blob: 46cec4af1f8e74703a1be39cf4eedfa6aff675ee [file] [log] [blame]
/*
*
* 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.hbase.io.hfile;
import java.io.Closeable;
import java.io.DataInput;
import java.io.IOException;
import java.net.InetSocketAddress;
import java.util.ArrayList;
import java.util.List;
import java.util.Optional;
import java.util.concurrent.atomic.LongAdder;
import org.apache.commons.io.IOUtils;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.PathFilter;
import org.apache.hadoop.hbase.Cell;
import org.apache.hadoop.hbase.CellComparator;
import org.apache.hadoop.hbase.HConstants;
import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper;
import org.apache.hadoop.hbase.io.MetricsIO;
import org.apache.hadoop.hbase.io.MetricsIOWrapperImpl;
import org.apache.hadoop.hbase.io.compress.Compression;
import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;
import org.apache.hadoop.hbase.io.hfile.ReaderContext.ReaderType;
import org.apache.hadoop.hbase.regionserver.CellSink;
import org.apache.hadoop.hbase.regionserver.ShipperListener;
import org.apache.hadoop.hbase.util.BloomFilterWriter;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.FSUtils;
import org.apache.hadoop.io.Writable;
import org.apache.yetus.audience.InterfaceAudience;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;
import org.apache.hbase.thirdparty.com.google.common.base.Preconditions;
/**
* File format for hbase.
* A file of sorted key/value pairs. Both keys and values are byte arrays.
* <p>
* The memory footprint of a HFile includes the following (below is taken from the
* <a
* href=https://issues.apache.org/jira/browse/HADOOP-3315>TFile</a> documentation
* but applies also to HFile):
* <ul>
* <li>Some constant overhead of reading or writing a compressed block.
* <ul>
* <li>Each compressed block requires one compression/decompression codec for
* I/O.
* <li>Temporary space to buffer the key.
* <li>Temporary space to buffer the value.
* </ul>
* <li>HFile index, which is proportional to the total number of Data Blocks.
* The total amount of memory needed to hold the index can be estimated as
* (56+AvgKeySize)*NumBlocks.
* </ul>
* Suggestions on performance optimization.
* <ul>
* <li>Minimum block size. We recommend a setting of minimum block size between
* 8KB to 1MB for general usage. Larger block size is preferred if files are
* primarily for sequential access. However, it would lead to inefficient random
* access (because there are more data to decompress). Smaller blocks are good
* for random access, but require more memory to hold the block index, and may
* be slower to create (because we must flush the compressor stream at the
* conclusion of each data block, which leads to an FS I/O flush). Further, due
* to the internal caching in Compression codec, the smallest possible block
* size would be around 20KB-30KB.
* <li>The current implementation does not offer true multi-threading for
* reading. The implementation uses FSDataInputStream seek()+read(), which is
* shown to be much faster than positioned-read call in single thread mode.
* However, it also means that if multiple threads attempt to access the same
* HFile (using multiple scanners) simultaneously, the actual I/O is carried out
* sequentially even if they access different DFS blocks (Reexamine! pread seems
* to be 10% faster than seek+read in my testing -- stack).
* <li>Compression codec. Use "none" if the data is not very compressable (by
* compressable, I mean a compression ratio at least 2:1). Generally, use "lzo"
* as the starting point for experimenting. "gz" overs slightly better
* compression ratio over "lzo" but requires 4x CPU to compress and 2x CPU to
* decompress, comparing to "lzo".
* </ul>
*
* For more on the background behind HFile, see <a
* href=https://issues.apache.org/jira/browse/HBASE-61>HBASE-61</a>.
* <p>
* File is made of data blocks followed by meta data blocks (if any), a fileinfo
* block, data block index, meta data block index, and a fixed size trailer
* which records the offsets at which file changes content type.
