Heartbeat-based scheduling

A core distributed systems mechanism: use periodic signals (heartbeats) to monitor node health, enabling rapid failure detection and automated cluster rebalancing.

Heartbeat-based scheduling is a critical fault-tolerance strategy for mission-critical systems and distributed computing (e.g., in Spark or Hadoop). The mechanism operates by having each node transmit a regular, periodic signal—the 'heartbeat'—to a central monitor or other cluster members. If a node fails to send its heartbeat within a predefined interval (typically a few seconds), the system immediately flags it as failed or unavailable. This rapid, deterministic detection triggers the scheduling process: the cluster manager automatically reallocates the failed node’s workload and resources to redundant, available nodes, ensuring high availability and meeting stringent Service-Level Agreements (SLAs).

https://en.wikipedia.org/wiki/Heartbeat_(computing)