Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance. Project price only 1 $
You can buy this project and download/modify it how often you want.
// -*- mode:Java; tab-width:2; c-basic-offset:2; indent-tabs-mode:t -*-
/**
*
* Licensed 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.
*
*
* Implements the Hadoop FS interfaces to allow applications to store
* files in Ceph.
*/
package org.apache.hadoop.fs.ceph;
import java.io.IOException;
import java.io.FileNotFoundException;
import java.io.OutputStream;
import java.net.URI;
import java.net.InetAddress;
import java.util.EnumSet;
import java.lang.Math;
import java.util.ArrayList;
import java.util.Map;
import java.util.TreeMap;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.BlockLocation;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FSInputStream;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.FileUtil;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.permission.FsPermission;
import org.apache.hadoop.fs.FileAlreadyExistsException;
import org.apache.hadoop.util.Progressable;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.net.DNS;
import org.apache.hadoop.fs.FsStatus;
import com.ceph.fs.CephFileAlreadyExistsException;
import com.ceph.fs.CephNotDirectoryException;
import com.ceph.fs.CephMount;
import com.ceph.fs.CephStat;
import com.ceph.fs.CephStatVFS;
import com.ceph.crush.Bucket;
import com.ceph.fs.CephFileExtent;
/**
* Known Issues:
*
* 1. Per-file replication and block size are ignored.
*/
public class CephFileSystem extends FileSystem {
private static final Log LOG = LogFactory.getLog(CephFileSystem.class);
private URI uri;
private Path workingDir;
private CephFS ceph = null;
private static final int CEPH_STRIPE_COUNT = 1;
private TreeMap datapools = null;
/**
* Create a new CephFileSystem.
*/
public CephFileSystem() {
}
/**
* Create an absolute path using the working directory.
*/
private Path makeAbsolute(Path path) {
if (path.isAbsolute()) {
return path;
}
return new Path(workingDir, path);
}
public URI getUri() {
return uri;
}
@Override
public void initialize(URI uri, Configuration conf) throws IOException {
super.initialize(uri, conf);
if (ceph == null) {
ceph = new CephTalker(conf, LOG);
}
ceph.initialize(uri, conf);
setConf(conf);
this.uri = URI.create(uri.getScheme() + "://" + uri.getAuthority());
this.workingDir = getHomeDirectory();
}
/**
* Open a Ceph file and attach the file handle to an FSDataInputStream.
* @param path The file to open
* @param bufferSize Ceph does internal buffering; but you can buffer in
* the Java code too if you like.
* @return FSDataInputStream reading from the given path.
* @throws IOException if the path DNE or is a
* directory, or there is an error getting data to set up the FSDataInputStream.
*/
public FSDataInputStream open(Path path, int bufferSize) throws IOException {
path = makeAbsolute(path);
int fd = ceph.open(path, CephMount.O_RDONLY, 0);
/* get file size */
CephStat stat = new CephStat();
ceph.fstat(fd, stat);
CephInputStream istream = new CephInputStream(getConf(), ceph, fd,
stat.size, bufferSize);
return new FSDataInputStream(istream);
}
/**
* Close down the CephFileSystem. Runs the base-class close method
* and then kills the Ceph client itself.
*/
@Override
public void close() throws IOException {
super.close(); // this method does stuff, make sure it's run!
ceph.shutdown();
}
/**
* Get an FSDataOutputStream to append onto a file.
* @param path The File you want to append onto
* @param bufferSize Ceph does internal buffering but you can buffer in the Java code as well if you like.
* @param progress The Progressable to report progress to.
* Reporting is limited but exists.
* @return An FSDataOutputStream that connects to the file on Ceph.
* @throws IOException If the file cannot be found or appended to.
*/
public FSDataOutputStream append(Path path, int bufferSize,
Progressable progress) throws IOException {
path = makeAbsolute(path);
if (progress != null) {
progress.progress();
}
int fd = ceph.open(path, CephMount.O_WRONLY|CephMount.O_APPEND, 0);
if (progress != null) {
progress.progress();
}
CephOutputStream ostream = new CephOutputStream(getConf(), ceph, fd,
bufferSize);
return new FSDataOutputStream(ostream, statistics);
}
public Path getWorkingDirectory() {
return workingDir;
}
@Override
public void setWorkingDirectory(Path dir) {
workingDir = makeAbsolute(dir);
}
/**
* Create a directory and any nonexistent parents. Any portion
* of the directory tree can exist without error.