* <pre>&lt;data blocks&gt;&lt;meta blocks&gt;&lt;fileinfo&gt;&lt;
* data index&gt;&lt;meta index&gt;&lt;trailer&gt;</pre>
* Each block has a bit of magic at its start. Block are comprised of
* key/values. In data blocks, they are both byte arrays. Metadata blocks are
* a String key and a byte array value. An empty file looks like this:
* <pre>&lt;fileinfo&gt;&lt;trailer&gt;</pre>. That is, there are not data nor meta
* blocks present.
* <p>
* TODO: Do scanners need to be able to take a start and end row?
* TODO: Should BlockIndex know the name of its file? Should it have a Path
* that points at its file say for the case where an index lives apart from
* an HFile instance?
*/
@InterfaceAudience.Private
public final class HFile {
// LOG is being used in HFileBlock and CheckSumUtil
static final Logger LOG = LoggerFactory.getLogger(HFile.class);
/**
* Maximum length of key in HFile.
*/
public final static int MAXIMUM_KEY_LENGTH = Integer.MAX_VALUE;
/**
* Default compression: none.
*/
public final static Compression.Algorithm DEFAULT_COMPRESSION_ALGORITHM =
Compression.Algorithm.NONE;
/** Minimum supported HFile format version */
public static final int MIN_FORMAT_VERSION = 2;
/** Maximum supported HFile format version
*/
public static final int MAX_FORMAT_VERSION = 3;
/**
* Minimum HFile format version with support for persisting cell tags
*/
public static final int MIN_FORMAT_VERSION_WITH_TAGS = 3;
/** Default compression name: none. */
public final static String DEFAULT_COMPRESSION =
DEFAULT_COMPRESSION_ALGORITHM.getName();
/** Meta data block name for bloom filter bits. */
public static final String BLOOM_FILTER_DATA_KEY = "BLOOM_FILTER_DATA";
/**
* We assume that HFile path ends with
* ROOT_DIR/TABLE_NAME/REGION_NAME/CF_NAME/HFILE, so it has at least this
* many levels of nesting. This is needed for identifying table and CF name
* from an HFile path.
*/
public final static int MIN_NUM_HFILE_PATH_LEVELS = 5;
/**
* The number of bytes per checksum.
*/
public static final int DEFAULT_BYTES_PER_CHECKSUM = 16 * 1024;
// For measuring number of checksum failures
static final LongAdder CHECKSUM_FAILURES = new LongAdder();
// For tests. Gets incremented when we read a block whether from HDFS or from Cache.
public static final LongAdder DATABLOCK_READ_COUNT = new LongAdder();
/** Static instance for the metrics so that HFileReaders access the same instance */
static final MetricsIO metrics = new MetricsIO(new MetricsIOWrapperImpl());
/**
* Shutdown constructor.
*/
private HFile() {}
/**
* Number of checksum verification failures. It also
* clears the counter.
*/
public static final long getAndResetChecksumFailuresCount() {
return CHECKSUM_FAILURES.sumThenReset();
}
/**
* Number of checksum verification failures. It also
* clears the counter.
*/
public static final long getChecksumFailuresCount() {
return CHECKSUM_FAILURES.sum();
}
public static final void updateReadLatency(long latencyMillis, boolean pread) {
if (pread) {
metrics.updateFsPreadTime(latencyMillis);
} else {
metrics.updateFsReadTime(latencyMillis);
}
}
public static final void updateWriteLatency(long latencyMillis) {
metrics.updateFsWriteTime(latencyMillis);
}
/** API required to write an {@link HFile} */
public interface Writer extends Closeable, CellSink, ShipperListener {
/** Max memstore (mvcc) timestamp in FileInfo */
public static final byte [] MAX_MEMSTORE_TS_KEY = Bytes.toBytes("MAX_MEMSTORE_TS_KEY");
/** Add an element to the file info map. */
void appendFileInfo(byte[] key, byte[] value) throws IOException;
/** @return the path to this {@link HFile} */
Path getPath();
/**
* Adds an inline block writer such as a multi-level block index writer or
* a compound Bloom filter writer.
