Testclusters: improove timeout handling (#43440)

This commit is contained in:
Alpar Torok 2019-07-01 11:29:54 +03:00
parent a58d231f4d
commit b4303290d9
6 changed files with 224 additions and 87 deletions

View File

@ -229,6 +229,7 @@ public class ElasticsearchCluster implements TestClusterConfiguration {
if (Version.fromString(node.getVersion()).getMajor() >= 7) {
node.defaultConfig.put("cluster.initial_master_nodes", "[" + nodeNames + "]");
node.defaultConfig.put("discovery.seed_providers", "file");
node.defaultConfig.put("discovery.seed_hosts", "[]");
}
}
node.start();
@ -286,14 +287,13 @@ public class ElasticsearchCluster implements TestClusterConfiguration {
}
public void waitForAllConditions() {
long startedAt = System.currentTimeMillis();
LOGGER.info("Waiting for nodes");
nodes.forEach(ElasticsearchNode::waitForAllConditions);
writeUnicastHostsFiles();
LOGGER.info("Starting to wait for cluster to form");
waitForConditions(waitConditions, startedAt, CLUSTER_UP_TIMEOUT, CLUSTER_UP_TIMEOUT_UNIT, this);
waitForConditions(waitConditions, System.currentTimeMillis(), CLUSTER_UP_TIMEOUT, CLUSTER_UP_TIMEOUT_UNIT, this);
}
@Override

View File

@ -37,6 +37,8 @@ import java.nio.charset.StandardCharsets;
import java.nio.file.Files;
import java.nio.file.Path;
import java.nio.file.StandardCopyOption;
import java.nio.file.StandardOpenOption;
import java.time.Instant;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collection;
@ -65,8 +67,10 @@ public class ElasticsearchNode implements TestClusterConfiguration {
private static final Logger LOGGER = Logging.getLogger(ElasticsearchNode.class);
private static final int ES_DESTROY_TIMEOUT = 20;
private static final TimeUnit ES_DESTROY_TIMEOUT_UNIT = TimeUnit.SECONDS;
private static final int NODE_UP_TIMEOUT = 60;
private static final TimeUnit NODE_UP_TIMEOUT_UNIT = TimeUnit.SECONDS;
private static final int NODE_UP_TIMEOUT = 2;
private static final TimeUnit NODE_UP_TIMEOUT_UNIT = TimeUnit.MINUTES;
private static final int ADDITIONAL_CONFIG_TIMEOUT = 15;
private static final TimeUnit ADDITIONAL_CONFIG_TIMEOUT_UNIT = TimeUnit.SECONDS;
private static final List<String> OVERRIDABLE_SETTINGS = Arrays.asList(
"path.repo",
"discovery.seed_providers"
@ -310,6 +314,7 @@ public class ElasticsearchNode implements TestClusterConfiguration {
try {
if (isWorkingDirConfigured == false) {
logToProcessStdout("Configuring working directory: " + workingDir);
// Only configure working dir once so we don't loose data on restarts
isWorkingDirConfigured = true;
createWorkingDir(distroArtifact);
@ -319,12 +324,16 @@ public class ElasticsearchNode implements TestClusterConfiguration {
}
createConfiguration();
if(plugins.isEmpty() == false) {
logToProcessStdout("Installing " + plugins.size() + " plugins");
plugins.forEach(plugin -> runElaticsearchBinScript(
"elasticsearch-plugin",
"install", "--batch", plugin.toString())
);
}
if (keystoreSettings.isEmpty() == false || keystoreFiles.isEmpty() == false) {
logToProcessStdout("Adding " + keystoreSettings.size() + " keystore settings and " + keystoreFiles.size() + " keystore files");
runElaticsearchBinScript("elasticsearch-keystore", "create");
checkSuppliers("Keystore", keystoreSettings.values());
@ -347,6 +356,7 @@ public class ElasticsearchNode implements TestClusterConfiguration {
copyExtraConfigFiles();
if (isSettingMissingOrTrue("xpack.security.enabled")) {
logToProcessStdout("Setting up " + credentials.size() + " users");
if (credentials.isEmpty()) {
user(Collections.emptyMap());
}
@ -358,9 +368,25 @@ public class ElasticsearchNode implements TestClusterConfiguration {
));
}
logToProcessStdout("Starting Elasticsearch process");
startElasticsearchProcess();
}
private void logToProcessStdout(String message) {
try {
if (Files.exists(esStdoutFile.getParent()) == false) {
Files.createDirectories(esStdoutFile.getParent());
}
Files.write(
esStdoutFile,
("[" + Instant.now().toString() + "] [BUILD] " + message + "\n").getBytes(StandardCharsets.UTF_8),
StandardOpenOption.CREATE, StandardOpenOption.APPEND
);
} catch (IOException e) {
throw new UncheckedIOException(e);
}
}
@Override
public void restart() {
LOGGER.info("Restarting {}", this);
@ -380,6 +406,9 @@ public class ElasticsearchNode implements TestClusterConfiguration {
}
private void copyExtraConfigFiles() {
if (extraConfigFiles.isEmpty() == false) {
logToProcessStdout("Setting up " + extraConfigFiles.size() + " additional config files");
}
extraConfigFiles.forEach((destination, from) -> {
if (Files.exists(from.toPath()) == false) {
throw new TestClustersException("Can't create extra config file from " + from + " for " + this +
@ -398,6 +427,7 @@ public class ElasticsearchNode implements TestClusterConfiguration {
private void installModules() {
if (distribution == Distribution.INTEG_TEST) {
logToProcessStdout("Installing " + modules.size() + "modules");
for (File module : modules) {
Path destination = workingDir.resolve("modules").resolve(module.getName().replace(".zip", "").replace("-" + version, ""));
@ -843,7 +873,23 @@ public class ElasticsearchNode implements TestClusterConfiguration {
}
void waitForAllConditions() {
waitForConditions(waitConditions, System.currentTimeMillis(), NODE_UP_TIMEOUT, NODE_UP_TIMEOUT_UNIT, this);
waitForConditions(
waitConditions,
System.currentTimeMillis(),
NODE_UP_TIMEOUT_UNIT.toMillis(NODE_UP_TIMEOUT) +
// Installing plugins at config time and loading them when nods start requires additional time we need to
// account for
ADDITIONAL_CONFIG_TIMEOUT_UNIT.toMillis(ADDITIONAL_CONFIG_TIMEOUT *
(
plugins.size() +
keystoreFiles.size() +
keystoreSettings.size() +
credentials.size()
)
),
TimeUnit.MILLISECONDS,
this
);
}
@Override

