diehard-async.core — org.clojars.vladislav/diehard-async 0.1.13

*elapsed-time-ms* clj

Available in retry block. Contextual value represents time elasped since first attempt

*executions* clj

Available in retry block. Contextual value represents execution times

*start-time-ms* clj

Available in retry block. Contextual value represents first attempt time

defbulkhead clj macro

(defbulkhead name opts)

Create bulkhead config from option map.

• concurrency the max number of concurrent executions

defcircuitbreaker clj macro

(defcircuitbreaker name opts)

Define a circuit breaker with option.

Available options

There options are available when creating circuit breaker in defcircuitbreaker.

Failure criteria

All the three fail options share same meaning with similar option in retry block.

• :fail-if
• :fail-on
• :fail-when
• :timeout-ms while give all you code a timeout is best practice in application level, circuit breaker also provides a timeout for marking a long running block as failure

Delay and threshold

• :delay-ms required. the delay for :open circuit breaker to turn into :half-open.
• :failure-threshold failures sets the number of failures that must occur before the circuit opens (count-based).
• :failure-threshold-ratio [failures executions] sets the failure ratio that must occur before the circuit opens (count-based). e.g. [4 10] means 4 out of the last 10 executions must fail to open the circuit.
• :failure-threshold-ratio-in-period [failures min-executions period-ms] sets the number of failures that must occur within a time period to open the circuit (time-based). Must also exceed min-executions within the time period to open the circuit.
• :failure-rate-threshold-in-period [failure-pct min-executions period-ms] sets the percent of executions that must fail (1-100) within the time period to open the circuit (time-based). Must also exceed min-executions within the time period to open the circuit.
• :success-threshold successes sets the number of successful executions that must occur when in :half-open to return to a :closed state.
• :success-threshold-ratio [successes executions] sets the success ratio that must occur when in :half-open to return to a :closed state. e.g. [6 10] means 6 out of the last 10 executions must succeed to close the circuit.

Listeners

• :on-open a function to be called when state goes :open
• :on-close a function to be called when state goes :closed
• :on-half-open a function to be called when state goes :half-open

defratelimiter clj macro

(defratelimiter name opts)

Create a rate limiter with options.

• :rate execution permits per second.
• :max-cached-tokens the max size of permits we can cache when idle

defretrypolicy clj macro

(defretrypolicy name opts)

Predefined retry policy.

Available options

Retry criteria

• :retry-when retry when return value is given value
• :retry-on retry on given exception / exceptions(vector) were thrown
• :retry-if specify a function (fn [return-value exception-thrown]), retry if the function returns truthy

Retry abortion criteria

• :abort-when abort retry when return value is given value
• :abort-on abort retry on given exception / exceptions(vector) were thrown
• :abort-if specify a function (fn [return-value exception-thrown]), abort retry if the function returns truthy
• :max-retries abort retry when max attempts reached
• :max-duration-ms abort retry when duration reached

Delay

• :backoff-ms specify a vector [initial-delay-ms max-delay-ms multiplier] to control the delay between each retry, the delay for nth retry will be (min (* initial-delay-ms (expt 2 (- n 1))) max-delay-ms)
• :delay-ms use constant delay between each retry
• :jitter-factor random factor for each delay
• :jitter-ms random time (-jitter-ms, jitter-ms) adds to each delay

Retry Listeners

• :on-abort accepts a function which takes result, exception as arguments, called when retry aborted
• :on-complete accepts a function which takes result, exception as arguments, called when exiting retry block
• :on-failed-attempt accepts a function which takes result, exception as arguments, called when execution failed (matches retry criteria)
• :on-failure accepts a function which takes result, exception as arguments, called when exiting retry block with failure (matches retry criteria)
• :on-success accepts a function which takes result as arguments, called when exiting retry block with success (mismatches retry criteria)
• :on-retry accepts a function which takes result, exception as arguments, called when a retry attempted.
• :on-retries-exceeded accepts a function which takes result, exception as arguments, called when max retries or max duration have been exceeded.

Use pre-defined policy

You can put together all those retry policies in a defretrypolicy. And use :policy option in option map.

(diehard/defretrypolicy policy
  {:max-retries 5
   :backoff-ms [1000 10000]})

(diehard/with-retry {:policy policy}
  ;; your code here
  )

with-bulkhead clj macro

(with-bulkhead opts & body)

Bulkhead block. Only given number of executions is allowed to be executed in parallel.

;; create a bulkhead that limit concurrency to 3
(defbulkhead mybh {:concurrency 3})

(with-bulkhead {:bulkhead mybh}
  ;; your task here
  )

By default it will wait until there is permits available for execution.

You can add max-wait-ms option for change this behavior. If no permits is available when max-wait-ms exceeded, an ex-info will be thrown with ex-data as {:bulkhead true :max-wait-ms wait-timeout}

(try
  (with-bulkhead {:bulkhead mybh
                  :max-wait-ms 1000}
    ;; your task here
    )
  (catch Exception e
    (is (::bulkhead (ex-data e)))))

with-circuit-breaker clj macro

(with-circuit-breaker cb & body)

Circuit breaker protected block.

