Kubernetes Operators · Rate Limits
Kubernetes Operators Rate Limits
Operators interact with the Kubernetes API server as authenticated clients. There is no vendor-imposed rate limit on the pattern itself; instead, rate limiting is enforced by the Kubernetes API server (per-user/per-priority-level) via API Priority and Fairness (APF) and by client-side QPS/burst settings configured in client-go (rest.Config).
2 Limits
Throttle: 429
Rate LimitingKubernetesOperatorsOpen Source
Limits
API Priority and Fairness (server-side) priority-level
cluster-configured PriorityLevelConfiguration concurrency shares
Client-side QPS / Burst (client-go) client
rest.Config.QPS (default 5) / Burst (default 10) — operators typically tune higher
Policies
APF flow control
The Kubernetes API server applies API Priority and Fairness; operators should set a sensible UserAgent and FlowSchema to avoid starvation under load.
Client-side throttling
client-go enforces local QPS/Burst before sending requests; operator authors raise these defaults to suit reconciliation workloads.
Backoff and resync
Use the controller-runtime / informer cache and rate-limited workqueues; do not poll the API server in tight loops — rely on watches.
Honor 429 and Retry-After
When the API server returns 429 (typically with a Retry-After), back off and respect the suggested delay rather than retrying immediately.