Pnetlab 5.3.11 May 2026

  • Auto‑Scaling Engine – A lightweight daemon monitors the metrics (CPU, memory, network I/O, custom Prometheus queries) across all running containers.
  • Provisioning – When a rule fires, Pnetlab pulls the appropriate Docker image (or LXC template) from the local registry, creates the container, attaches it to the existing virtual network, and updates the lab’s topology graph on‑the‑fly.
  • Graceful De‑Provisioning – When the metric falls back below the defined hysteresis threshold, the engine gracefully drains traffic from the extra node(s) and shuts them down, preserving any stateful data you’ve chosen to retain (e.g., config files saved to a persistent volume).

    • Fix: This version introduced more aggressive memory ballooning. Disable it globally: System > Preferences > QEMU > Disable Memory Ballooning (Check) > Save.

    | Aspect | Benefit | |--------|----------| | Resource Efficiency | Nodes that are idle for a configurable period are automatically powered down or placed in a lightweight “sleep” mode, freeing CPU, RAM, and storage for other labs. | | Instantaneous Scaling | When a lab’s traffic spikes (e.g., during a simulated DDoS or a large‑scale routing convergence test), Pnetlab instantly spins up additional instances of the required device types to handle the load. | | Policy‑Driven Control | Administrators can define policies per‑lab, per‑user group, or per‑device type (e.g., “always keep 2 routers online, but allow up to 5 when CPU > 70 %”). | | Cost‑Optimized Cloud Deployments | In cloud‑hosted environments (AWS, Azure, GCP), the feature ties into the provider’s auto‑scaling groups, reducing operational spend by scaling only when needed. | | Seamless Integration with Existing Workflows | The auto‑scaling engine works transparently with the classic “Start/Stop” UI, the API, and the CLI, so scripts and automation pipelines require no changes. | | Granular Monitoring & Alerts | Built‑in dashboards show real‑time scaling events, and you can hook into Prometheus or Grafana for custom alerts (e.g., “scale‑up event triggered on Lab‑A at 14:03 UTC”). | Pnetlab 5.3.11