Skip to main content

DisasterOperation

Failover flow

Editable source: failover-flow.excalidraw

DisasterOperation represents a long-running operation such as failover, reprotect, undo, cancel, sync, or drill execution.

Why Operations Are Resources​

Long workflows need observation, retries, compensation, and history. Modeling them as resources makes each step visible through status and Events.

Failover Steps​

  • PreCheck
  • PauseSchedules
  • FinalSync
  • ScaleDownSource
  • ScaleUpTarget
  • CheckReplicas
  • SwitchRoles

Failure Handling​

When a step fails, the operation records state, reason, message, and step history. Depending on the trigger point, the system may enter cancel or rollback-style compensation.