fleetlock

Notable changes between versions.

Latest

v0.4.0

Identify Kubelet nodes via MachineID instead of SystemUUID (action required) (#96)
- Host SystemUUID and MachineID are identical on most platforms, but not on Azure or bare-metal
- Kubelet detected SystemUUID via sysfs, but MachineID requires Kubelet be able to read /etc/machine-id from the host (ro mount when Kubelet run as a container) (action required)
- Add support for Azure and bare-metal plaforms by changing to MachineID
Reset fleet_lock_state gauge on first lock or unlock call (#71)
Improve reject reply when a client attempts to unlock a lock it doesn't own (#71)

v0.3.0

Add support for Kubernetes node draining (#51)
Automate base image, Go version, and module dependency updates
Build multi-arch container images (amd64, arm64) (#15)
Switch to using the Prometheus collectors package (#37)

v0.2.0

Add Prometheus /metrics endpoint (#4)
Add JSON error responses (#9)
Fix -version command output (#6)

v0.1.0

Implement the FleetLock protocol backed by Kubernetes coordination API
Support reboot groups with separate fleetlock-group Leases
Use a Role with coordination lease create, get, and update
- Respect NAMESPACE if set via downward API, default to "default"