*/
void addInlineBlockWriter(InlineBlockWriter bloomWriter);
// The below three methods take Writables. We'd like to undo Writables but undoing the below
// would be pretty painful. Could take a byte [] or a Message but we want to be backward
// compatible around hfiles so would need to map between Message and Writable or byte [] and
// current Writable serialization. This would be a bit of work to little gain. Thats my
// thinking at moment. St.Ack 20121129
void appendMetaBlock(String bloomFilterMetaKey, Writable metaWriter);
/**
* Store general Bloom filter in the file. This does not deal with Bloom filter
* internals but is necessary, since Bloom filters are stored differently
* in HFile version 1 and version 2.
*/
void addGeneralBloomFilter(BloomFilterWriter bfw);
/**
* Store delete family Bloom filter in the file, which is only supported in
* HFile V2.
*/
void addDeleteFamilyBloomFilter(BloomFilterWriter bfw) throws IOException;
/**
* Return the file context for the HFile this writer belongs to
*/
HFileContext getFileContext();
}
/**
* This variety of ways to construct writers is used throughout the code, and
* we want to be able to swap writer implementations.
*/
public static class WriterFactory {
protected final Configuration conf;
protected final CacheConfig cacheConf;
protected FileSystem fs;
protected Path path;
protected FSDataOutputStream ostream;
protected InetSocketAddress[] favoredNodes;
private HFileContext fileContext;
protected boolean shouldDropBehind = false;
WriterFactory(Configuration conf, CacheConfig cacheConf) {
this.conf = conf;
this.cacheConf = cacheConf;
}
public WriterFactory withPath(FileSystem fs, Path path) {
Preconditions.checkNotNull(fs);
Preconditions.checkNotNull(path);
this.fs = fs;
this.path = path;
return this;
}
public WriterFactory withOutputStream(FSDataOutputStream ostream) {
Preconditions.checkNotNull(ostream);
this.ostream = ostream;
return this;
}
public WriterFactory withFavoredNodes(InetSocketAddress[] favoredNodes) {
// Deliberately not checking for null here.
this.favoredNodes = favoredNodes;
return this;
}
public WriterFactory withFileContext(HFileContext fileContext) {
this.fileContext = fileContext;
return this;
}
public WriterFactory withShouldDropCacheBehind(boolean shouldDropBehind) {
this.shouldDropBehind = shouldDropBehind;
return this;
}
public Writer create() throws IOException {
if ((path != null ? 1 : 0) + (ostream != null ? 1 : 0) != 1) {
throw new AssertionError("Please specify exactly one of " +
"filesystem/path or path");
}
if (path != null) {
ostream = HFileWriterImpl.createOutputStream(conf, fs, path, favoredNodes);
try {
ostream.setDropBehind(shouldDropBehind && cacheConf.shouldDropBehindCompaction());
} catch (UnsupportedOperationException uoe) {
LOG.trace("Unable to set drop behind on {}", path, uoe);
LOG.debug("Unable to set drop behind on {}", path.getName());
}
}
return new HFileWriterImpl(conf, cacheConf, path, ostream, fileContext);
}
}
/** The configuration key for HFile version to use for new files */
public static final String FORMAT_VERSION_KEY = "hfile.format.version";
public static int getFormatVersion(Configuration conf) {
int version = conf.getInt(FORMAT_VERSION_KEY, MAX_FORMAT_VERSION);
checkFormatVersion(version);
return version;
}
/**
* Returns the factory to be used to create {@link HFile} writers.
* Disables block cache access for all writers created through the
* returned factory.
*/
public static final WriterFactory getWriterFactoryNoCache(Configuration
conf) {
return HFile.getWriterFactory(conf, CacheConfig.DISABLED);
}
/**
* Returns the factory to be used to create {@link HFile} writers
*/
public static final WriterFactory getWriterFactory(Configuration conf,
CacheConfig cacheConf) {
int version = getFormatVersion(conf);
switch (version) {
case 2:
throw new IllegalArgumentException("This should never happen. " +
"Did you change hfile.format.version to read v2? This version of the software writes v3" +
" hfiles only (but it can read v2 files without having to update hfile.format.version " +
"in hbase-site.xml)");
case 3:
return new HFile.WriterFactory(conf, cacheConf);
default:
throw new IllegalArgumentException("Cannot create writer for HFile " +
"format version " + version);
}
}
/**
* An abstraction used by the block index.
