Transition-only notification has a blind spot spotted on nas: while
stale <metadata> entries keep 'zpool status -x' unhealthy, the
watchdog parks in the alarm state and can never signal anything new.
Now it re-notifies every -i hours (default 24) while the problem
persists.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A SUSPENDED pool keeps its vdev ONLINE, so statechange-notify never
fires; data-class events need ZED_NOTIFY_DATA. Three nuc incidents
went unnoticed for days. This mails on healthy<->problem transitions
only, so it is silent in normal operation.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Julien LutranandClaude Fable 5