LXC AutoScaleAutomatic scaling for Proxmox LXC containers

CPU, memory, and core pinning — managed by a lightweight daemon.

Get Started

View on GitHub

Vertical Scaling

Automatically adjusts CPU cores and memory for each container based on real-time usage thresholds.

Dual Backend

Operate via pct CLI commands, locally on the node or over SSH.

Tier System

Group containers by workload profile with per-tier thresholds, limits, and CPU core pinning.

cgroup-based Metrics

Reads CPU and memory from host-side cgroup accounting. No pct exec into containers, no LXCFS required.

Async Architecture

Fully non-blocking event loop. Concurrent data collection, fire-and-forget notifications, zero-sleep measurements.

Security Hardening

SSH host key verification, secret masking in logs, environment variable expansion, Pydantic config validation.

Overview

LXC AutoScale is an async resource management daemon for Proxmox environments. It monitors LXC container CPU and memory usage every polling cycle (default: 5 minutes) and adjusts allocations based on predefined thresholds.

It runs as a systemd service on the Proxmox host, or drives one remotely over SSH. Configuration is validated at startup via Pydantic models and supports per-container tier overrides.

Key capabilities

Async event loop with concurrent container data collection
Pydantic-validated YAML configuration with ${ENV_VAR} expansion
Automatic vertical scaling of CPU cores and memory
Per-container or per-group configuration via YAML tiers
CPU and memory measured from host-side cgroup (no pct exec needed)
CPU core pinning by L3 cache domain (CCD/CCX) or NUMA node on any CPU, and by P/E core on hybrid Intel (12th gen+)
Horizontal scaling via container cloning (experimental)
Timezone-aware off-peak energy mode
SSH connection pool with host key verification (default: reject)
Secret masking in all log output
Notifications via Email, Gotify, Uptime Kuma (async, with failure backoff)
Docker support with optional non-root user for API-only mode