org.apache.cassandra.io.util.SequentialWriter Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of cassandra-all Show documentation
Show all versions of cassandra-all Show documentation
The Apache Cassandra Project develops a highly scalable second-generation distributed database, bringing together Dynamo's fully distributed design and Bigtable's ColumnFamily-based data model.
/*
* 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.cassandra.io.util;
import java.io.File;
import java.io.IOException;
import java.nio.ByteBuffer;
import java.nio.channels.FileChannel;
import java.nio.file.StandardOpenOption;
import org.apache.cassandra.io.FSReadError;
import org.apache.cassandra.io.FSWriteError;
import org.apache.cassandra.utils.SyncUtil;
import org.apache.cassandra.utils.concurrent.Transactional;
import static org.apache.cassandra.utils.Throwables.merge;
/**
* Adds buffering, mark, and fsyncing to OutputStream. We always fsync on close; we may also
* fsync incrementally if Config.trickle_fsync is enabled.
*/
public class SequentialWriter extends BufferedDataOutputStreamPlus implements Transactional
{
// absolute path to the given file
private final String filePath;
// Offset for start of buffer relative to underlying file
protected long bufferOffset;
protected final FileChannel fchannel;
//Allow derived classes to specify writing to the channel
//directly shouldn't happen because they intercept via doFlush for things
//like compression or checksumming
//Another hack for this value is that it also indicates that flushing early
//should not occur, flushes aligned with buffer size are desired
//Unless... it's the last flush. Compression and checksum formats
//expect block (same as buffer size) alignment for everything except the last block
private final boolean strictFlushing;
// whether to do trickling fsync() to avoid sudden bursts of dirty buffer flushing by kernel causing read
// latency spikes
private final SequentialWriterOption option;
private int bytesSinceTrickleFsync = 0;
protected long lastFlushOffset;
protected Runnable runPostFlush;
private final TransactionalProxy txnProxy = txnProxy();
// due to lack of multiple-inheritance, we proxy our transactional implementation
protected class TransactionalProxy extends AbstractTransactional
{
@Override
protected Throwable doPreCleanup(Throwable accumulate)
{
// close is idempotent
try { channel.close(); }
catch (Throwable t) { accumulate = merge(accumulate, t); }
if (buffer != null)
{
try { FileUtils.clean(buffer); }
catch (Throwable t) { accumulate = merge(accumulate, t); }
buffer = null;
}
return accumulate;
}
protected void doPrepare()
{
syncInternal();
}
protected Throwable doCommit(Throwable accumulate)
{
return accumulate;
}
protected Throwable doAbort(Throwable accumulate)
{
return accumulate;
}
}
// TODO: we should specify as a parameter if we permit an existing file or not
private static FileChannel openChannel(File file)
{
try
{
if (file.exists())
{
return FileChannel.open(file.toPath(), StandardOpenOption.READ, StandardOpenOption.WRITE);
}
else
{
FileChannel channel = FileChannel.open(file.toPath(), StandardOpenOption.READ, StandardOpenOption.WRITE, StandardOpenOption.CREATE_NEW);
try
{
SyncUtil.trySyncDir(file.getParentFile());
}
catch (Throwable t)
{
try { channel.close(); }
catch (Throwable t2) { t.addSuppressed(t2); }
}
return channel;
}
}
catch (IOException e)
{
throw new RuntimeException(e);
}
}
/**
* Create heap-based, non-compressed SequenialWriter with default buffer size(64k).
*
* @param file File to write
*/
public SequentialWriter(File file)
{
this(file, SequentialWriterOption.DEFAULT);
}
/**
* Create SequentialWriter for given file with specific writer option.
*
* @param file File to write
* @param option Writer option
*/
public SequentialWriter(File file, SequentialWriterOption option)
{
this(file, option, true);
}
/**
* Create SequentialWriter for given file with specific writer option.