View File

@ -0,0 +1,59 @@
package org.elasticsearch.gradle.testclusters;
import org.gradle.api.logging.Logger;
import org.gradle.api.logging.Logging;
import java.util.Collection;
import java.util.HashSet;
import java.util.Iterator;
import java.util.Set;
/**
* Keep an inventory of all running Clusters and stop them when interrupted
*
* This takes advantage of the fact that Gradle interrupts all the threads in the daemon when the build completes.
*/
public class TestClusterCleanupOnShutdown implements Runnable {
private final Logger logger = Logging.getLogger(TestClusterCleanupOnShutdown.class);
private Set<ElasticsearchCluster> clustersToWatch = new HashSet<>();
public void watch(Collection<ElasticsearchCluster> cluster) {
synchronized (clustersToWatch) {
clustersToWatch.addAll(clustersToWatch);
}
}
public void unWatch(Collection<ElasticsearchCluster> cluster) {
synchronized (clustersToWatch) {
clustersToWatch.removeAll(clustersToWatch);
}
}
@Override
public void run() {
try {
while (true) {
Thread.sleep(Long.MAX_VALUE);
}
} catch (InterruptedException interrupted) {
synchronized (clustersToWatch) {
if (clustersToWatch.isEmpty()) {
return;
}
logger.info("Cleanup thread was interrupted, shutting down all clusters");
Iterator<ElasticsearchCluster> iterator = clustersToWatch.iterator();
while (iterator.hasNext()) {
ElasticsearchCluster cluster = iterator.next();
iterator.remove();
try {
cluster.stop(false);
} catch (Exception e) {
logger.warn("Could not shut down {}", cluster, e);
}
}
}
}
}
}

