HDFS-15305. Extend ViewFS and provide ViewFileSystemOverloadScheme implementation with scheme configurable. Contributed by Uma Maheswara Rao G.

(cherry picked from commit 9c8236d04d)
This commit is contained in:
Uma Maheswara Rao G 2020-05-04 17:55:40 -07:00 committed by Virajith Jalaparti
parent 7aade932c7
commit edf52d29f1
8 changed files with 937 additions and 10 deletions

View File

@ -42,4 +42,6 @@ public interface FsConstants {
*/
public static final URI VIEWFS_URI = URI.create("viewfs:///");
public static final String VIEWFS_SCHEME = "viewfs";
String FS_VIEWFS_OVERLOAD_SCHEME_TARGET_FS_IMPL_PATTERN =
"fs.viewfs.overload.scheme.target.%s.impl";
}

View File

@ -18,9 +18,9 @@
package org.apache.hadoop.fs.viewfs;
import static org.apache.hadoop.fs.impl.PathCapabilitiesSupport.validatePathCapabilityArgs;
import static org.apache.hadoop.fs.viewfs.Constants.PERMISSION_555;
import static org.apache.hadoop.fs.viewfs.Constants.CONFIG_VIEWFS_ENABLE_INNER_CACHE;
import static org.apache.hadoop.fs.viewfs.Constants.CONFIG_VIEWFS_ENABLE_INNER_CACHE_DEFAULT;
import static org.apache.hadoop.fs.viewfs.Constants.PERMISSION_555;
import java.io.FileNotFoundException;
import java.io.IOException;
@ -35,9 +35,9 @@ import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.Objects;
import java.util.Set;
import java.util.Map.Entry;
import org.apache.hadoop.classification.InterfaceAudience;
import org.apache.hadoop.classification.InterfaceStability;
@ -96,16 +96,49 @@ public class ViewFileSystem extends FileSystem {
return readOnlyMountTable(operation, p.toString());
}
/**
* File system instance getter.
*/
static class FsGetter {
/**
* Gets new file system instance of given uri.
*/
public FileSystem getNewInstance(URI uri, Configuration conf)
throws IOException {
return FileSystem.newInstance(uri, conf);
}
/**
* Gets file system instance of given uri.
*/
public FileSystem get(URI uri, Configuration conf) throws IOException {
return FileSystem.get(uri, conf);
}
}
/**
* Gets file system creator instance.
*/
protected FsGetter fsGetter() {
return new FsGetter();
}
/**
* Caching children filesystems. HADOOP-15565.
*/
static class InnerCache {
private Map<Key, FileSystem> map = new HashMap<>();
private FsGetter fsCreator;
InnerCache(FsGetter fsCreator) {
this.fsCreator = fsCreator;
}
FileSystem get(URI uri, Configuration config) throws IOException {
Key key = new Key(uri);
if (map.get(key) == null) {
FileSystem fs = FileSystem.newInstance(uri, config);
FileSystem fs = fsCreator.getNewInstance(uri, config);
map.put(key, fs);
return fs;
} else {
@ -193,7 +226,7 @@ public class ViewFileSystem extends FileSystem {
final long creationTime; // of the the mount table
final UserGroupInformation ugi; // the user/group of user who created mtable
URI myUri;
private URI myUri;
private Path workingDir;
Configuration config;
InodeTree<FileSystem> fsState; // the fs state; ie the mount table
@ -255,13 +288,13 @@ public class ViewFileSystem extends FileSystem {
config = conf;
enableInnerCache = config.getBoolean(CONFIG_VIEWFS_ENABLE_INNER_CACHE,
CONFIG_VIEWFS_ENABLE_INNER_CACHE_DEFAULT);
final InnerCache innerCache = new InnerCache();
FsGetter fsGetter = fsGetter();
final InnerCache innerCache = new InnerCache(fsGetter);
// Now build client side view (i.e. client side mount table) from config.
final String authority = theUri.getAuthority();
try {
myUri = new URI(FsConstants.VIEWFS_SCHEME, authority, "/", null, null);
myUri = new URI(getScheme(), authority, "/", null, null);
fsState = new InodeTree<FileSystem>(conf, authority) {
@Override
protected FileSystem getTargetFileSystem(final URI uri)
throws URISyntaxException, IOException {
@ -269,7 +302,7 @@ public class ViewFileSystem extends FileSystem {
if (enableInnerCache) {
fs = innerCache.get(uri, config);
} else {
fs = FileSystem.get(uri, config);
fs = fsGetter.get(uri, config);
}
return new ChRootedFileSystem(fs, uri);
}