* @param path The directory path to create
* @param perms The permissions to apply to the created directories.
* @return true if successful, false otherwise
* @throws IOException if the path is a child of a file.
*/
@Override
public boolean mkdirs(Path path, FsPermission perms) throws IOException {
path = makeAbsolute(path);
boolean result = false;
try {
ceph.mkdirs(path, (int) perms.toShort());
result = true;
} catch (CephFileAlreadyExistsException e) {
result = true;
}
return result;
}
/**
* Create a directory and any nonexistent parents. Any portion
* of the directory tree can exist without error.
* Apply umask from conf
* @param f The directory path to create
* @return true if successful, false otherwise
* @throws IOException if the path is a child of a file.
*/
@Override
public boolean mkdirs(Path f) throws IOException {
return mkdirs(f, FsPermission.getDirDefault().applyUMask(FsPermission.getUMask(getConf())));
}
/**
* Get stat information on a file. This does not fill owner or group, as
* Ceph's support for these is a bit different than HDFS'.
* @param path The path to stat.
* @return FileStatus object containing the stat information.
* @throws FileNotFoundException if the path could not be resolved.
*/
public FileStatus getFileStatus(Path path) throws IOException {
path = makeAbsolute(path);
CephStat stat = new CephStat();
ceph.lstat(path, stat);
FileStatus status = new FileStatus(stat.size, stat.isDir(),
ceph.get_file_replication(path), stat.blksize, stat.m_time,
stat.a_time, new FsPermission((short) stat.mode),
System.getProperty("user.name"), null, path.makeQualified(this));
return status;
}
/**
* Get the FileStatus for each listing in a directory.
* @param path The directory to get listings from.
* @return FileStatus[] containing one FileStatus for each directory listing;
* null if path does not exist.
*/
public FileStatus[] listStatus(Path path) throws IOException {
path = makeAbsolute(path);
String[] dirlist = ceph.listdir(path);
if (dirlist != null) {
FileStatus[] status = new FileStatus[dirlist.length];
for (int i = 0; i < status.length; i++) {
status[i] = getFileStatus(new Path(path, dirlist[i]));
}
return status;
}
if (isFile(path))
return new FileStatus[] { getFileStatus(path) };
return null;
}
@Override
public void setPermission(Path path, FsPermission permission) throws IOException {
path = makeAbsolute(path);
ceph.chmod(path, permission.toShort());
}
@Override
public void setTimes(Path path, long mtime, long atime) throws IOException {
path = makeAbsolute(path);
CephStat stat = new CephStat();
int mask = 0;
if (mtime != -1) {
mask |= CephMount.SETATTR_MTIME;
stat.m_time = mtime;
}
if (atime != -1) {
mask |= CephMount.SETATTR_ATIME;
stat.a_time = atime;
}
ceph.setattr(path, stat, mask);
}
/**
* Get data pools from configuration.
*
* Package-private: used by unit tests
*/
String[] getConfiguredDataPools() {
String pool_list = getConf().get(
CephConfigKeys.CEPH_DATA_POOLS_KEY,
CephConfigKeys.CEPH_DATA_POOLS_DEFAULT);
if (pool_list != null)
return pool_list.split(",");
return new String[0];
}
/**
* Lookup pool size by name.
*
* Package-private: used by unit tests
*/
int getPoolReplication(String pool_name) throws IOException {
int pool_id = ceph.get_pool_id(pool_name);
return ceph.get_pool_replication(pool_id);
}
/**
* Select a data pool given the requested replication factor.
*/
private String selectDataPool(Path path, int repl_wanted) throws IOException {
/* map pool size -> pool name */
TreeMap pools = new TreeMap();
/*
* Start with a mapping for the default pool. An error here would indicate
* something bad, so we throw any exceptions. For configured pools we
* ignore some errors.
*/
int fd = ceph.__open(new Path("/"), CephMount.O_RDONLY, 0);
String pool_name = ceph.get_file_pool_name(fd);
ceph.close(fd);
int replication = getPoolReplication(pool_name);
pools.put(new Integer(replication), pool_name);
/*
* Insert extra data pools from configuration. Errors are logged (most
* likely a non-existant pool), and a configured pool will override the
* default pool.