View File

@ -0,0 +1,74 @@
package org.elasticsearch.gradle.testclusters;
import org.gradle.api.Project;
import org.gradle.api.logging.Logger;
import org.gradle.api.logging.Logging;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.TimeUnit;
/**
* This extensions was meant to be used internally by testclusters
*
* It holds synchronization primitives needed to implement the rate limiting.
* This is tricky because we can't use Gradle workers as there's no way to make sure that tests and their clusters are
* allocated atomically, so we could be in a situation where all workers are tests waiting for clusters to start up.
*
* Also auto configures cleanup of executors to make sure we don't leak threads in the daemon.
*/
public class TestClustersCleanupExtension {
private static final int EXECUTOR_SHUTDOWN_TIMEOUT = 1;
private static final TimeUnit EXECUTOR_SHUTDOWN_TIMEOUT_UNIT = TimeUnit.MINUTES;
private static final Logger logger = Logging.getLogger(TestClustersCleanupExtension.class);
private final ExecutorService executorService;
private final TestClusterCleanupOnShutdown cleanupThread;
public TestClustersCleanupExtension() {
executorService = Executors.newSingleThreadExecutor();
cleanupThread = new TestClusterCleanupOnShutdown();
executorService.submit(cleanupThread);
}
public static void createExtension(Project project) {
if (project.getRootProject().getExtensions().findByType(TestClustersCleanupExtension.class) != null) {
return;
}
// Configure the extension on the root project so we have a single instance per run
TestClustersCleanupExtension ext = project.getRootProject().getExtensions().create(
"__testclusters_rate_limit",
TestClustersCleanupExtension.class
);
Thread shutdownHook = new Thread(ext.cleanupThread::run);
Runtime.getRuntime().addShutdownHook(shutdownHook);
project.getGradle().buildFinished(buildResult -> {
ext.executorService.shutdownNow();
try {
if (ext.executorService.awaitTermination(EXECUTOR_SHUTDOWN_TIMEOUT, EXECUTOR_SHUTDOWN_TIMEOUT_UNIT) == false) {
throw new IllegalStateException(
"Failed to shut down executor service after " +
EXECUTOR_SHUTDOWN_TIMEOUT + " " + EXECUTOR_SHUTDOWN_TIMEOUT_UNIT
);
}
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
try {
if (false == Runtime.getRuntime().removeShutdownHook(shutdownHook)) {
logger.warn("Trying to deregister shutdown hook when it was not registered.");
}
} catch (IllegalStateException ese) {
// Thrown when shutdown is in progress
logger.warn("Can't remove shutdown hook", ese);
}
});
}
public TestClusterCleanupOnShutdown getCleanupThread() {
return cleanupThread;
}
}