View File

@ -0,0 +1,204 @@
/**
* 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.hadoop.fs.viewfs;
import java.io.IOException;
import java.lang.reflect.Constructor;
import java.lang.reflect.InvocationTargetException;
import java.net.URI;
import org.apache.hadoop.classification.InterfaceAudience;
import org.apache.hadoop.classification.InterfaceStability;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.FsConstants;
import org.apache.hadoop.fs.UnsupportedFileSystemException;
/******************************************************************************
* This class is extended from the ViewFileSystem for the overloaded scheme
* file system. Mount link configurations and in-memory mount table
* building behaviors are inherited from ViewFileSystem. Unlike ViewFileSystem
* scheme (viewfs://), the users would be able to use any scheme.
*
* To use this class, the following configurations need to be added in
* core-site.xml file.
* 1) fs.<scheme>.impl
* = org.apache.hadoop.fs.viewfs.ViewFileSystemOverloadScheme
* 2) fs.viewfs.overload.scheme.target.<scheme>.impl
* = <hadoop compatible file system implementation class name for the
* <scheme>"
*
* Here <scheme> can be any scheme, but with that scheme there should be a
* hadoop compatible file system available. Second configuration value should
* be the respective scheme's file system implementation class.
* Example: if scheme is configured with "hdfs", then the 2nd configuration
* class name will be org.apache.hadoop.hdfs.DistributedFileSystem.
* if scheme is configured with "s3a", then the 2nd configuration class name
* will be org.apache.hadoop.fs.s3a.S3AFileSystem.
*
* Use Case 1:
* ===========
* If users want some of their existing cluster (hdfs://Cluster)
* data to mount with other hdfs and object store clusters(hdfs://NN1,
* o3fs://bucket1.volume1/, s3a://bucket1/)
*
* fs.viewfs.mounttable.Cluster./user = hdfs://NN1/user
* fs.viewfs.mounttable.Cluster./data = o3fs://bucket1.volume1/data
* fs.viewfs.mounttable.Cluster./backup = s3a://bucket1/backup/
*
* Op1: Create file hdfs://Cluster/user/fileA will go to hdfs://NN1/user/fileA
* Op2: Create file hdfs://Cluster/data/datafile will go to
* o3fs://bucket1.volume1/data/datafile
* Op3: Create file hdfs://Cluster/backup/data.zip will go to
* s3a://bucket1/backup/data.zip
*
* Use Case 2:
* ===========
* If users want some of their existing cluster (s3a://bucketA/)
* data to mount with other hdfs and object store clusters
* (hdfs://NN1, o3fs://bucket1.volume1/)
*
* fs.viewfs.mounttable.bucketA./user = hdfs://NN1/user
* fs.viewfs.mounttable.bucketA./data = o3fs://bucket1.volume1/data
* fs.viewfs.mounttable.bucketA./salesDB = s3a://bucketA/salesDB/
*
* Op1: Create file s3a://bucketA/user/fileA will go to hdfs://NN1/user/fileA
* Op2: Create file s3a://bucketA/data/datafile will go to
* o3fs://bucket1.volume1/data/datafile
* Op3: Create file s3a://bucketA/salesDB/dbfile will go to
* s3a://bucketA/salesDB/dbfile
*****************************************************************************/
@InterfaceAudience.LimitedPrivate({ "MapReduce", "HBase", "Hive" })
@InterfaceStability.Evolving
public class ViewFileSystemOverloadScheme extends ViewFileSystem {
private URI myUri;
