macOS (vz)
Same kling. Another hypervisor.
Three ways to use a Mac
Section titled “Three ways to use a Mac”| Way | What runs where |
|---|---|
| CLI → Linux host | kling context add lab ssh://… |
| Native (vz) | daemon on the Mac, one kling-vz per microVM |
| Lima | Firecracker in a nested arm64 Linux VM (M3+, macOS 15+) |
The backend is chosen in the config
Section titled “The backend is chosen in the config”kling config set daemon.vmm vz # "vz" or "firecracker"kling status -v # tells which backend it uses: backend: vzLeft empty, it uses the platform’s: vz on macOS, firecracker on Linux.
Where it lives
Section titled “Where it lives”| macOS | Linux | |
|---|---|---|
| data | ~/Library/Application Support/kindling |
/var/lib/kindling |
| socket | ~/Library/Application Support/kindling/kling.sock |
/run/kling.sock |
| daemon user | you | root (the VMM drops to kindling) |
What changes
Section titled “What changes”- A restore costs ~350 MiB: the golden’s memory is not shared.
kling psshows a loopback port, not an IP:running 127.0.0.1:61234.kling db upon an M4: 526–562 ms with the key already rotated.- With Lima, a cold
initializedrops from ~16 s to ~2.5 s with-bundleand-cpu-pct 100.