Components Overview

LXC AutoScale ML consists of three main components that work together to provide intelligent autoscaling.

Architecture

┌─────────────────────────────────────────────────────────────┐
│                      Proxmox Host                           │
│                                                             │
│  ┌───────────────┐  ┌───────────────┐  ┌───────────────┐   │
│  │    Monitor    │──│     Model     │──│      API      │   │
│  │   (Collect)   │  │   (Decide)    │  │   (Execute)   │   │
│  └───────────────┘  └───────────────┘  └───────────────┘   │
│         │                  │                   │            │
│         ▼                  ▼                   ▼            │
│  lxc_metrics.json    Predictions         pct commands      │
│                                                             │
│  ┌─────────┐ ┌─────────┐ ┌─────────┐                       │
│  │ LXC 101 │ │ LXC 102 │ │ LXC 103 │  ...                  │
│  └─────────┘ └─────────┘ └─────────┘                       │
└─────────────────────────────────────────────────────────────┘

Component Summary

Component	Service	Purpose	Port
API	lxc_autoscale_api	RESTful interface for scaling operations	5000
Model	lxc_autoscale_ml	ML engine for scaling decisions	-
Monitor	lxc_monitor	Metrics collection from containers	-

Data Flow

1. Metrics Collection

The Monitor service runs continuously, collecting resource metrics from all running LXC containers every 10 seconds (configurable).

Metrics collected:

• CPU usage percentage
• Memory usage (MB)
• Swap usage (MB)
• Disk usage (GB)
• Network I/O (bytes)
• Disk I/O (operations)
• Process count

Output: /var/log/lxc_metrics.json

2. ML Analysis

The Model service reads historical metrics and trains an IsolationForest model to detect anomalies in resource usage patterns.

Processing steps:

1. Load metrics from JSON file
2. Preprocess and engineer features
3. Train IsolationForest model
4. Batch fetch current container configurations
5. For each container:
   • Predict if usage is anomalous
   • Determine scaling action based on thresholds
   • Execute scaling via API

3. Scaling Execution

The API service receives scaling requests and executes them on the Proxmox host.

Capabilities:

• Scale CPU cores
• Scale memory (RAM)
• Increase storage
• Create/list/rollback snapshots
• Clone/delete containers
• Query resource status

Service Management

Start All Services

systemctl start lxc_autoscale_api lxc_monitor lxc_autoscale_ml

Stop All Services

systemctl stop lxc_autoscale_api lxc_monitor lxc_autoscale_ml

Check Status

systemctl status lxc_autoscale_api lxc_monitor lxc_autoscale_ml

View Logs

journalctl -u lxc_autoscale_api -f
journalctl -u lxc_monitor -f
journalctl -u lxc_autoscale_ml -f

Configuration Files

Component	Configuration Path
API	/etc/lxc_autoscale_ml/lxc_autoscale_api.yaml
Model	/etc/lxc_autoscale_ml/lxc_autoscale_ml.yaml
Monitor	/etc/lxc_autoscale_ml/lxc_monitor.yaml

Dependencies Between Components

Monitor ──writes──▶ lxc_metrics.json ◀──reads── Model
                                                  │
                                                  │ API calls
                                                  ▼
                                                 API
                                                  │
                                                  │ pct commands
                                                  ▼
                                            LXC Containers

The components can run independently:

• Monitor only: Collect metrics without scaling
• API only: Manual scaling via API
• All three: Full automated scaling

Next Steps

• API Component: RESTful interface details
• Model Component: ML engine and scaling logic
• Monitor Component: Metrics collection