public ViewFileSystemOverloadScheme() throws IOException {
super();
}
@Override
public String getScheme() {
return myUri.getScheme();
}
@Override
public void initialize(URI theUri, Configuration conf) throws IOException {
if (LOG.isDebugEnabled()) {
LOG.debug("Initializing the ViewFileSystemOverloadScheme with the uri: "
+ theUri);
}
this.myUri = theUri;
super.initialize(theUri, conf);
}
/**
* This method is overridden because in ViewFileSystemOverloadScheme if
* overloaded cheme matches with mounted target fs scheme, file system should
* be created without going into fs.<scheme>.impl based resolution. Otherwise
* it will end up in an infinite loop as the target will be resolved again
* to ViewFileSystemOverloadScheme as fs.<scheme>.impl points to
* ViewFileSystemOverloadScheme. So, below method will initialize the
* fs.viewfs.overload.scheme.target.<scheme>.impl. Other schemes can
* follow fs.newInstance
*/
@Override
protected FsGetter fsGetter() {
return new FsGetter() {
@Override
public FileSystem getNewInstance(URI uri, Configuration conf)
throws IOException {
if (uri.getScheme().equals(getScheme())) {
if (LOG.isDebugEnabled()) {
LOG.debug(
"The file system initialized uri scheme is matching with the "
+ "given target uri scheme. The target uri is: " + uri);
}
/*
* Avoid looping when target fs scheme is matching to overloaded
* scheme.
*/
return createFileSystem(uri, conf);
} else {
return FileSystem.newInstance(uri, conf);
}
}
/**
* When ViewFileSystemOverloadScheme scheme and target uri scheme are
* matching, it will not take advantage of FileSystem cache as it will
* create instance directly. For caching needs please set
* "fs.viewfs.enable.inner.cache" to true.
*/
@Override
public FileSystem get(URI uri, Configuration conf) throws IOException {
if (uri.getScheme().equals(getScheme())) {
// Avoid looping when target fs scheme is matching to overloaded
// scheme.
if (LOG.isDebugEnabled()) {
LOG.debug(
"The file system initialized uri scheme is matching with the "
+ "given target uri scheme. So, the target file system "
+ "instances will not be cached. To cache fs instances, "
+ "please set fs.viewfs.enable.inner.cache to true. "
+ "The target uri is: " + uri);
}
return createFileSystem(uri, conf);
} else {
return FileSystem.get(uri, conf);
}
}
private FileSystem createFileSystem(URI uri, Configuration conf)
throws IOException {
final String fsImplConf = String.format(
FsConstants.FS_VIEWFS_OVERLOAD_SCHEME_TARGET_FS_IMPL_PATTERN,
uri.getScheme());
Class<?> clazz = conf.getClass(fsImplConf, null);
if (clazz == null) {
throw new UnsupportedFileSystemException(
String.format("%s=null: %s: %s", fsImplConf,
"No overload scheme fs configured", uri.getScheme()));
}
FileSystem fs = (FileSystem) newInstance(clazz, uri, conf);
fs.initialize(uri, conf);
return fs;
}
private <T> T newInstance(Class<T> theClass, URI uri,
Configuration conf) {
T result;
try {
Constructor<T> meth = theClass.getConstructor();
meth.setAccessible(true);
result = meth.newInstance();
} catch (InvocationTargetException e) {
Throwable cause = e.getCause();
if (cause instanceof RuntimeException) {
throw (RuntimeException) cause;
} else {
throw new RuntimeException(cause);
}
} catch (Exception e) {
throw new RuntimeException(e);
}
return result;
}
};
}
}

