Acquire attempts to acquire some number of tokens from the semaphore on behalf of the given priority. If Acquire returns nil, these tokens should be r

Wait blocks until the given number of tokens is available for the given priority. It returns nil if the tokens were successfully acquired or ErrReject

setMode appropriately sets the codelMode. Called internally.

AcceptedError wraps the given err as "accepted" by the backend for the purposes of AdaptiveThrottle. This should be done for regular protocol errors t

WithAdaptiveThrottle is used to send a request to a backend using the given AdaptiveThrottle for client-rejections. If f returns an error, at conside

true if there are no waiters in the codel.

NewAdaptiveThrottle returns an AdaptiveThrottle. priorities is the number of priorities that the throttle will accept. Giving a priority outside of `

NewRateLimiter returns a rate limiter with the given number of priorities, allowing the given aggregate rate, and up to the given burst. That is, the

NewSemaphore returns a semaphore with the given number of priorities, and will allow at most capacity concurrency. The other options do not frequentl

Release returns the given number of tokens to the semaphore. It should only be called if these tokens are known to be acquired from the semaphore with

returns the next item that would be removed by pop, if there is one. calling reap, setMode, or push invalidates this value.

pop pops the next waiter from the codel. Returns zero, false if the next waiter has already been abandoned or if the queue is empty. popping the next

push adds a waiter to the codel.

reap removes and wakes-unsuccessfully waiters that have timed out based on the current codelMode. codel assumes this will be called at least once ever

wait blocks until the waiter is woken up by the codel or ctx expires. returns ErrRejected if timed out by the codel.

Accepted records that a request was accepted by the backend.

AdaptiveThrottleWindow sets the time window over which the throttle remembers requests for use in figuring out the success rate.

Attempt returns true if the request should be attempted, and false if it should be rejected.

Close frees background resources used by the rate limiter.

Close frees background resources used by the semaphore.

The time it takes for 100% debt to be completely forgiven. Debt decays linearly over time since load patterns change and a previously learned debt amo

The proportion of debt that is forgiven for lower priorities whenever a higher-priority request succeeds, in [0, 1].

The long timeout for the internal CoDels. See the README for more on CoDel.

The short timeout for the internal CoDels. See the README for more on CoDel.

Rejected records that a request was rejected by the backend.

The time it takes for 100% debt to be completely forgiven. Debt decays linearly over time since load patterns change and a previously learned debt amo

The proportion of debt that is forgiven for lower priorities whenever a higher-priority request succeeds, in [0, 1].

The long timeout for the internal CoDels. See the README for more on CoDel.

The short timeout for the internal CoDels. See the README for more on CoDel.

admit attempts to successfully-wake the waiter, returns true if the waiter will return nil from wait.

drop unsuccessfully-wakes the waiter, it will return ErrRejected from wait.

AdaptiveThrottleMinimumRate sets the minimum number of requests per second that the adaptive throttle will allow (approximately) through to the upstre

AdaptiveThrottleRatio sets the ratio of the measured success rate and the rate that the throttle will admit. For example, when k is 2 the throttle wil

SetCapacity sets the maximum number of outstanding tokens for the semaphore. If more tokens than this new value are already outstanding the semaphore