View File

@ -43,13 +43,9 @@ import java.util.ArrayList;
import java.util.Collections;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.TimeUnit;
import java.util.stream.Collectors;
public class TestClustersPlugin implements Plugin<Project> {
@ -58,18 +54,14 @@ public class TestClustersPlugin implements Plugin<Project> {
public static final String EXTENSION_NAME = "testClusters";
private static final String HELPER_CONFIGURATION_PREFIX = "testclusters";
private static final String SYNC_ARTIFACTS_TASK_NAME = "syncTestClustersArtifacts";
private static final int EXECUTOR_SHUTDOWN_TIMEOUT = 1;
private static final TimeUnit EXECUTOR_SHUTDOWN_TIMEOUT_UNIT = TimeUnit.MINUTES;
private static final Logger logger = Logging.getLogger(TestClustersPlugin.class);
private static final String TESTCLUSTERS_INSPECT_FAILURE = "testclusters.inspect.failure";
private final Map<Task, List<ElasticsearchCluster>> usedClusters = new HashMap<>();
private final Map<ElasticsearchCluster, Integer> claimsInventory = new HashMap<>();
private final Set<ElasticsearchCluster> runningClusters =new HashSet<>();
private final Thread shutdownHook = new Thread(this::shutDownAllClusters);
private final Set<ElasticsearchCluster> runningClusters = new HashSet<>();
private final Boolean allowClusterToSurvive = Boolean.valueOf(System.getProperty(TESTCLUSTERS_INSPECT_FAILURE, "false"));
private ExecutorService executorService = Executors.newSingleThreadExecutor();
public static String getHelperConfigurationName(String version) {
return HELPER_CONFIGURATION_PREFIX + "-" + version;
@ -82,6 +74,8 @@ public class TestClustersPlugin implements Plugin<Project> {
// enable the DSL to describe clusters
NamedDomainObjectContainer<ElasticsearchCluster> container = createTestClustersContainerExtension(project);
TestClustersCleanupExtension.createExtension(project);
// provide a task to be able to list defined clusters.
createListClustersTask(project, container);
@ -100,9 +94,6 @@ public class TestClustersPlugin implements Plugin<Project> {
// After each task we determine if there are clusters that are no longer needed.
configureStopClustersHook(project);
// configure hooks to make sure no test cluster processes survive the build
configureCleanupHooks(project);
// Since we have everything modeled in the DSL, add all the required dependencies e.x. the distribution to the
// configuration so the user doesn't have to repeat this.
autoConfigureClusterDependencies(project, rootProject, container);
@ -196,8 +187,19 @@ public class TestClustersPlugin implements Plugin<Project> {
@Override
public void beforeActions(Task task) {
// we only start the cluster before the actions, so we'll not start it if the task is up-to-date
usedClusters.getOrDefault(task, Collections.emptyList()).stream()
List<ElasticsearchCluster> neededButNotRunning = usedClusters.getOrDefault(
task,
Collections.emptyList()
)
.stream()
.filter(cluster -> runningClusters.contains(cluster) == false)
.collect(Collectors.toList());
project.getRootProject().getExtensions()
.getByType(TestClustersCleanupExtension.class)
.getCleanupThread()
.watch(neededButNotRunning);
neededButNotRunning
.forEach(elasticsearchCluster -> {
elasticsearchCluster.start();
runningClusters.add(elasticsearchCluster);
@ -220,22 +222,36 @@ public class TestClustersPlugin implements Plugin<Project> {
task,
Collections.emptyList()
);
if (clustersUsedByTask.isEmpty()) {
return;
}
logger.info("Clusters were used, stopping and releasing permits");
final int permitsToRelease;
if (state.getFailure() != null) {
// If the task fails, and other tasks use this cluster, the other task will likely never be
// executed at all, so we will never get to un-claim and terminate it.
// executed at all, so we will never be called again to un-claim and terminate it.
clustersUsedByTask.forEach(cluster -> stopCluster(cluster, true));
permitsToRelease = clustersUsedByTask.stream()
.map(cluster -> cluster.getNumberOfNodes())
.reduce(Integer::sum).get();
} else {
clustersUsedByTask.forEach(
cluster -> claimsInventory.put(cluster, claimsInventory.getOrDefault(cluster, 0) - 1)
);
claimsInventory.entrySet().stream()
List<ElasticsearchCluster> stoppingClusers = claimsInventory.entrySet().stream()
.filter(entry -> entry.getValue() == 0)
.filter(entry -> runningClusters.contains(entry.getKey()))
.map(Map.Entry::getKey)
.forEach(cluster -> {
.collect(Collectors.toList());
stoppingClusers.forEach(cluster -> {
stopCluster(cluster, false);
runningClusters.remove(cluster);
});
project.getRootProject().getExtensions()
.getByType(TestClustersCleanupExtension.class)
.getCleanupThread()
.unWatch(stoppingClusers);
}
}
@Override
@ -406,62 +422,4 @@ public class TestClustersPlugin implements Plugin<Project> {
})));
}
private void configureCleanupHooks(Project project) {
// When the Gradle daemon is used, it will interrupt all threads when the build concludes.
// This is our signal to clean up
executorService.submit(() -> {
while (true) {
try {
Thread.sleep(Long.MAX_VALUE);
} catch (InterruptedException interrupted) {
shutDownAllClusters();
Thread.currentThread().interrupt();
return;
}
}
});
// When the Daemon is not used, or runs into issues, rely on a shutdown hook
// When the daemon is used, but does not work correctly and eventually dies off (e.x. due to non interruptible
// thread in the build) process will be stopped eventually when the daemon dies.
Runtime.getRuntime().addShutdownHook(shutdownHook);
// When we don't run into anything out of the ordinary, and the build completes, makes sure to clean up
project.getGradle().buildFinished(buildResult -> {
shutdownExecutorService();
if (false == Runtime.getRuntime().removeShutdownHook(shutdownHook)) {
logger.info("Trying to deregister shutdown hook when it was not registered.");
}
});
}
private void shutdownExecutorService() {
executorService.shutdownNow();
try {
if (executorService.awaitTermination(EXECUTOR_SHUTDOWN_TIMEOUT, EXECUTOR_SHUTDOWN_TIMEOUT_UNIT) == false) {
throw new IllegalStateException(
"Failed to shut down executor service after " +
EXECUTOR_SHUTDOWN_TIMEOUT + " " + EXECUTOR_SHUTDOWN_TIMEOUT_UNIT
);
}
} catch (InterruptedException e) {
logger.info("Wait for testclusters shutdown interrupted", e);
Thread.currentThread().interrupt();
}
}
private void shutDownAllClusters() {
synchronized (runningClusters) {
if (runningClusters.isEmpty()) {
return;
}
Iterator<ElasticsearchCluster> iterator = runningClusters.iterator();
while (iterator.hasNext()) {
ElasticsearchCluster next = iterator.next();
iterator.remove();
next.stop(false);
}
}
}
}

View File

@ -1,3 +1,3 @@
org.gradle.daemon=true
org.gradle.jvmargs=-Xmx2g -XX:+HeapDumpOnOutOfMemoryError -Xss2m
org.gradle.jvmargs=-Xmx3g -XX:+HeapDumpOnOutOfMemoryError -Xss2m
options.forkOptions.memoryMaximumSize=2g