View File

@ -862,7 +862,8 @@ public abstract class FileSystemContractBaseTest {
found);
}
private void assertListStatusFinds(Path dir, Path subdir) throws IOException {
protected void assertListStatusFinds(Path dir, Path subdir)
throws IOException {
FileStatus[] stats = fs.listStatus(dir);
boolean found = false;
StringBuilder builder = new StringBuilder();

View File

@ -0,0 +1,149 @@
/**
* 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.hadoop.fs.viewfs;
import java.io.IOException;
import java.net.URI;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.FileSystemTestHelper;
import org.apache.hadoop.fs.FsConstants;
import org.apache.hadoop.fs.LocalFileSystem;
import org.apache.hadoop.fs.Path;
import org.junit.After;
import org.junit.Assert;
import org.junit.Before;
import org.junit.Test;
/**
*
* Test the TestViewFileSystemOverloadSchemeLF using a file with authority:
* file://mountTableName/ i.e, the authority is used to load a mount table.
*/
public class TestViewFileSystemOverloadSchemeLocalFileSystem {
private static final String FILE = "file";
private static final Log LOG =
LogFactory.getLog(TestViewFileSystemOverloadSchemeLocalFileSystem.class);
private FileSystem fsTarget;
private Configuration conf;
private Path targetTestRoot;
private FileSystemTestHelper fileSystemTestHelper =
new FileSystemTestHelper();
@Before
public void setUp() throws Exception {
conf = new Configuration();
conf.set(String.format("fs.%s.impl", FILE),
ViewFileSystemOverloadScheme.class.getName());
conf.set(String.format(
FsConstants.FS_VIEWFS_OVERLOAD_SCHEME_TARGET_FS_IMPL_PATTERN, FILE),
LocalFileSystem.class.getName());
fsTarget = new LocalFileSystem();
fsTarget.initialize(new URI("file:///"), conf);
// create the test root on local_fs
targetTestRoot = fileSystemTestHelper.getAbsoluteTestRootPath(fsTarget);
fsTarget.delete(targetTestRoot, true);
fsTarget.mkdirs(targetTestRoot);
}
/**
* Tests write file and read file with ViewFileSystemOverloadScheme.
*/
@Test
public void testLocalTargetLinkWriteSimple() throws IOException {
LOG.info("Starting testLocalTargetLinkWriteSimple");
final String testString = "Hello Local!...";
final Path lfsRoot = new Path("/lfsRoot");
ConfigUtil.addLink(conf, lfsRoot.toString(),
URI.create(targetTestRoot + "/local"));
try (FileSystem lViewFs = FileSystem.get(URI.create("file:///"), conf)) {
final Path testPath = new Path(lfsRoot, "test.txt");
try (FSDataOutputStream fsDos = lViewFs.create(testPath)) {
fsDos.writeUTF(testString);
}
try (FSDataInputStream lViewIs = lViewFs.open(testPath)) {
Assert.assertEquals(testString, lViewIs.readUTF());
}
}
}
/**
* Tests create file and delete file with ViewFileSystemOverloadScheme.
*/
@Test
public void testLocalFsCreateAndDelete() throws Exception {
LOG.info("Starting testLocalFsCreateAndDelete");
ConfigUtil.addLink(conf, "mt", "/lfsroot",
URI.create(targetTestRoot + "/wd2"));
final URI mountURI = URI.create("file://mt/");
try (FileSystem lViewFS = FileSystem.get(mountURI, conf)) {
Path testPath = new Path(mountURI.toString() + "/lfsroot/test");
lViewFS.createNewFile(testPath);
Assert.assertTrue(lViewFS.exists(testPath));
lViewFS.delete(testPath, true);
Assert.assertFalse(lViewFS.exists(testPath));
}
}
/**
* Tests root level file with linkMergeSlash with
* ViewFileSystemOverloadScheme.
*/
@Test
public void testLocalFsLinkSlashMerge() throws Exception {
LOG.info("Starting testLocalFsLinkSlashMerge");
ConfigUtil.addLinkMergeSlash(conf, "mt",
URI.create(targetTestRoot + "/wd2"));
final URI mountURI = URI.create("file://mt/");
try (FileSystem lViewFS = FileSystem.get(mountURI, conf)) {
Path fileOnRoot = new Path(mountURI.toString() + "/NewFile");
lViewFS.createNewFile(fileOnRoot);
Assert.assertTrue(lViewFS.exists(fileOnRoot));
}
}
/**
* Tests with linkMergeSlash and other mounts in
* ViewFileSystemOverloadScheme.
*/
@Test(expected = IOException.class)
public void testLocalFsLinkSlashMergeWithOtherMountLinks() throws Exception {
LOG.info("Starting testLocalFsLinkSlashMergeWithOtherMountLinks");
ConfigUtil.addLink(conf, "mt", "/lfsroot",
URI.create(targetTestRoot + "/wd2"));
ConfigUtil.addLinkMergeSlash(conf, "mt",
URI.create(targetTestRoot + "/wd2"));
final URI mountURI = URI.create("file://mt/");
FileSystem.get(mountURI, conf);
Assert.fail("A merge slash cannot be configured with other mount links.");
}
@After
public void tearDown() throws Exception {
if (null != fsTarget) {
fsTarget.delete(fileSystemTestHelper.getTestRootPath(fsTarget), true);
fsTarget.close();
}
}
}

View File

@ -50,6 +50,7 @@ import org.apache.hadoop.fs.permission.AclStatus;
import org.apache.hadoop.fs.permission.AclUtil;
import org.apache.hadoop.fs.permission.FsAction;
import org.apache.hadoop.fs.permission.FsPermission;
import org.apache.hadoop.fs.viewfs.ViewFileSystem.FsGetter;
import org.apache.hadoop.fs.viewfs.ViewFileSystem.MountPoint;
import org.apache.hadoop.security.AccessControlException;
import org.apache.hadoop.security.Credentials;
@ -1279,7 +1280,8 @@ abstract public class ViewFileSystemBaseTest {
@Test
public void testViewFileSystemInnerCache() throws Exception {
ViewFileSystem.InnerCache cache = new ViewFileSystem.InnerCache();
ViewFileSystem.InnerCache cache =
new ViewFileSystem.InnerCache(new FsGetter());
FileSystem fs = cache.get(fsTarget.getUri(), conf);
// InnerCache caches filesystem.

