Cribl · Rate Limits
Cribl Rate Limits
Cribl's APIs (Cloud, Stream, Edge, Search, Lake, As Code) authenticate via OAuth 2.0 client credentials against the Cribl Cloud control plane. Cribl does not publish per-API-key request rate limits on its public documentation; effective throughput is bounded primarily by the platform-tier worker capacity (worker groups, worker processes, search executors) and daily TB/day data-processing ceiling. Hard request-rate throttles, if any, are enforced at the gateway and surfaced via standard HTTP 429 / 503 responses; exact numeric limits are not published.
5 Limits
Throttle: 429
ConfigurationData LakeData PipelinesData RoutingEdge ComputingObservabilitySearchStream ProcessingTelemetryRate LimitingQuotasThrottling
Limits
Daily data processed (Free) account
1
Free tier ceiling for processed data volume. Enforced as a quota, not as a per-second throttle.
Daily data processed (Standard) account
5
Standard tier ceiling.
Worker processes (Free) account
10
Worker process ceiling that bounds concurrent pipeline throughput on the Free tier.
Worker processes (Standard) account
50
API request rate api-key
not publicly documented
Cribl Cloud API enforces gateway-level throttling but does not publish per-second numbers. Honor Retry-After when 429 is returned.
Policies
OAuth 2.0 client credentials
All Cribl Cloud API calls authenticate with OAuth 2.0 client credentials; tokens are short-lived, so clients must cache and refresh access tokens rather than re-authenticating per request.
Backoff Strategy
On 429 or 5xx responses, implement exponential backoff with jitter and respect Retry-After headers when present.
Capacity, not RPS
Cribl's effective consumption ceiling is the tier-bound daily TB/day and worker capacity rather than a fixed requests-per-second budget. Scale by upgrading the tier or adding worker groups / processes / executors.
Edge fleet scoping
For Cribl Edge, each edge node enforces its own local processing budget; control-plane API rate limits apply only to the Cloud / Stream / Edge management endpoints, not to the data path on the edge agents.