What is LXC AutoScale ML?

LXC AutoScale ML adjusts the CPU and memory allocated to Proxmox LXC containers. It collects per-container metrics, trains an anomaly detection model on each container's own history, and applies changes by running pct on the host. Resizing CPU and memory with pct set does not restart the container.

Key Features

ML-Driven Autoscaling

The system uses an IsolationForest machine learning model to detect anomalies in resource usage patterns. When unusual activity is detected, the model evaluates whether scaling is needed based on:

• Current CPU and memory utilization
• Historical usage trends
• Configured thresholds and limits

Incremental Scaling

Resources scale gradually to avoid instability:

• CPU: Adjusts by ±1 core per cycle (configurable)
• RAM: Adjusts by ±512 MB per cycle (configurable)

This prevents the system from jumping between minimum and maximum allocations.

Concurrent configuration fetching

Before deciding, the model reads each container's current allocation from the API. Those reads are issued concurrently, up to api.max_concurrent at a time (10 by default), so a cycle does not grow linearly with the number of containers. How much that saves depends on how quickly the API answers, which depends on how quickly pct config returns on your host.

Security

• API key authentication, off by default. When enabled, every endpoint except /health/check and /metrics requires a key, compared in constant time.
• Per-IP rate limiting, 120 requests per minute by default, with localhost exempt so the model is not throttled.
• Input validation on every parameter. Commands are handed to the kernel as argument lists and never go through a shell.

The API runs as root, because pct requires it, and binds to every interface by default. See Configuration before exposing it.

Fault Tolerance

The circuit breaker pattern automatically skips failed API endpoints, preventing cascading failures. The system recovers gracefully from crashes through automatic stale lock cleanup.

Architecture Overview

LXC AutoScale ML consists of three main components:

Component	Purpose	Service Name
API	RESTful interface for scaling operations	lxc_autoscale_api
Monitor	Collects resource metrics from containers	lxc_monitor
Model	ML engine for scaling decisions	lxc_autoscale_ml

Data Flow

LXC Containers
      │
      ▼ (metrics collection)
┌─────────────┐
│   Monitor   │──▶ /var/log/lxc_metrics.json
└─────────────┘
      │
      ▼ (read metrics)
┌─────────────┐
│    Model    │──▶ Train IsolationForest
└─────────────┘    Detect anomalies
      │            Predict scaling needs
      ▼ (API calls)
┌─────────────┐
│     API     │──▶ Apply scaling actions
└─────────────┘
      │
      ▼
LXC Containers (scaled)

Use Cases

Dynamic Web Applications

Scale container resources during peak traffic hours and reduce them during off-peak periods.

Development Environments

Automatically adjust resources based on build activity and test workloads.

Database Servers

Respond to query load variations by scaling memory and CPU as needed.

Batch Processing

Increase resources when batch jobs run and scale down after completion.

Comparison with Manual Scaling

Aspect	Manual Scaling	LXC AutoScale ML
Response time	Minutes to hours	Seconds
Accuracy	Based on estimates	Based on real data
Consistency	Human-dependent	Automated
Overnight coverage	Requires on-call staff	Continuous
Resource efficiency	Often over-provisioned	Optimized

Next Steps

• Getting Started: Install and configure LXC AutoScale ML
• Architecture: Understand the system design
• Configuration: Customize scaling behavior