View File

@ -0,0 +1,126 @@
/**
* 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.hadoop.fs.viewfs;
import static org.junit.Assume.assumeTrue;
import java.io.File;
import java.io.IOException;
import java.net.URI;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.CommonConfigurationKeys;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.FileSystemContractBaseTest;
import org.apache.hadoop.fs.FsConstants;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hdfs.AppendTestUtil;
import org.apache.hadoop.hdfs.DistributedFileSystem;
import org.apache.hadoop.hdfs.HdfsConfiguration;
import org.apache.hadoop.hdfs.MiniDFSCluster;
import org.apache.hadoop.hdfs.TestHDFSFileSystemContract;
import org.apache.hadoop.security.AccessControlException;
import org.apache.hadoop.security.UserGroupInformation;
import org.apache.hadoop.test.GenericTestUtils;
import org.junit.AfterClass;
import org.junit.Before;
import org.junit.BeforeClass;
import org.junit.Ignore;
import org.junit.Test;
/**
* Tests ViewFileSystemOverloadScheme with file system contract tests.
*/
public class TestViewFileSystemOverloadSchemeHdfsFileSystemContract
extends TestHDFSFileSystemContract {
private static MiniDFSCluster cluster;
private static String defaultWorkingDirectory;
private static Configuration conf = new HdfsConfiguration();
@BeforeClass
public static void init() throws IOException {
final File basedir = GenericTestUtils.getRandomizedTestDir();
conf.set(CommonConfigurationKeys.FS_PERMISSIONS_UMASK_KEY,
FileSystemContractBaseTest.TEST_UMASK);
cluster = new MiniDFSCluster.Builder(conf, basedir)
.numDataNodes(2)
.build();
defaultWorkingDirectory =
"/user/" + UserGroupInformation.getCurrentUser().getShortUserName();
}
@Before
public void setUp() throws Exception {
conf.set(String.format("fs.%s.impl", "hdfs"),
ViewFileSystemOverloadScheme.class.getName());
conf.set(String.format(
FsConstants.FS_VIEWFS_OVERLOAD_SCHEME_TARGET_FS_IMPL_PATTERN,
"hdfs"),
DistributedFileSystem.class.getName());
URI defaultFSURI =
URI.create(conf.get(CommonConfigurationKeys.FS_DEFAULT_NAME_KEY));
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), "/user",
defaultFSURI);
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), "/append",
defaultFSURI);
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(),
"/FileSystemContractBaseTest/",
new URI(defaultFSURI.toString() + "/FileSystemContractBaseTest/"));
fs = FileSystem.get(conf);
}
@AfterClass
public static void tearDownAfter() throws Exception {
if (cluster != null) {
cluster.shutdown();
cluster = null;
}
}
@Override
protected String getDefaultWorkingDirectory() {
return defaultWorkingDirectory;
}
@Override
@Test
public void testAppend() throws IOException {
AppendTestUtil.testAppend(fs, new Path("/append/f"));
}
@Override
@Test(expected = AccessControlException.class)
public void testRenameRootDirForbidden() throws Exception {
super.testRenameRootDirForbidden();
}
@Override
@Test
public void testListStatusRootDir() throws Throwable {
assumeTrue(rootDirTestEnabled());
Path dir = path("/");
Path child = path("/FileSystemContractBaseTest");
assertListStatusFinds(dir, child);
}
@Override
@Ignore // This test same as above in this case.
public void testLSRootDir() throws Throwable {
}
}