* Implementations will check cache for any asked-for block and return cached block if found.
* Otherwise, after reading from fs, will try and put block into cache before returning.
*/
public interface CachingBlockReader {
/**
* Read in a file block.
* @param offset offset to read.
* @param onDiskBlockSize size of the block
* @param isCompaction is this block being read as part of a compaction
* @param expectedBlockType the block type we are expecting to read with this read operation,
* or null to read whatever block type is available and avoid checking (that might reduce
* caching efficiency of encoded data blocks)
* @param expectedDataBlockEncoding the data block encoding the caller is expecting data blocks
* to be in, or null to not perform this check and return the block irrespective of the
* encoding. This check only applies to data blocks and can be set to null when the caller is
* expecting to read a non-data block and has set expectedBlockType accordingly.
* @return Block wrapped in a ByteBuffer.
*/
HFileBlock readBlock(long offset, long onDiskBlockSize,
boolean cacheBlock, final boolean pread, final boolean isCompaction,
final boolean updateCacheMetrics, BlockType expectedBlockType,
DataBlockEncoding expectedDataBlockEncoding)
throws IOException;
}
/** An interface used by clients to open and iterate an {@link HFile}. */
public interface Reader extends Closeable, CachingBlockReader {
/**
* Returns this reader's "name". Usually the last component of the path.
* Needs to be constant as the file is being moved to support caching on
* write.
*/
String getName();
CellComparator getComparator();
HFileScanner getScanner(boolean cacheBlocks, final boolean pread, final boolean isCompaction);
HFileBlock getMetaBlock(String metaBlockName, boolean cacheBlock) throws IOException;
Optional<Cell> getLastKey();
Optional<Cell> midKey() throws IOException;
long length();
long getEntries();
Optional<Cell> getFirstKey();
long indexSize();
Optional<byte[]> getFirstRowKey();
Optional<byte[]> getLastRowKey();
FixedFileTrailer getTrailer();
void setDataBlockIndexReader(HFileBlockIndex.CellBasedKeyBlockIndexReader reader);
HFileBlockIndex.CellBasedKeyBlockIndexReader getDataBlockIndexReader();
void setMetaBlockIndexReader(HFileBlockIndex.ByteArrayKeyBlockIndexReader reader);
HFileBlockIndex.ByteArrayKeyBlockIndexReader getMetaBlockIndexReader();
HFileScanner getScanner(boolean cacheBlocks, boolean pread);
/**
* Retrieves general Bloom filter metadata as appropriate for each
* {@link HFile} version.
* Knows nothing about how that metadata is structured.
*/
DataInput getGeneralBloomFilterMetadata() throws IOException;
/**
* Retrieves delete family Bloom filter metadata as appropriate for each
* {@link HFile} version.
* Knows nothing about how that metadata is structured.
*/
DataInput getDeleteBloomFilterMetadata() throws IOException;
Path getPath();
/** Close method with optional evictOnClose */
void close(boolean evictOnClose) throws IOException;
DataBlockEncoding getDataBlockEncoding();
boolean hasMVCCInfo();
/**
* Return the file context of the HFile this reader belongs to
*/
HFileContext getFileContext();
boolean isPrimaryReplicaReader();
DataBlockEncoding getEffectiveEncodingInCache(boolean isCompaction);
@VisibleForTesting
HFileBlock.FSReader getUncachedBlockReader();
@VisibleForTesting
boolean prefetchComplete();
/**
* To close the stream's socket. Note: This can be concurrently called from multiple threads and
* implementation should take care of thread safety.
*/
void unbufferStream();
ReaderContext getContext();
HFileInfo getHFileInfo();
void setDataBlockEncoder(HFileDataBlockEncoder dataBlockEncoder);
}
/**
* Method returns the reader given the specified arguments.
