Ethanlm commented on a change in pull request #3244: [STORM-3600] Add caching in Cluster.calculateSharedOffHeapNodeMemory URL: https://github.com/apache/storm/pull/3244#discussion_r406455496
########## File path: storm-server/src/test/java/org/apache/storm/scheduler/resource/strategies/scheduling/TestLargeCluster.java ########## @@ -0,0 +1,458 @@ +/** + * 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 + * <p> + * http://www.apache.org/licenses/LICENSE-2.0 + * <p> + * 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.storm.scheduler.resource.strategies.scheduling; + +import org.apache.logging.log4j.Level; +import org.apache.logging.log4j.core.config.Configurator; +import org.apache.storm.Config; +import org.apache.storm.DaemonConfig; +import org.apache.storm.generated.StormTopology; +import org.apache.storm.metric.StormMetricsRegistry; +import org.apache.storm.scheduler.Cluster; +import org.apache.storm.scheduler.ExecutorDetails; +import org.apache.storm.scheduler.INimbus; +import org.apache.storm.scheduler.IScheduler; +import org.apache.storm.scheduler.SupervisorDetails; +import org.apache.storm.scheduler.Topologies; +import org.apache.storm.scheduler.TopologyDetails; +import org.apache.storm.scheduler.WorkerSlot; +import org.apache.storm.scheduler.resource.ResourceAwareScheduler; +import org.apache.storm.scheduler.resource.TestUtilsForResourceAwareScheduler; +import org.apache.storm.scheduler.resource.normalization.NormalizedResources; +import org.apache.storm.scheduler.resource.normalization.NormalizedResourcesExtension; +import org.apache.storm.scheduler.resource.normalization.ResourceMetrics; +import org.apache.storm.utils.ObjectReader; +import org.apache.storm.utils.Time; +import org.apache.storm.utils.Utils; +import org.junit.Assert; +import org.junit.jupiter.api.AfterEach; +import org.junit.jupiter.api.Test; +import org.junit.jupiter.api.extension.ExtendWith; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import java.io.BufferedReader; +import java.io.ByteArrayOutputStream; +import java.io.IOException; +import java.io.InputStream; +import java.io.InputStreamReader; +import java.util.ArrayList; +import java.util.Arrays; +import java.util.Collection; +import java.util.Collections; +import java.util.HashMap; +import java.util.HashSet; +import java.util.LinkedList; +import java.util.List; +import java.util.Map; +import java.util.Set; +import java.util.TreeMap; + +@ExtendWith({NormalizedResourcesExtension.class}) +public class TestLargeCluster { + private static final Logger LOG = LoggerFactory.getLogger(TestLargeCluster.class); + + public static final String TEST_CLUSTER_NAME = "largeCluster01"; + public static final String TEST_RESOURCE_PATH = "clusterconf/" + TEST_CLUSTER_NAME; + + private static IScheduler scheduler = null; + + @AfterEach + public void cleanup() { + if (scheduler != null) { + scheduler.cleanup(); + scheduler = null; + } + } + + /** + * Get the list of serialized topology (*code.ser) and configuration (*conf.ser) + * resource files in the path. The resources are sorted so that paired topology and conf + * files are sequential. Unpaired files may be ignored by the caller. + * + * @param path directory in which resources exist. + * @return + * @throws IOException + */ + public static List<String> getResourceFiles(String path) throws IOException { + List<String> fileNames = new ArrayList<>(); + + try ( + InputStream in = getResourceAsStream(path); + BufferedReader br = new BufferedReader(new InputStreamReader(in)) + ) { + String resource; + + while ((resource = br.readLine()) != null) { + if (resource.endsWith("code.ser") || resource.endsWith("conf.ser")) { + fileNames.add(path + "/" + resource); + } + } + Collections.sort(fileNames); + } + return fileNames; + } + + /** + * InputStream to read the fully qualified resource path. + * + * @param resource + * @return + */ + public static InputStream getResourceAsStream(String resource) { + final InputStream in = getContextClassLoader().getResourceAsStream(resource); + return in == null ? ClassLoader.getSystemClassLoader().getResourceAsStream(resource) : in; + } + + /** + * Read