kuke daemon

Manage the kukeond daemon cell lifecycle.

kuke daemon [command]

These commands act on the kukeond cell provisioned by kuke init (kuke-system / kukeon / kukeon / kukeond). They run in-process because the daemon they manage may not be running at the time the command is invoked.

Subcommands

Command	What it does
kuke daemon start	Start the kukeond daemon cell
kuke daemon stop	Gracefully stop the kukeond cell (SIGTERM, escalating to SIGKILL)
kuke daemon kill	Immediately SIGKILL the kukeond daemon cell
kuke daemon restart	Stop then start the kukeond daemon cell
kuke daemon reset	Stop, delete metadata + cgroups, clear /run/kukeon/kukeond.{sock,pid}
kuke daemon recreate	Recreate the kukeond daemon cell (tear down, re-provision, start)
kuke daemon logs	Print the kukeond daemon's stdout/stderr (use -f to follow)

All subcommands are idempotent: they succeed with a clear message when the daemon is already in the requested state.

Targeting a specific instance

Every kuke daemon … subcommand accepts --server-configuration <path> (default /etc/kukeon/kukeond.yaml) to point at a non-default kukeond instance. The precedence chain matches kuke init and kuke uninstall:

1. --server-configuration <path> (flag)
2. KUKEOND_CONFIGURATION environment variable
3. /etc/kukeon/kukeond.yaml (default file)
4. Hardcoded defaults (absent file)

The same chain kukeond itself uses, so a --server-configuration ./kukeond-dev.yaml on any admin command points at the same document the daemon honors. Running sudo kuke daemon stop --server-configuration ./kukeond-dev.yaml signals the dev instance only; the prod kukeond (under the default /etc/kukeon/kukeond.yaml) is untouched.

kuke daemon start

Bring up the existing kukeond cell provisioned by kuke init. Returns success when the daemon is already running. Errors when the host has not been initialized — run kuke init first.

kuke daemon stop

Send SIGTERM and wait up to --timeout (default 10s) for the daemon to exit; if the grace period expires, escalate to SIGKILL.

Flag	Default	Description
--timeout	10s	Grace period before escalating from SIGTERM to SIGKILL

kuke daemon kill

Force-kill the kukeond cell with no grace period. This is the escape hatch for a hung or unresponsive daemon — use kuke daemon stop for the graceful path.

kuke daemon restart

Compose kuke daemon stop and kuke daemon start into a single verb. When the daemon is already stopped, the stop phase is skipped and the start phase still runs.

Flag	Default	Description
--timeout	10s	Grace period for the stop phase before escalating from SIGTERM to SIGKILL

kuke daemon recreate

sudo kuke daemon recreate --kukeond-image <image> [--timeout <duration>] [--server-configuration <path>]

Recreate the kukeond daemon cell (tear down, re-provision, start).

Compose kuke daemon reset and kuke init's kukeond cell provisioning into a single verb.

Tears down the existing kukeond cell (stop, delete, clear socket+pid) and re-provisions it from scratch using the specified --kukeond-image. The cell-creation path is shared with kuke init, so the two cannot drift.

Requires --kukeond-image and errors with ErrHostNotInitialized when the host has not been bootstrapped by kuke init yet.

Image-load is out of scope: the desired image must already be present in the kuke-system realm (via kuke build or kuke image load) before invoking this command.

Flag	Default	Description
--kukeond-image	(required)	Container image for kukeond. May be omitted if kukeondImage is set in /etc/kukeon/kukeond.yaml; otherwise required.
--timeout	10s (lifecycle.DefaultTimeout)	Grace period for the stop phase before escalating from SIGTERM to SIGKILL
--server-configuration	/etc/kukeon/kukeond.yaml	Path to the kukeond server-configuration YAML (precedence chain: flag > KUKEOND_CONFIGURATION env > default file > hardcoded defaults)

kuke daemon reset

Lightweight dev-loop teardown of the kukeond daemon. Stops the cell (with the same SIGTERM → SIGKILL escalation as daemon stop), deletes the cell metadata + cgroups, and clears the transient files /run/kukeon/kukeond.{sock,pid}. User-realm data under /opt/kukeon/default/** is left intact, so re-running kuke init after kuke daemon reset produces a clean re-bootstrap.

Always runs as root: it touches /sys/fs/cgroup, containerd namespaces, and /opt/kukeon. Fails fast with a clear remediation if you forget sudo.

Distinct from kuke uninstall, which is the per-host teardown (every realm, the kukeon system user/group, and the run path itself).

Flag	Default	Description
--purge-system	false	Also remove /opt/kukeon/kuke-system (user-realm data is still preserved)
--timeout	10s	Grace period for the stop phase before escalating from SIGTERM to SIGKILL

kuke daemon logs

Print the kukeond container's stdout/stderr stream. Shortcut for kuke log --realm kuke-system --space kukeon --stack kukeon kukeond — the coordinates are static and filled in for you.

kuke daemon logs [-f]
kuke daemon log  [-f]      # alias

By default the current contents are printed and the command exits; pass -f/--follow to tail until SIGINT.

Flag	Default	Description
--follow, -f	false	Tail until SIGINT instead of printing current contents and exiting

Examples

# Bring the daemon back after a host reboot
sudo kuke daemon start

# Graceful restart after editing /etc/kukeon/kukeond.yaml
sudo kuke daemon restart

# Force-kill an unresponsive daemon
sudo kuke daemon kill

# Dev loop: blow away the daemon cell and re-init
sudo kuke daemon reset
sudo kuke init --kukeond-image docker.io/library/kukeon-local:dev

# Pick up a new kukeond image without a full kuke init
sudo kuke daemon recreate --kukeond-image docker.io/library/kukeon-local:dev

# Wipe /opt/kukeon/kuke-system too (user-realm data stays)
sudo kuke daemon reset --purge-system

# Tail the daemon log live
sudo kuke daemon logs -f

Related

• kuke init — provisions the daemon cell that kuke daemon … manages
• kuke uninstall — full-host teardown; the next step up from daemon reset --purge-system
• kukeond — the daemon binary itself