kevinrr888 commented on code in PR #28: URL: https://github.com/apache/accumulo-classloaders/pull/28#discussion_r2456204854
########## modules/classloaders/hdfs-class-loader/src/main/java/org/apache/accumulo/classloader/hdfs/HDFSContextClassLoaderFactory.java: ########## @@ -0,0 +1,402 @@ +/* + * 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 + * + * https://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.accumulo.classloader.hdfs; + +import java.io.ByteArrayInputStream; +import java.io.IOException; +import java.io.InputStream; +import java.io.InputStreamReader; +import java.net.URL; +import java.net.URLClassLoader; +import java.nio.charset.StandardCharsets; +import java.nio.file.Files; +import java.security.MessageDigest; +import java.security.NoSuchAlgorithmException; +import java.util.List; +import java.util.UUID; +import java.util.concurrent.ExecutionException; +import java.util.concurrent.TimeUnit; +import java.util.concurrent.atomic.AtomicBoolean; + +import org.apache.accumulo.core.spi.common.ContextClassLoaderEnvironment; +import org.apache.accumulo.core.spi.common.ContextClassLoaderFactory; +import org.apache.accumulo.core.util.Pair; +import org.apache.accumulo.core.util.Retry; +import org.apache.commons.io.FileUtils; +import org.apache.hadoop.conf.Configuration; +import org.apache.hadoop.fs.FileSystem; +import org.apache.hadoop.fs.Path; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import com.google.common.annotations.VisibleForTesting; +import com.google.common.cache.Cache; +import com.google.common.cache.CacheBuilder; +import com.google.gson.Gson; + +import edu.umd.cs.findbugs.annotations.SuppressFBWarnings; + +/** + * A {@link ContextClassLoaderFactory} implementation which uses a {@link URLClassLoader} per + * defined context. To use this class, need to set the Accumulo configuration property + * <b>general.context.class.loader.factory</b> to the fully qualified name of this class. <br> + * <br> + * Configuration of this class is done by having a base directory in HDFS which stores context + * directories (e.g., "hdfs://test:1234/contexts" could be the base directory) and where the + * children are specific contexts (e.g., contextA, contextB). The "contexts" directory should not + * contain any other files or directories. Each context directory should contain a manifest file + * <b>manifest.json</b> and JAR files. The manifest file defines the context name and the JAR info + * for what JARs should be in the {@link URLClassLoader} for that context. For example: + * + * <pre> + * { + * "context": "contextA", + * "jars": [ + * { + * "name": "Iterators.jar", + * "checksum": "f2ca1bb6c7e907d06dafe4687e579fce76b37e4e93b7605022da52e6ccc26fd2" + * }, + * { + * "name": "IteratorsV2.jar", + * "checksum": "934ee77f70dc82403618c154aa63af6f4ebbe3ac1eaf22df21e7e64f0fb6643d" + * } + * ] + * } + * </pre> + * + * Two system properties need to be set to use this class: <br> + * <b>hdfs.contexts.class.loader.base.dir</b> <br> + * This defines where the context directories exist within HDFS (e.g., "hdfs://test:1234/contexts" + * in our example) <br> + * <b>local.contexts.class.loader.download.dir</b> <br> + * This defines where the context info will be locally downloaded to. Will only download the JARs. + * For example, if set to "/path/to/contexts" and + * {@link HDFSContextClassLoaderFactory#getClassLoader(String)} is called with the argument + * "contextA", will create the local directory "/path/to/contexts/contextA" (note that all parent + * directories will be created if needed), will download the JARs Iterators.jar and IteratorsV2.jar + * from HDFS to the contextA directory. The {@link URLClassLoader} will be constructed from these + * local JARs, will be associated with "contextA", and will be cached. <br> + * Another optional system property: <br> + * <b>manifest.file.check.interval.sec</b> <br> + * may be set as well. This defines how often (in sec) the cached contexts will be checked and/or + * updated based on changes to their associated manifest file. If not set, default interval is 10s. + * Changes to this will take effect on next interval. + */ +public class HDFSContextClassLoaderFactory implements ContextClassLoaderFactory { + private static final Logger LOG = LoggerFactory.getLogger(HDFSContextClassLoaderFactory.class); + public static final String MANIFEST_FILE_NAME = "manifest.json"; + private static final String HASH_ALG = "SHA-256"; + private static final long DEFAULT_MANIFEST_CHECK_INTERVAL = TimeUnit.SECONDS.toMillis(10); + public static final String HDFS_CONTEXTS_BASE_DIR = "hdfs.contexts.class.loader.base.dir"; + public static final String LOCAL_CONTEXTS_DOWNLOAD_DIR = + "local.contexts.class.loader.download.dir"; + public static final String MANIFEST_FILE_CHECK_INTERVAL = "manifest.file.check.interval.sec"; + // Cache the class loaders for re-use + // WeakReferences are used so that the class loaders can be cleaned up when no longer needed + // Classes that are loaded contain a reference to the class loader used to load them + // so the class loader will be garbage collected when no more classes are loaded that reference it + private final Cache<Path,Pair<String,URLClassLoader>> classLoaders = + CacheBuilder.newBuilder().weakValues().build(); + private FileSystem hdfs; + private Path HDFSContextsDir; + private java.nio.file.Path localContextsDir; + private final Configuration hadoopConf = new Configuration(); + private final Thread manifestFileChecker = new Thread(new ManifestFileChecker()); + private final AtomicBoolean shutdownManifestFileChecker = new AtomicBoolean(false); + + @VisibleForTesting + public class ManifestFileChecker implements Runnable { + @Override + public void run() { + while (!shutdownManifestFileChecker.get()) { + var manifestFileCheckInterval = System.getProperty(MANIFEST_FILE_CHECK_INTERVAL); + long sleepTime = manifestFileCheckInterval == null ? DEFAULT_MANIFEST_CHECK_INTERVAL + : TimeUnit.SECONDS.toMillis(Integer.parseInt(manifestFileCheckInterval)); + + classLoaders.asMap().keySet().forEach(HDFSManifestFile -> classLoaders.asMap() + .computeIfPresent(HDFSManifestFile, (key, existingVal) -> { + var existingChecksum = existingVal.getFirst(); + try (var HDFSManifestFileIn = hdfs.open(HDFSManifestFile)) { + var HDFSManifestFileBytes = HDFSManifestFileIn.readAllBytes(); + var computedChecksum = checksum(HDFSManifestFileBytes); Review Comment: This is a nice improvement, added. -- 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]