View File

@ -0,0 +1,410 @@
/**
* 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.hadoop.fs.viewfs;
import java.io.File;
import java.io.IOException;
import java.net.URI;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.CommonConfigurationKeys;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.FsConstants;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.RawLocalFileSystem;
import org.apache.hadoop.fs.UnsupportedFileSystemException;
import org.apache.hadoop.hdfs.DFSConfigKeys;
import org.apache.hadoop.hdfs.DistributedFileSystem;
import org.apache.hadoop.hdfs.MiniDFSCluster;
import org.apache.hadoop.security.AccessControlException;
import org.apache.hadoop.test.PathUtils;
import org.junit.After;
import org.junit.Assert;
import org.junit.Before;
import org.junit.Test;
/**
* Tests ViewFileSystemOverloadScheme with configured mount links.
*/
public class TestViewFileSystemOverloadSchemeWithHdfsScheme {
private static final String FS_IMPL_PATTERN_KEY = "fs.%s.impl";
private static final String HDFS_SCHEME = "hdfs";
private Configuration conf = null;
private MiniDFSCluster cluster = null;
private URI defaultFSURI;
private File localTargetDir;
private static final String TEST_ROOT_DIR = PathUtils
.getTestDirName(TestViewFileSystemOverloadSchemeWithHdfsScheme.class);
private static final String HDFS_USER_FOLDER = "/HDFSUser";
private static final String LOCAL_FOLDER = "/local";
@Before
public void startCluster() throws IOException {
conf = new Configuration();
conf.setBoolean(DFSConfigKeys.DFS_NAMENODE_DELEGATION_TOKEN_ALWAYS_USE_KEY,
true);
conf.set(String.format(FS_IMPL_PATTERN_KEY, HDFS_SCHEME),
ViewFileSystemOverloadScheme.class.getName());
conf.set(String.format(
FsConstants.FS_VIEWFS_OVERLOAD_SCHEME_TARGET_FS_IMPL_PATTERN,
HDFS_SCHEME), DistributedFileSystem.class.getName());
cluster = new MiniDFSCluster.Builder(conf).numDataNodes(2).build();
cluster.waitClusterUp();
defaultFSURI =
URI.create(conf.get(CommonConfigurationKeys.FS_DEFAULT_NAME_KEY));
localTargetDir = new File(TEST_ROOT_DIR, "/root/");
Assert.assertEquals(HDFS_SCHEME, defaultFSURI.getScheme()); // hdfs scheme.
}
@After
public void tearDown() throws IOException {
if (cluster != null) {
FileSystem.closeAll();
cluster.shutdown();
}
}
private void createLinks(boolean needFalbackLink, Path hdfsTargetPath,
Path localTragetPath) {
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), HDFS_USER_FOLDER,
hdfsTargetPath.toUri());
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), LOCAL_FOLDER,
localTragetPath.toUri());
if (needFalbackLink) {
ConfigUtil.addLinkFallback(conf, defaultFSURI.getAuthority(),
hdfsTargetPath.toUri());
}
}
/**
* Create mount links as follows.
* hdfs://localhost:xxx/HDFSUser --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/local --> file://TEST_ROOT_DIR/root/
*
* create file /HDFSUser/testfile should create in hdfs
* create file /local/test should create directory in local fs
*/
@Test(timeout = 30000)
public void testMountLinkWithLocalAndHDFS() throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
final Path localTragetPath = new Path(localTargetDir.toURI());
createLinks(false, hdfsTargetPath, localTragetPath);
// /HDFSUser/testfile
Path hdfsFile = new Path(HDFS_USER_FOLDER + "/testfile");
// /local/test
Path localDir = new Path(LOCAL_FOLDER + "/test");
try (ViewFileSystemOverloadScheme fs
= (ViewFileSystemOverloadScheme) FileSystem.get(conf)) {
Assert.assertEquals(2, fs.getMountPoints().length);
fs.createNewFile(hdfsFile); // /HDFSUser/testfile
fs.mkdirs(localDir); // /local/test
}
// Initialize HDFS and test files exist in ls or not
try (DistributedFileSystem dfs = new DistributedFileSystem()) {
dfs.initialize(defaultFSURI, conf);
Assert.assertTrue(dfs.exists(
new Path(Path.getPathWithoutSchemeAndAuthority(hdfsTargetPath),
hdfsFile.getName()))); // should be in hdfs.
Assert.assertFalse(dfs.exists(
new Path(Path.getPathWithoutSchemeAndAuthority(localTragetPath),
localDir.getName()))); // should not be in local fs.
}
try (RawLocalFileSystem lfs = new RawLocalFileSystem()) {
lfs.initialize(localTragetPath.toUri(), conf);
Assert.assertFalse(lfs.exists(
new Path(Path.getPathWithoutSchemeAndAuthority(hdfsTargetPath),
hdfsFile.getName()))); // should not be in hdfs.
Assert.assertTrue(lfs.exists(
new Path(Path.getPathWithoutSchemeAndAuthority(localTragetPath),
localDir.getName()))); // should be in local fs.
}
}
/**
* Create mount links as follows.
* hdfs://localhost:xxx/HDFSUser --> nonexistent://NonExistent/User/
* It should fail to add non existent fs link.
*/
@Test(expected = IOException.class, timeout = 30000)