* TODO This is a bad abstraction. See HBASE-6635.
*
* @param context Reader context info
* @param fileInfo HFile info
* @param cacheConf Cache configuation values, cannot be null.
* @param conf Configuration
* @return an appropriate instance of HFileReader
* @throws IOException If file is invalid, will throw CorruptHFileException flavored IOException
*/
@edu.umd.cs.findbugs.annotations.SuppressWarnings(value="SF_SWITCH_FALLTHROUGH",
justification="Intentional")
public static Reader createReader(ReaderContext context, HFileInfo fileInfo,
CacheConfig cacheConf, Configuration conf) throws IOException {
try {
if (context.getReaderType() == ReaderType.STREAM) {
// stream reader will share trailer with pread reader, see HFileStreamReader#copyFields
return new HFileStreamReader(context, fileInfo, cacheConf, conf);
}
FixedFileTrailer trailer = fileInfo.getTrailer();
switch (trailer.getMajorVersion()) {
case 2:
LOG.debug("Opening HFile v2 with v3 reader");
// Fall through. FindBugs: SF_SWITCH_FALLTHROUGH
case 3:
return new HFilePreadReader(context, fileInfo, cacheConf, conf);
default:
throw new IllegalArgumentException("Invalid HFile version " + trailer.getMajorVersion());
}
} catch (Throwable t) {
IOUtils.closeQuietly(context.getInputStreamWrapper());
throw new CorruptHFileException("Problem reading HFile Trailer from file "
+ context.getFilePath(), t);
} finally {
context.getInputStreamWrapper().unbuffer();
}
}
/**
* Creates reader with cache configuration disabled
* @param fs filesystem
* @param path Path to file to read
* @param conf Configuration
* @return an active Reader instance
* @throws IOException Will throw a CorruptHFileException
* (DoNotRetryIOException subtype) if hfile is corrupt/invalid.
*/
public static Reader createReader(FileSystem fs, Path path, Configuration conf)
throws IOException {
// The primaryReplicaReader is mainly used for constructing block cache key, so if we do not use
// block cache then it is OK to set it as any value. We use true here.
return createReader(fs, path, CacheConfig.DISABLED, true, conf);
}
/**
* @param fs filesystem
* @param path Path to file to read
* @param cacheConf This must not be null. @see
* {@link org.apache.hadoop.hbase.io.hfile.CacheConfig#CacheConfig(Configuration)}
* @param primaryReplicaReader true if this is a reader for primary replica
* @param conf Configuration
* @return an active Reader instance
* @throws IOException Will throw a CorruptHFileException (DoNotRetryIOException subtype) if hfile
* is corrupt/invalid.
*/
public static Reader createReader(FileSystem fs, Path path, CacheConfig cacheConf,
boolean primaryReplicaReader, Configuration conf) throws IOException {
Preconditions.checkNotNull(cacheConf, "Cannot create Reader with null CacheConf");
FSDataInputStreamWrapper stream = new FSDataInputStreamWrapper(fs, path);
ReaderContext context = new ReaderContextBuilder()
.withFilePath(path)
.withInputStreamWrapper(stream)
.withFileSize(fs.getFileStatus(path).getLen())
.withFileSystem(stream.getHfs())
.withPrimaryReplicaReader(primaryReplicaReader)
.withReaderType(ReaderType.PREAD)
.build();
HFileInfo fileInfo = new HFileInfo(context, conf);
Reader reader = createReader(context, fileInfo, cacheConf, conf);
fileInfo.initMetaAndIndex(reader);
return reader;
}
/**
* Returns true if the specified file has a valid HFile Trailer.
* @param fs filesystem
* @param path Path to file to verify
* @return true if the file has a valid HFile Trailer, otherwise false
* @throws IOException if failed to read from the underlying stream
*/
public static boolean isHFileFormat(final FileSystem fs, final Path path) throws IOException {
return isHFileFormat(fs, fs.getFileStatus(path));
}
/**
* Returns true if the specified file has a valid HFile Trailer.
