Octopus
1.x
Open

Reading1 min
Updated31 Jul 2026
Sourcev1/state/index.mdx

State

Several gateway features need to remember things between requests — and, when you run more than one replica, between instances. Octopus stores that shared runtime state in a pluggable state store so features behave consistently whether you run one process or many.

What uses the state store01

  • Rate limiting — request counters and windows, so limits are enforced across all replicas.

  • Caching — cached upstream responses.

  • Sessions & tokens — authentication caches and session data.

  • Circuit breakers — shared failure counts and breaker state.

  • Coordination — locks and other coordination primitives.

Backends02

BackendScopeUse case
In-memoryPer processLocal development and single-instance deployments.
RedisSharedMulti-replica deployments needing fast shared counters and caches.
PostgresShared, durableShared state that must survive restarts.
HybridTieredA fast local tier backed by a shared tier.