public void testMountLinkWithNonExistentLink() throws Exception {
final String userFolder = "/User";
final Path nonExistTargetPath =
new Path("nonexistent://NonExistent" + userFolder);
/**
* Below addLink will create following mount points
* hdfs://localhost:xxx/User --> nonexistent://NonExistent/User/
*/
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), userFolder,
nonExistTargetPath.toUri());
FileSystem.get(conf);
Assert.fail("Expected to fail with non existent link");
}
/**
* Create mount links as follows.
* hdfs://localhost:xxx/HDFSUser --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/local --> file://TEST_ROOT_DIR/root/
* ListStatus on / should list the mount links.
*/
@Test(timeout = 30000)
public void testListStatusOnRootShouldListAllMountLinks() throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
final Path localTragetPath = new Path(localTargetDir.toURI());
createLinks(false, hdfsTargetPath, localTragetPath);
try (FileSystem fs = FileSystem.get(conf)) {
FileStatus[] ls = fs.listStatus(new Path("/"));
Assert.assertEquals(2, ls.length);
String lsPath1 =
Path.getPathWithoutSchemeAndAuthority(ls[0].getPath()).toString();
String lsPath2 =
Path.getPathWithoutSchemeAndAuthority(ls[1].getPath()).toString();
Assert.assertTrue(
HDFS_USER_FOLDER.equals(lsPath1) || LOCAL_FOLDER.equals(lsPath1));
Assert.assertTrue(
HDFS_USER_FOLDER.equals(lsPath2) || LOCAL_FOLDER.equals(lsPath2));
}
}
/**
* Create mount links as follows
* hdfs://localhost:xxx/HDFSUser --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/local --> file://TEST_ROOT_DIR/root/
* ListStatus non mount directory should fail.
*/
@Test(expected = IOException.class, timeout = 30000)
public void testListStatusOnNonMountedPath() throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
final Path localTragetPath = new Path(localTargetDir.toURI());
createLinks(false, hdfsTargetPath, localTragetPath);
try (FileSystem fs = FileSystem.get(conf)) {
fs.listStatus(new Path("/nonMount"));
Assert.fail("It should fail as no mount link with /nonMount");
}
}
/**
* Create mount links as follows hdfs://localhost:xxx/HDFSUser -->
* hdfs://localhost:xxx/HDFSUser/ hdfs://localhost:xxx/local -->
* file://TEST_ROOT_DIR/root/ fallback --> hdfs://localhost:xxx/HDFSUser/
* Creating file or directory at non root level should succeed with fallback
* links.
*/
@Test(timeout = 30000)
public void testWithLinkFallBack() throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
final Path localTragetPath = new Path(localTargetDir.toURI());
createLinks(true, hdfsTargetPath, localTragetPath);
try (FileSystem fs = FileSystem.get(conf)) {
fs.createNewFile(new Path("/nonMount/myfile"));
FileStatus[] ls = fs.listStatus(new Path("/nonMount"));
Assert.assertEquals(1, ls.length);
Assert.assertEquals(
Path.getPathWithoutSchemeAndAuthority(ls[0].getPath()).getName(),
"myfile");
}
}
/**
* Create mount links as follows
* hdfs://localhost:xxx/HDFSUser --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/local --> file://TEST_ROOT_DIR/root/
*
* It cannot find any mount link. ViewFS expects a mount point from root.
*/
@Test(expected = NotInMountpointException.class, timeout = 30000)
public void testCreateOnRootShouldFailWhenMountLinkConfigured()
throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
final Path localTragetPath = new Path(localTargetDir.toURI());
createLinks(false, hdfsTargetPath, localTragetPath);
try (FileSystem fs = FileSystem.get(conf)) {
fs.createNewFile(new Path("/newFileOnRoot"));
Assert.fail("It should fail as root is read only in viewFS.");
}
}
/**
* Create mount links as follows
* hdfs://localhost:xxx/HDFSUser --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/local --> file://TEST_ROOT_DIR/root/
* fallback --> hdfs://localhost:xxx/HDFSUser/
*
* It will find fallback link, but root is not accessible and read only.
*/
@Test(expected = AccessControlException.class, timeout = 30000)
public void testCreateOnRootShouldFailEvenFallBackMountLinkConfigured()
throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
final Path localTragetPath = new Path(localTargetDir.toURI());
createLinks(true, hdfsTargetPath, localTragetPath);
try (FileSystem fs = FileSystem.get(conf)) {
fs.createNewFile(new Path("/onRootWhenFallBack"));
Assert.fail(
"It should fail as root is read only in viewFS, even when configured"
+ " with fallback.");
}
}
/**
* Create mount links as follows
* hdfs://localhost:xxx/HDFSUser --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/local --> file://TEST_ROOT_DIR/root/
* fallback --> hdfs://localhost:xxx/HDFSUser/
*
* Note: Above links created because to make fs initialization success.
