Traefik Mesh sidecar resource limits not scaling with traffic

Traefik Mesh Sidecar Scaling

Goal

Reduce operational costs for a low‑traffic microservice by minimizing the CPU and memory overhead introduced by the Envoy sidecar injected by Traefik Mesh.

Constraints and Uncertainty

Each pod receives a sidecar by default, and its resource requests and limits can be tuned via pod annotations or namespace defaults. The Mesh Gateway feature can expose services externally without injecting a sidecar, but requires explicit configuration. The control plane can be scaled down to a single replica for cost savings, though this trades high‑availability. It is unclear whether Traefik Mesh automatically adjusts sidecar resource limits in response to traffic patterns or if manual tuning is required.

Unresolved Decision / Behavior

Does Traefik Mesh provide automatic scaling of Envoy sidecar resource limits based on observed traffic? If not, what mechanisms can be employed to trigger dynamic adjustment, and how can these changes be verified?

1. Does Traefik Mesh automatically adjust Envoy sidecar resource limits as traffic fluctuates?
2. If automatic scaling is absent, what configuration options exist to apply dynamic limits or trigger manual scaling?
3. How can resource usage be monitored and compared before and after applying Mesh Gateway or manual limit adjustments?