Home

LXC AutoScale Wiki

An async daemon that adjusts CPU and memory of Proxmox LXC containers from real-time usage, running on the node or driving one over SSH.

This wiki is the operational companion to the documentation site. The docs are the reference: installation, configuration, tiers, notifications, Docker, security. The wiki covers three things the reference cannot: which advertised features actually work right now, what went wrong for other people, and how the numbers this daemon acts on are measured.

Reference material lives on the docs site

• Getting started, Docker, service management
• Configuration, tiers, defaults
• Troubleshooting, logging, security

In this wiki

• Feature status: what is production-ready, what is experimental, and what was recently broken and is now fixed but only on main. Read this before enabling anything beyond vertical scaling.
• How usage is measured: cgroup accounting, the LXCFS fallback, and why the daemon's numbers can disagree with the Proxmox web UI. Most reports in this tracker start here.
• Field notes: closed reports, the cause each turned out to have, and what it means for you.
• FAQ: scope, the sibling projects, and what "experimental" means in practice.

Asking a question

Q&A covers the four things worth checking before posting. If you use cpu_pinning, read the advisory first.

If you are installing today

Vertical scaling of CPU and memory is the mature path: it is what the test suite covers and what most of this tracker is about. Horizontal scaling is still explicitly experimental.

CPU pinning worked in none of its forms until recently. The three defects behind that are fixed and closed, but the fixes are on main and not in any tagged release yet, and a bad cpuset written by the old bug stays in the container config until you replace or clear it. If you have ever set cpu_pinning, read Feature status before assuming either that it works or that it does not.