*/
String[] conf_pools = getConfiguredDataPools();
for (String name : conf_pools) {
try {
replication = getPoolReplication(name);
pools.put(new Integer(replication), name);
} catch (IOException e) {
LOG.warn("Error looking up replication of pool: " + name + ", " + e);
}
}
/* Choose smallest entry >= target, or largest in map. */
Map.Entry entry = pools.ceilingEntry(new Integer(repl_wanted));
if (entry == null)
entry = pools.lastEntry();
/* should always contain default pool */
assert(entry != null);
replication = entry.getKey().intValue();
pool_name = entry.getValue();
/* log non-exact match cases */
if (replication != repl_wanted) {
LOG.info("selectDataPool path=" + path + " pool:repl=" +
pool_name + ":" + replication + " wanted=" + repl_wanted);
}
return pool_name;
}
/**
* Create a new file and open an FSDataOutputStream that's connected to it.
* @param path The file to create.
* @param permission The permissions to apply to the file.
* @param overwrite If true, overwrite any existing file with
* this name; otherwise don't.
* @param bufferSize Ceph does internal buffering, but you can buffer
* in the Java code too if you like.
* @param replication Replication factor. See documentation on the
* "ceph.data.pools" configuration option.
* @param blockSize Ignored by Ceph. You can set client-wide block sizes
* via the fs.ceph.blockSize param if you like.
* @param progress A Progressable to report back to.
* Reporting is limited but exists.
* @return An FSDataOutputStream pointing to the created file.
* @throws IOException if the path is an
* existing directory, or the path exists but overwrite is false, or there is a
* failure in attempting to open for append with Ceph.
*/
public FSDataOutputStream create(Path path, FsPermission permission,
boolean overwrite, int bufferSize, short replication, long blockSize,
Progressable progress) throws IOException {
path = makeAbsolute(path);
boolean exists = exists(path);
if (progress != null) {
progress.progress();
}
int flags = CephMount.O_WRONLY | CephMount.O_CREAT;
if (exists) {
if (overwrite)
flags |= CephMount.O_TRUNC;
else
throw new FileAlreadyExistsException();
} else {
Path parent = path.getParent();
if (parent != null)
if (!mkdirs(parent, permission))
throw new IOException("mkdirs failed for " + parent.toString());
}
if (progress != null) {
progress.progress();
}
/* Sanity check. Ceph interface uses int for striping strategy */
if (blockSize > Integer.MAX_VALUE) {
blockSize = Integer.MAX_VALUE;
LOG.info("blockSize too large. Rounding down to " + blockSize);
}
/*
* If blockSize <= 0 then we complain. We need to explicitly check for the
* < 0 case (as opposed to allowing Ceph to raise an exception) because
* the ceph_open_layout interface accepts -1 to request Ceph-specific
* defaults.
*/
if (blockSize <= 0)
throw new IllegalArgumentException("Invalid block size: " + blockSize);
/*
* Ceph may impose alignment restrictions on file layout. In this case we
* check if the requested block size is aligned to the granularity of a
* stripe unit used in the file system. When the block size is not aligned
* we automatically adjust to the next largest multiple of stripe unit
* granularity.
*/
int su = ceph.get_stripe_unit_granularity();
if (blockSize % su != 0) {
long newBlockSize = blockSize - (blockSize % su) + su;
LOG.debug("fix alignment: blksize " + blockSize + " new blksize " + newBlockSize);
blockSize = newBlockSize;
}
/*
* The default Ceph data pool is selected to store files unless a specific
* data pool is provided when a file is created. Since a pool has a fixed
* replication factor, in order to achieve a requested replication factor,
* we must select an appropriate data pool to place the file into.
*/
String datapool = selectDataPool(path, replication);
int fd = ceph.open(path, flags, (int)permission.toShort(), (int)blockSize,
CEPH_STRIPE_COUNT, (int)blockSize, datapool);
if (progress != null) {
progress.progress();
}
OutputStream ostream = new CephOutputStream(getConf(), ceph, fd,
bufferSize);
return new FSDataOutputStream(ostream, statistics);
}
/**
* Opens an FSDataOutputStream at the indicated Path with write-progress
* reporting. Same as create(), except fails if parent directory doesn't
* already exist.
* @param path the file name to open
* @param permission
* @param overwrite if a file with this name already exists, then if true,
* the file will be overwritten, and if false an error will be thrown.
* @param bufferSize the size of the buffer to be used.
* @param replication required block replication for the file.