the contents of the fully qualified resource path. + * + * @param resource + * @return + * @throws Exception + */ + public static byte[] getResourceAsBytes(String resource) throws Exception { + InputStream in = getResourceAsStream(resource); + if (in == null) { + return null; + } + try (ByteArrayOutputStream out = new ByteArrayOutputStream()) { + while (in.available() > 0) { + out.write(in.read()); + } + return out.toByteArray(); + } + } + + public static ClassLoader getContextClassLoader() { + return Thread.currentThread().getContextClassLoader(); + } + + /** + * Create an array of TopologyDetails by reading serialized files for topology and configuration in the + * resource path. + * + * @param failOnParseError throw exception if there are unmatched files, otherwise ignore unmatched and read errors. + * @return An array of TopologyDetails representing resource files. + * @throws Exception + */ + public static TopologyDetails[] createTopoDetailsArray(boolean failOnParseError) throws Exception { + List<TopologyDetails> topoDetailsList = new ArrayList<>(); + List<String> errors = new ArrayList<>(); + List<String> resources = getResourceFiles(TEST_RESOURCE_PATH); + Map<String, String> codeResourceMap = new TreeMap<>(); + Map<String, String> confResourceMap = new HashMap<>(); + for (int i = 0 ; i < resources.size() ; i++) { + String resource = resources.get(i); + int idxOfSlash = resource.lastIndexOf("/"); + int idxOfDash = resource.lastIndexOf("-"); + String nm = idxOfDash > idxOfSlash ? resource.substring(idxOfSlash + 1, idxOfDash) : resource.substring(idxOfSlash + 1, resource.length() - 7); + if (resource.endsWith("code.ser")) { + codeResourceMap.put(nm, resource); + } else if (resource.endsWith("conf.ser")) { + confResourceMap.put(nm, resource); + } else { + LOG.info("Ignoring unsupported resource file " + resource); + } + } + String[] examinedConfParams = { + Config.TOPOLOGY_NAME, + Config.TOPOLOGY_SCHEDULER_STRATEGY, + Config.TOPOLOGY_PRIORITY, + Config.TOPOLOGY_WORKERS, + Config.TOPOLOGY_WORKER_MAX_HEAP_SIZE_MB, + Config.TOPOLOGY_SUBMITTER_USER, + Config.TOPOLOGY_ACKER_CPU_PCORE_PERCENT, + Config.TOPOLOGY_ACKER_RESOURCES_OFFHEAP_MEMORY_MB, + Config.TOPOLOGY_ACKER_RESOURCES_ONHEAP_MEMORY_MB, + }; + + for (String nm : codeResourceMap.keySet()) { + String codeResource = codeResourceMap.get(nm); + if (!confResourceMap.containsKey(nm)) { + String err = String.format("Ignoring topology file %s because of missing config file for %s", codeResource, nm); + errors.add(err); + LOG.error(err); + continue; + } + String confResource = confResourceMap.get(nm); + LOG.info("Found matching topology and config files: {}, {}", codeResource, confResource); + StormTopology stormTopology; + try { + stormTopology = Utils.deserialize(getResourceAsBytes(codeResource), StormTopology.class); + } catch (Exception ex) { + String err = String.format("Cannot read topology from resource %s", codeResource); + errors.add(err); + LOG.error(err, ex); + continue; + } + + Map<String, Object> conf; + try { + conf = Utils.fromCompressedJsonConf(getResourceAsBytes(confResource)); + } catch (RuntimeException | IOException ex) { + String err = String.format("Cannot read configuration from resource %s", confResource); + errors.add(err); + LOG.error(err, ex); + continue; + } + // fix 0.10 conf class names + String[] configParamsToFix = {Config.TOPOLOGY_SCHEDULER_STRATEGY, Config.STORM_NETWORK_TOPOGRAPHY_PLUGIN, + DaemonConfig.RESOURCE_AWARE_SCHEDULER_PRIORITY_STRATEGY }; + for (String configParam: configParamsToFix) { + if (!conf.containsKey(configParam)) { + continue; + } + String className = (String) conf.get(configParam); + if (className.startsWith("backtype")) { + className = className.replace("backtype", "org.apache"); + conf.put(configParam, className); + } + } + // fix conf params used by ConstraintSolverStrategy + if (!conf.containsKey(DaemonConfig.RESOURCE_AWARE_SCHEDULER_MAX_STATE_SEARCH)) { + conf.put(DaemonConfig.RESOURCE_AWARE_SCHEDULER_MAX_STATE_SEARCH, 10_000); + } + if (!conf.containsKey(Config.TOPOLOGY_RAS_CONSTRAINT_MAX_STATE_SEARCH)) { + conf.put(Config.TOPOLOGY_RAS_CONSTRAINT_MAX_STATE_SEARCH, 