Skip to content
All concepts

Concept

Thundering herd

A thundering herd is a large number of clients or requests all doing the same thing at the same moment. Typical triggers are a cache entry expiring, a server coming back, or a shared timer firing. Each request is reasonable alone, but together they overload whatever they all arrive at.

Watch two gateways restart close together: every dropped client reconnects at once to the one that is left. A live step from the lesson Reconnect storms after a deploy (Real-time & chat). Timings and counts are illustrative.

Where you'll see it

Sources: Kubernetes documentation: Pod lifecycle, termination of Pods ↗ · Discord: Behind the scenes of the 3/25/26 voice outage ↗ · Report a problem

The definition is in our own words; the sources are those of the lesson Reconnect storms after a deploy.