual behavior
The CLI panics/errors out before reaching the API mutation:
Cannot query field "railwayMetal" on type "Region"
Impact
Blocks any agent-based or scripted ops automation that needs to scale workers programmatically (e.g., scaling up for a bulk job, scaling back down).
Forces every scaling action through the dashboard, which is impractical when driving Railway from a CI step or an autonomous agent.
Workflows that previously relied on railway scale in scripts now silently fail or require human intervention.
Likely root cause
The CLI's GraphQL query references a railwayMetal field on the Region type that has been removed (or renamed) on the Railway API server. The CLI needs either to drop the obsolete field from its query, or the API needs to expose it again for backward compatibility.
Workaround
Use the Railway dashboard: Project → Service → Settings → Deploy → Replicas.
Summary
Both
railway scale(top-level) andrailway service scale(subcommand) panicwith a GraphQL field error, making CLI-based replica scaling impossible. The
Railway dashboard works fine; only the CLI surface is broken.
Environment
railway --version)multi-replica scaling (queue worker)
Reproduction
railway loginrailway linkto a project that has a service supporting multi-replicaExpected behavior
The command should adjust the replica count for the targeted service in the active environment and return either a success confirmation or the new replica state.
Actual behavior
The CLI panics/errors out before reaching the API mutation:
Cannot query field "railwayMetal" on type "Region"
Impact
Blocks any agent-based or scripted ops automation that needs to scale workers programmatically (e.g., scaling up for a bulk job, scaling back down).
Forces every scaling action through the dashboard, which is impractical when driving Railway from a CI step or an autonomous agent.
Workflows that previously relied on railway scale in scripts now silently fail or require human intervention.
Likely root cause
The CLI's GraphQL query references a railwayMetal field on the Region type that has been removed (or renamed) on the Railway API server. The CLI needs either to drop the obsolete field from its query, or the API needs to expose it again for backward compatibility.
Workaround
Use the Railway dashboard: Project → Service → Settings → Deploy → Replicas.