(require '[diehard-async.core :as dha])

(dha/defcircuitbreaker test-cb {:failure-threshold-ratio [35 50]
                                    :delay-ms 1000})

(dha/with-circuit-breaker test-cb
  ;; your protected code here
  )

In this scenario, if the circuit breaker protected code block fails 35 times in 50 executions, as defined in :failure-threshold-ratio, the test-cb is entering into :open state. When circuit breaker is open, all execution requests will be rejected immediately.

After :delay-ms, the circuit breaker will be :half-open. At the moment, 50 execution will be allowed, to test the state to see if it's recovered. If success, the circuit breaker is back to :closed state. Otherwise, it will be :open again.

The block will throw CircuitBreakerOpenException when the circuit breaker is open and skip execution of inner forms. Otherwise it will return the value or throw the exception raised from inner.

You can always check circuit breaker state with dha.circuitbreaker/state.

with-rate-limiter clj macro

(with-rate-limiter opts & body)

Rate Limiter protected block. Code execution in this block is throttled to given rate. Use defratelimiter to define a ratelimiter and use it as option:

;; create a rate limiter for 100 executions for second
(defratelimiter myfl {:rate 100})

(with-rate-limiter {:ratelimiter myfl}
  ;; your task here
  )

By default it will wait forever until there is permits available. You can also specify a max-wait-ms to wait for a given time. If there's no permits in this period, this block will throw a Clojure ex-info, with ex-data as

(try
  (with-rate-limiter {:ratelimiter myfl
                      :max-wait-ms 1000}
    ;; your task here
    )
  (catch Exception e
    (is (:throttled (ex-data e)))))

If your execution has a greater graininess, you can customize the permits for this execution by setting :permits option.

(with-rate-limiter {:ratelimiter myfl
                    :permits (size-of-the-task)}
  ;; your task here
  )

with-retry clj macro

(with-retry opt & body)

Retry policy protected block. If the return value of or exception thrown from the code block matches the criteria of your retry policy, the code block will be executed again, until it mismatch the retry policy or matches the abort criteria. The block will return the value or throw exception from the last execution. If :circuit-breaker is set, it will throw CircuitBreakerOpenException when the breaker becomes open.

Available options

Retry criteria

• :retry-when retry when return value is given value
• :retry-on retry on given exception / exceptions(vector) were thrown
• :retry-if specify a function (fn [return-value exception-thrown]), retry if the function returns truthy

Retry abortion criteria

• :abort-when abort retry when return value is given value
• :abort-on abort retry on given exception / exceptions(vector) were thrown
• :abort-if specify a function (fn [return-value exception-thrown]), abort retry if the function returns truthy
• :max-retries abort retry when max attempts reached
• :max-duration-ms abort retry when duration reached

Delay

• :backoff-ms specify a vector [initial-delay-ms max-delay-ms multiplier] to control the delay between each retry, the delay for nth retry will be (min (* initial-delay-ms (expt 2 (- n 1))) max-delay-ms)
• :delay-ms use constant delay between each retry
• :jitter-factor random factor for each delay
• :jitter-ms random time (-jitter-ms, jitter-ms) adds to each delay

Retry Listeners

• :on-abort accepts a function which takes result, exception as arguments, called when retry aborted
• :on-failed-attempt accepts a function which takes result, exception as arguments, called when execution failed (matches retry criteria)
• :on-failure accepts a function which takes result, exception as arguments, called when existing retry block with failure (matches retry criteria)
• :on-success accepts a function which takes result as arguments, called when existing retry block with success (mismatches retry criteria)
• :on-retry accepts a function which takes result and exception as arguments, called when a retry attempted.
• :on-retries-exceeded accepts a function which takes result and exception as arguments, called when retries exceeded

Use pre-defined policy

You can put together all those retry policies in a defretrypolicy. And use :policy option in option map.

(diehard/defretrypolicy policy
  {:max-retries 5
   :backoff-ms [1000 10000]})

(diehard/with-retry {:policy policy}
  ;; your code here
  )

Retry Listeners

• :on-abort accepts a function which takes result, exception as arguments, called when retry aborted
• :on-complete accepts a function which takes result, exception as arguments, called when exiting retry block
• :on-failed-attempt accepts a function which takes result, exception as arguments, called when execution failed (matches retry criteria)
• :on-failure accepts a function which takes result, exception as arguments, called when exiting retry block with failure (matches retry criteria)
• :on-success accepts a function which takes result as arguments, called when exiting retry block with success (mismatches retry criteria)
• :on-retry accepts a function which takes result as arguments, called when a retry attempted.

Fallback

• :fallback fallback value or handler function when retry blocks exists with failure.

;; return 5 when attempts failure
(with-retry {:fallback 5}
  ;; ...
  )

;; return fallback handler function result when failed
(with-retry {:fallback (fn [value exception]
                         ;; value: value returned from last attempt
                         ;; exp: exception thrown from last attempt
                         )}
  ;; ...
  )

Circuit breaker

• :circuit-breaker a circuit breaker created from defcircuitbreaker. It will work together with retry policy as quit criteria.

with-timeout clj macro

(with-timeout opts & body)

Timeout block. This block will throw TimeoutExceededException when configured time elapsed.

Available options:

• timeout-ms: required timeout
• on-success: the callback when block execution succeeded
• on-failure: the failure when block execution timed out
• interrupt?: cancel the execution if it times out. boolean, default false