10_000); + } + + String topoId = nm; + String topoName = (String) conf.getOrDefault(Config.TOPOLOGY_NAME, nm); + + // conf + StringBuffer sb = new StringBuffer("Config for " + nm + ": "); + for (String param : examinedConfParams) { + Object val = conf.getOrDefault(param, "<null>"); + sb.append(param).append("=").append(val).append(", "); + } + LOG.info(sb.toString()); + + // topo + Map<ExecutorDetails, String> execToComp = TestUtilsForResourceAwareScheduler.genExecsAndComps(stormTopology); + LOG.info("Topology \"{}\" spouts={}, bolts={}, execToComp size is {}", topoName, + stormTopology.get_spouts_size(), stormTopology.get_bolts_size(), execToComp.size()); + int numWorkers = Integer.parseInt("" + conf.getOrDefault(Config.TOPOLOGY_WORKERS, "0")); + TopologyDetails topo = new TopologyDetails(topoId, conf, stormTopology, numWorkers, + execToComp, Time.currentTimeSecs(), "user"); + topo.getComponents(); // sanity check - normally this should not fail + + topoDetailsList.add(topo); + } + if (!errors.isEmpty() && failOnParseError) { + throw new Exception("Unable to parse all serialized objects\n\t" + String.join("\n\t", errors)); + } + return topoDetailsList.toArray(new TopologyDetails[0]); + } + + /** + * Check if the files in the resource directory are matched, can be read properly, and code/config files occur + * in matched pairs. + * + * @throws Exception showing bad and unmatched resource files. + */ + @Test + public void testReadSerializedTopologiesAndConfigs() throws Exception { + List<String> resources = getResourceFiles(TEST_RESOURCE_PATH); + Assert.assertTrue("No resource files found in " + TEST_RESOURCE_PATH, !resources.isEmpty()); + TopologyDetails[] topoDetailsArray = createTopoDetailsArray(true); + } + + /** + * Create one supervisor and add to the supervisors list. + * + * @param rack rack-number + * @param superInRack supervisor number in the rack + * @param cpu percentage + * @param mem in megabytes + * @param numPorts number of ports on this supervisor + * @param sups returned map os supervisors + */ + private static void createAndAddOneSupervisor( + int rack, int superInRack, double cpu, double mem, int numPorts, + Map<String, SupervisorDetails> sups) { + + List<Number> ports = new LinkedList<>(); + for (int p = 0; p < numPorts; p++) { + ports.add(p); + } + String superId = String.format("r%03ds%03d", rack, superInRack); + String hostId = String.format("host-%03d-rack-%03d", superInRack, rack); + Map<String, Double> resourceMap = new HashMap<>(); + resourceMap.put(Config.SUPERVISOR_CPU_CAPACITY, cpu); + resourceMap.put(Config.SUPERVISOR_MEMORY_CAPACITY_MB, mem); + resourceMap.put("network.resource.units", 50.0); + SupervisorDetails sup = new SupervisorDetails(superId, + hostId, null, ports, + NormalizedResources.RESOURCE_NAME_NORMALIZER.normalizedResourceMap(resourceMap)); + sups.put(sup.getId(), sup); + } + + /** + * Create supervisors. + * + * @param uniformSupervisors true if all supervisors are of the same size, false otherwise. + * @return supervisor details indexed by id + */ + private static Map<String, SupervisorDetails> createSupervisors(boolean uniformSupervisors) { + Map<String, SupervisorDetails> retVal; + if (uniformSupervisors) { + int numRacks = 16; + int numSupersPerRack = 82; + int numPorts = 50; // scheduling will fill up 2-6, leaving of 90% workerslots unused - does this cause slow scheduling? + int rackStart = 0; + int superInRackStart = 1; + double cpu = 7200; // %percent + double mem = 356_000; // MB + Map<String, Double> miscResources = new HashMap<>(); + miscResources.put("network.resource.units", 100.0); + + return TestUtilsForResourceAwareScheduler.genSupervisorsWithRacks( + numRacks, numSupersPerRack, numPorts, rackStart, superInRackStart, cpu, mem, miscResources); + + } else { + // this non-uniform supervisor distribution closelt (but not exactly) mimics IridiumBlue cluster Review comment: `IridiumBlue` doesn't mean anything in the community code. Maybe reword it. Typo: `closelt` ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