* @param blockSize
* @param progress
* @throws IOException
* @see #setPermission(Path, FsPermission)
* @deprecated API only for 0.20-append
*/
@Deprecated
public FSDataOutputStream createNonRecursive(Path path, FsPermission permission,
boolean overwrite,
int bufferSize, short replication, long blockSize,
Progressable progress) throws IOException {
path = makeAbsolute(path);
Path parent = path.getParent();
if (parent != null) {
CephStat stat = new CephStat();
ceph.lstat(parent, stat); // handles FileNotFoundException case
if (stat.isFile())
throw new FileAlreadyExistsException(parent.toString());
}
return this.create(path, permission, overwrite,
bufferSize, replication, blockSize, progress);
}
/**
* Rename a file or directory.
* @param src The current path of the file/directory
* @param dst The new name for the path.
* @return true if the rename succeeded, false otherwise.
*/
@Override
public boolean rename(Path src, Path dst) throws IOException {
src = makeAbsolute(src);
dst = makeAbsolute(dst);
try {
CephStat stat = new CephStat();
ceph.lstat(dst, stat);
if (stat.isDir())
return rename(src, new Path(dst, src.getName()));
return false;
} catch (FileNotFoundException e) {}
try {
ceph.rename(src, dst);
} catch (FileNotFoundException e) {
return false;
}
return true;
}
/**
* Get a BlockLocation object for each block in a file.
*
* @param file A FileStatus object corresponding to the file you want locations for.
* @param start The offset of the first part of the file you are interested in.
* @param len The amount of the file past the offset you are interested in.
* @return A BlockLocation[] where each object corresponds to a block within
* the given range.
*/
@Override
public BlockLocation[] getFileBlockLocations(FileStatus file, long start, long len) throws IOException {
Path abs_path = makeAbsolute(file.getPath());
int fh = ceph.open(abs_path, CephMount.O_RDONLY, 0);
if (fh < 0) {
LOG.error("getFileBlockLocations:got error " + fh + ", exiting and returning null!");
return null;
}
ArrayList blocks = new ArrayList();
long curPos = start;
long endOff = curPos + len;
do {
CephFileExtent extent = ceph.get_file_extent(fh, curPos);
int[] osds = extent.getOSDs();
String[] names = new String[osds.length];
String[] hosts = new String[osds.length];
String[] racks = new String[osds.length];
for (int i = 0; i < osds.length; i++) {
InetAddress addr = ceph.get_osd_address(osds[i]);
names[i] = addr.getHostAddress();
/*
* Grab the hostname and rack from the crush hierarchy. Current we
* hard code the item types. For a more general treatment, we'll need
* a new configuration option that allows users to map their custom
* crush types to hosts and topology.
*/
Bucket[] path = ceph.get_osd_crush_location(osds[i]);
for (Bucket bucket : path) {
String type = bucket.getType();
if (type.compareTo("host") == 0)
hosts[i] = bucket.getName();
else if (type.compareTo("rack") == 0)
racks[i] = bucket.getName();
}
}
blocks.add(new BlockLocation(names, hosts, racks,
extent.getOffset(), extent.getLength()));
curPos += extent.getLength();
} while(curPos < endOff);
ceph.close(fh);
BlockLocation[] locations = new BlockLocation[blocks.size()];
locations = blocks.toArray(locations);
return locations;
}
@Deprecated
public boolean delete(Path path) throws IOException {
return delete(path, false);
}
public boolean delete(Path path, boolean recursive) throws IOException {
path = makeAbsolute(path);
/* path exists? */
FileStatus status;
try {
status = getFileStatus(path);
} catch (FileNotFoundException e) {
return false;
}
/* we're done if its a file */
if (!status.isDir()) {
ceph.unlink(path);
return true;
}
/* get directory contents */
FileStatus[] dirlist = listStatus(path);
if (dirlist == null)
return false;
if (!recursive && dirlist.length > 0)
throw new IOException("Directory " + path.toString() + "is not empty.");
for (FileStatus fs : dirlist) {
if (!delete(fs.getPath(), recursive))
return false;
}
ceph.rmdir(path);
return true;
}
@Override
public short getDefaultReplication() {
return ceph.getDefaultReplication();
}
@Override
public long getDefaultBlockSize() {
return getConf().getLong(
CephConfigKeys.CEPH_OBJECT_SIZE_KEY,
CephConfigKeys.CEPH_OBJECT_SIZE_DEFAULT);
}
@Override
public FsStatus getStatus(Path p) throws IOException {
CephStatVFS stat = new CephStatVFS();
ceph.statfs(p, stat);
FsStatus status = new FsStatus(stat.bsize * stat.blocks,
stat.bsize * (stat.blocks - stat.bavail),
stat.bsize * stat.bavail);
return status;
}
}
