damondouglas commented on code in PR #28952: URL: https://github.com/apache/beam/pull/28952#discussion_r1361246057
########## sdks/java/io/rrio/src/main/java/org/apache/beam/io/requestresponseio/CallShouldBackoffBasedOnRejectionProbability.java: ########## @@ -0,0 +1,102 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package org.apache.beam.io.requestresponseio; + +import org.checkerframework.checker.nullness.qual.Nullable; + +/** Reports whether to apply backoff based on https://sre.google/sre-book/handling-overload/. */ +class CallShouldBackoffBasedOnRejectionProbability<ResponseT> + implements CallShouldBackoff<ResponseT> { + + // Default multiplier value recommended by https://sre.google/sre-book/handling-overload/ + private static final double DEFAULT_MULTIPLIER = 2.0; + + // The threshold is the value that pReject() must exceed in order to report a value() of true. If + // null, then the computation relies on a random value. + private @Nullable Double threshold; + + // The multiplier drives the impact of accepts on the rejection probability. See <a + // https://sre.google/sre-book/handling-overload/ for details. + private final double multiplier; + + // The number of total requests called to a remote API. + private double requests = 0; + + // The number of total accepts called to a remote API. + private double accepts = 0; + + /** Instantiate class with the {@link #DEFAULT_MULTIPLIER}. */ + CallShouldBackoffBasedOnRejectionProbability() { + this(DEFAULT_MULTIPLIER); + } + + /** + * Instantiates class with the provided multiplier. The multiplier drives the impact of accepts on + * the rejection probability. See https://sre.google/sre-book/handling-overload/ for details. + */ + CallShouldBackoffBasedOnRejectionProbability(double multiplier) { + this.multiplier = multiplier; + } + + /** + * Setter for the threshold that overrides usage of a random value. The threshold is the value + * that pReject() must exceed in order to report a value() of true. + */ + CallShouldBackoffBasedOnRejectionProbability<ResponseT> setThreshold(double threshold) { + this.threshold = threshold; + return this; + } + + /** Update the state of whether to backoff using information about the exception. */ + @Override + public void update(UserCodeExecutionException exception) { + this.requests++; + } + + /** Update the state of whether to backoff using information about the response. */ + @Override + public void update(ResponseT response) { + this.requests++; + this.accepts++; + } + + /** Provide a threshold to evaluate backoff. */ + double threshold() { + if (this.threshold != null) { + return this.threshold; + } + return Math.random(); + } + + /** + * Compute the probability of API call rejection based on + * https://sre.google/sre-book/handling-overload/. + */ + double pReject() { + double numerator = requests - multiplier * accepts; + double denominator = requests + 1; Review Comment: I'd rather keep it as is. It looks more readable in my opinion. Also, the threshold method helpful to keep separate as it has the conditional whether someone supplied a hard code for testing or whether it is using the random. The code is simple enough but I wanted to keep the use of Math.random out of this core method. I will go ahead and resolve this but please re-open if you feel strongly otherwise. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
