Struct RateLimitConfig 

pub struct RateLimitConfig {
    pub qps: u32,
    pub burst: u32,
    pub per_tenant_default: PerTenantRateLimitConfig,
}

Static configuration for the rate limiter.

Two layers, both enforced (whichever is tighter wins):

1. Per-tenant bucket keyed on (TokenId, TenantId) (per_tenant_default) — primary defence. Prevents one tenant from saturating a shared token’s quota.
2. Per-token bucket keyed on TokenId (qps, burst) — backstop. Caps aggregate usage by a single token across all tenants.

Token-level knobs come from TENSOR_WASM_API_RATE_LIMIT_QPS and TENSOR_WASM_API_RATE_LIMIT_BURST at server startup; per-tenant defaults to PerTenantRateLimitConfig::default. If both knobs are zero (or unset) the token-level backstop is disabled, but the per-tenant layer is still in force unless explicitly cleared — see RateLimitConfig::is_disabled.

Note: this type is no longer Eq because PerTenantRateLimitConfig::qps is f64. Use PartialEq for comparisons in tests.

Fields

qps: u32

Steady-state requests-per-second admitted per token (backstop layer).

burst: u32

Maximum burst — the per-token bucket capacity, in permits.

per_tenant_default: PerTenantRateLimitConfig

Default per-tenant configuration applied to every (token, tenant) pair. The primary defence against a single tenant exhausting a shared token’s quota.

Implementations

impl RateLimitConfig

pub const ENV_QPS: &'static str = "TENSOR_WASM_API_RATE_LIMIT_QPS"

Environment variable carrying the steady-state QPS allowance per token.

pub const ENV_BURST: &'static str = "TENSOR_WASM_API_RATE_LIMIT_BURST"

Environment variable carrying the burst (bucket capacity) per token.

pub const DEFAULT_QPS: u32 = 100

Default QPS applied when ENV_QPS is unset but ENV_BURST is set.

pub const DEFAULT_BURST: u32 = 200

Default burst applied when ENV_BURST is unset but ENV_QPS is set.

pub const fn disabled() -> Self

Disabled config: every layer off. The middleware is a pass-through.

pub const fn is_disabled(&self) -> bool

true if every layer of the limiter is disabled and the middleware would unconditionally admit. Used by rate_limit to short-circuit the bucket lookup entirely.

pub const fn is_token_layer_disabled(&self) -> bool

true if the per-token (backstop) layer is disabled.

pub fn from_env() -> Self

Load from the process environment.

• Both vars unset / either "0" / either unparseable => token-layer disabled. The per-tenant layer still defaults to PerTenantRateLimitConfig::default.
• Otherwise: missing-but-other-side-set falls back to DEFAULT_QPS / DEFAULT_BURST.

Trait Implementations

impl Clone for RateLimitConfig

fn clone(&self) -> RateLimitConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for RateLimitConfig

impl Debug for RateLimitConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for RateLimitConfig

fn default() -> Self

Default is disabled. Operators opt in by setting both env vars (or by constructing a config explicitly).

impl PartialEq for RateLimitConfig

fn eq(&self, other: &RateLimitConfig) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for RateLimitConfig