Gradient2Options
Defined in: src/adaptive-concurrency.ts:77
Properties
Section titled “Properties”backoffRatio?
Section titled “backoffRatio?”
optionalbackoffRatio?:number
Defined in: src/adaptive-concurrency.ts:89
Multiplier applied on a drop. Default 0.9.
longWindow?
Section titled “longWindow?”
optionallongWindow?:number
Defined in: src/adaptive-concurrency.ts:83
Samples averaged by the long-term latency EMA. Default 600.
queueSize?
Section titled “queueSize?”
optionalqueueSize?: (limit) =>number
Defined in: src/adaptive-concurrency.ts:87
Headroom added on top of limit * gradient, so the limit can grow. Default sqrt(limit).
Parameters
Section titled “Parameters”number
Returns
Section titled “Returns”number
shortWindow?
Section titled “shortWindow?”
optionalshortWindow?:number
Defined in: src/adaptive-concurrency.ts:81
Samples averaged by the short-term latency EMA. Default 10.
smoothing?
Section titled “smoothing?”
optionalsmoothing?:number
Defined in: src/adaptive-concurrency.ts:85
Weight of each new limit estimate (0..1]; lower is steadier. Default 0.2.
tolerance?
Section titled “tolerance?”
optionaltolerance?:number
Defined in: src/adaptive-concurrency.ts:79
How much the short-term latency may exceed the long-term one before the limit shrinks. Default 1.5.