* @param fs filesystem
* @param fileStatus the file to verify
* @return true if the file has a valid HFile Trailer, otherwise false
* @throws IOException if failed to read from the underlying stream
*/
public static boolean isHFileFormat(final FileSystem fs, final FileStatus fileStatus)
throws IOException {
final Path path = fileStatus.getPath();
final long size = fileStatus.getLen();
try (FSDataInputStreamWrapper fsdis = new FSDataInputStreamWrapper(fs, path)) {
boolean isHBaseChecksum = fsdis.shouldUseHBaseChecksum();
assert !isHBaseChecksum; // Initially we must read with FS checksum.
FixedFileTrailer.readFromStream(fsdis.getStream(isHBaseChecksum), size);
return true;
} catch (IllegalArgumentException e) {
return false;
}
}
/**
* Get names of supported compression algorithms. The names are acceptable by
* HFile.Writer.
*
* @return Array of strings, each represents a supported compression
* algorithm. Currently, the following compression algorithms are
* supported.
* <ul>
* <li>"none" - No compression.
* <li>"gz" - GZIP compression.
* </ul>
*/
public static String[] getSupportedCompressionAlgorithms() {
return Compression.getSupportedAlgorithms();
}
// Utility methods.
/*
* @param l Long to convert to an int.
* @return <code>l</code> cast as an int.
*/
static int longToInt(final long l) {
// Expecting the size() of a block not exceeding 4GB. Assuming the
// size() will wrap to negative integer if it exceeds 2GB (From tfile).
return (int)(l & 0x00000000ffffffffL);
}
/**
* Returns all HFiles belonging to the given region directory. Could return an
* empty list.
*
* @param fs The file system reference.
* @param regionDir The region directory to scan.
* @return The list of files found.
* @throws IOException When scanning the files fails.
*/
public static List<Path> getStoreFiles(FileSystem fs, Path regionDir)
throws IOException {
List<Path> regionHFiles = new ArrayList<>();
PathFilter dirFilter = new FSUtils.DirFilter(fs);
FileStatus[] familyDirs = fs.listStatus(regionDir, dirFilter);
for(FileStatus dir : familyDirs) {
FileStatus[] files = fs.listStatus(dir.getPath());
for (FileStatus file : files) {
if (!file.isDirectory() &&
(!file.getPath().toString().contains(HConstants.HREGION_OLDLOGDIR_NAME)) &&
(!file.getPath().toString().contains(HConstants.RECOVERED_EDITS_DIR))) {
regionHFiles.add(file.getPath());
}
}
}
return regionHFiles;
}
/**
* Checks the given {@link HFile} format version, and throws an exception if
* invalid. Note that if the version number comes from an input file and has
* not been verified, the caller needs to re-throw an {@link IOException} to
* indicate that this is not a software error, but corrupted input.
*
* @param version an HFile version
* @throws IllegalArgumentException if the version is invalid
*/
public static void checkFormatVersion(int version)
throws IllegalArgumentException {
if (version < MIN_FORMAT_VERSION || version > MAX_FORMAT_VERSION) {
throw new IllegalArgumentException("Invalid HFile version: " + version
+ " (expected to be " + "between " + MIN_FORMAT_VERSION + " and "
+ MAX_FORMAT_VERSION + ")");
}
}
public static void checkHFileVersion(final Configuration c) {
int version = c.getInt(FORMAT_VERSION_KEY, MAX_FORMAT_VERSION);
if (version < MAX_FORMAT_VERSION || version > MAX_FORMAT_VERSION) {
throw new IllegalArgumentException("The setting for " + FORMAT_VERSION_KEY +
" (in your hbase-*.xml files) is " + version + " which does not match " +
MAX_FORMAT_VERSION +
"; are you running with a configuration from an older or newer hbase install (an " +
"incompatible hbase-default.xml or hbase-site.xml on your CLASSPATH)?");
}
}
public static void main(String[] args) throws Exception {
// delegate to preserve old behavior
HFilePrettyPrinter.main(args);
}
}