Bare-metal Kura nodes (Dedibox, OVH, Scaleway Elastic Metal) now auto-recover from silent kernel lockups instead of staying wedged and blocking every production deploy. Provisioned boxes get a hardware watchdog and lockup-aware panic sysctls that reboot a stuck node in about 30 seconds while keeping the warm /var/cache/kura store intact. Existing boxes are covered by a per-fleet MachineHealthCheck that remediates a Node that stays NotReady for 10 minutes, so deploys are no longer gated on a single frozen box waiting for a manual IPMI reset.
Hive
Kura self-heals bare-metal nodes that hit a silent kernel lockup
Published
Jul 02, 2026 · 09:52 UTC
Repository
tuist/tuist