* Otherwise will not proceed if no mount links.
*
* Don't set fs.viewfs.overload.scheme.target.hdfs.impl property.
* So, OverloadScheme target fs initialization will fail.
*/
@Test(expected = UnsupportedFileSystemException.class, timeout = 30000)
public void testInvalidOverloadSchemeTargetFS() throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
final Path localTragetPath = new Path(localTargetDir.toURI());
conf = new Configuration();
createLinks(true, hdfsTargetPath, localTragetPath);
conf.set(CommonConfigurationKeys.FS_DEFAULT_NAME_KEY,
defaultFSURI.toString());
conf.set(String.format(FS_IMPL_PATTERN_KEY, HDFS_SCHEME),
ViewFileSystemOverloadScheme.class.getName());
try (FileSystem fs = FileSystem.get(conf)) {
fs.createNewFile(new Path("/onRootWhenFallBack"));
Assert.fail("OverloadScheme target fs should be valid.");
}
}
/**
* Create mount links as follows
* hdfs://localhost:xxx/HDFSUser --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/local --> file://TEST_ROOT_DIR/root/
*
* It should be able to create file using ViewFileSystemOverloadScheme.
*/
@Test(timeout = 30000)
public void testViewFsOverloadSchemeWhenInnerCacheDisabled()
throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
final Path localTragetPath = new Path(localTargetDir.toURI());
createLinks(false, hdfsTargetPath, localTragetPath);
conf.setBoolean(Constants.CONFIG_VIEWFS_ENABLE_INNER_CACHE, false);
try (FileSystem fs = FileSystem.get(conf)) {
Path testFile = new Path(HDFS_USER_FOLDER + "/testFile");
fs.createNewFile(testFile);
Assert.assertTrue(fs.exists(testFile));
}
}
/**
* Create mount links as follows
* hdfs://localhost:xxx/HDFSUser0 --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/HDFSUser1 --> hdfs://localhost:xxx/HDFSUser/
*
* 1. With cache, only one hdfs child file system instance should be there.
* 2. Without cache, there should 2 hdfs instances.
*/
@Test(timeout = 30000)
public void testViewFsOverloadSchemeWithInnerCache()
throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), HDFS_USER_FOLDER + 0,
hdfsTargetPath.toUri());
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), HDFS_USER_FOLDER + 1,
hdfsTargetPath.toUri());
// 1. Only 1 hdfs child file system should be there with cache.
try (ViewFileSystemOverloadScheme vfs =
(ViewFileSystemOverloadScheme) FileSystem.get(conf)) {
Assert.assertEquals(1, vfs.getChildFileSystems().length);
}
// 2. Two hdfs file systems should be there if no cache.
conf.setBoolean(Constants.CONFIG_VIEWFS_ENABLE_INNER_CACHE, false);
try (ViewFileSystemOverloadScheme vfs =
(ViewFileSystemOverloadScheme) FileSystem.get(conf)) {
Assert.assertEquals(2, vfs.getChildFileSystems().length);
}
}
/**
* Create mount links as follows
* hdfs://localhost:xxx/HDFSUser0 --> hdfs://localhost:xxx/HDFSUser/
* hdfs://localhost:xxx/HDFSUser1 --> hdfs://localhost:xxx/HDFSUser/
*
* When InnerCache disabled, all matching ViewFileSystemOverloadScheme
* initialized scheme file systems would not use FileSystem cache.
*/
@Test(timeout = 3000)
public void testViewFsOverloadSchemeWithNoInnerCacheAndHdfsTargets()
throws Exception {
final Path hdfsTargetPath = new Path(defaultFSURI + HDFS_USER_FOLDER);
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), HDFS_USER_FOLDER + 0,
hdfsTargetPath.toUri());
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), HDFS_USER_FOLDER + 1,
hdfsTargetPath.toUri());
conf.setBoolean(Constants.CONFIG_VIEWFS_ENABLE_INNER_CACHE, false);
// Two hdfs file systems should be there if no cache.
try (ViewFileSystemOverloadScheme vfs =
(ViewFileSystemOverloadScheme) FileSystem.get(conf)) {
Assert.assertEquals(2, vfs.getChildFileSystems().length);
}
}
/**
* Create mount links as follows
* hdfs://localhost:xxx/local0 --> file://localPath/
* hdfs://localhost:xxx/local1 --> file://localPath/
*
* When InnerCache disabled, all non matching ViewFileSystemOverloadScheme
* initialized scheme file systems should continue to take advantage of
* FileSystem cache.
*/
@Test(timeout = 3000)
public void testViewFsOverloadSchemeWithNoInnerCacheAndLocalSchemeTargets()
throws Exception {
final Path localTragetPath = new Path(localTargetDir.toURI());
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), LOCAL_FOLDER + 0,
localTragetPath.toUri());
ConfigUtil.addLink(conf, defaultFSURI.getAuthority(), LOCAL_FOLDER + 1,
localTragetPath.toUri());
// Only one local file system should be there if no InnerCache, but fs
// cache should work.
conf.setBoolean(Constants.CONFIG_VIEWFS_ENABLE_INNER_CACHE, false);
try (ViewFileSystemOverloadScheme vfs =
(ViewFileSystemOverloadScheme) FileSystem.get(conf)) {
Assert.assertEquals(1, vfs.getChildFileSystems().length);
}
}
}