* @param file
* @param option
* @param strictFlushing
*/
public SequentialWriter(File file, SequentialWriterOption option, boolean strictFlushing)
{
super(openChannel(file), option.allocateBuffer());
this.strictFlushing = strictFlushing;
this.fchannel = (FileChannel)channel;
this.filePath = file.getAbsolutePath();
this.option = option;
}
public void skipBytes(int numBytes) throws IOException
{
flush();
fchannel.position(fchannel.position() + numBytes);
bufferOffset = fchannel.position();
}
/**
* Synchronize file contents with disk.
*/
public void sync()
{
syncInternal();
}
protected void syncDataOnlyInternal()
{
try
{
SyncUtil.force(fchannel, false);
}
catch (IOException e)
{
throw new FSWriteError(e, getPath());
}
}
/*
* This is only safe to call before truncation or close for CompressedSequentialWriter
* Otherwise it will leave a non-uniform size compressed block in the middle of the file
* and the compressed format can't handle that.
*/
protected void syncInternal()
{
doFlush(0);
syncDataOnlyInternal();
}
@Override
protected void doFlush(int count)
{
flushData();
if (option.trickleFsync())
{
bytesSinceTrickleFsync += buffer.position();
if (bytesSinceTrickleFsync >= option.trickleFsyncByteInterval())
{
syncDataOnlyInternal();
bytesSinceTrickleFsync = 0;
}
}
// Remember that we wrote, so we don't write it again on next flush().
resetBuffer();
}
public void setPostFlushListener(Runnable runPostFlush)
{
assert this.runPostFlush == null;
this.runPostFlush = runPostFlush;
}
/**
* Override this method instead of overriding flush()
* @throws FSWriteError on any I/O error.
*/
protected void flushData()
{
try
{
buffer.flip();
channel.write(buffer);
lastFlushOffset += buffer.position();
}
catch (IOException e)
{
throw new FSWriteError(e, getPath());
}
if (runPostFlush != null)
runPostFlush.run();
}
public boolean hasPosition()
{
return true;
}
public long position()
{
return current();
}
/**
* Returns the current file pointer of the underlying on-disk file.
* Note that since write works by buffering data, the value of this will increase by buffer
* size and not every write to the writer will modify this value.
* Furthermore, for compressed files, this value refers to compressed data, while the
* writer getFilePointer() refers to uncompressedFile
*
* @return the current file pointer
*/
public long getOnDiskFilePointer()
{
return position();
}
public long getEstimatedOnDiskBytesWritten()
{
return getOnDiskFilePointer();
}
public long length()
{
try
{
return Math.max(current(), fchannel.size());
}
catch (IOException e)
{
throw new FSReadError(e, getPath());
}
}
public String getPath()
{
return filePath;
}
protected void resetBuffer()
{
bufferOffset = current();
buffer.clear();
}
protected long current()
{
return bufferOffset + (buffer == null ? 0 : buffer.position());
}
public DataPosition mark()
{
return new BufferedFileWriterMark(current());
}
/**
* Drops all buffered data that's past the limits of our new file mark + buffer capacity, or syncs and truncates
* the underlying file to the marked position
*/
public void resetAndTruncate(DataPosition mark)
{
assert mark instanceof BufferedFileWriterMark;
long previous = current();
long truncateTarget = ((BufferedFileWriterMark) mark).pointer;
// If we're resetting to a point within our buffered data, just adjust our buffered position to drop bytes to
// the right of the desired mark.
if (previous - truncateTarget <= buffer.position())
{
buffer.position(buffer.position() - ((int) (previous - truncateTarget)));
return;
}
// synchronize current buffer with disk - we don't want any data loss
syncInternal();
// truncate file to given position
truncate(truncateTarget);
try
{
fchannel.position(truncateTarget);
}
catch (IOException e)
{
throw new FSReadError(e, getPath());
}
bufferOffset = truncateTarget;
resetBuffer();
}
public long getLastFlushOffset()
{
return lastFlushOffset;
}
public void truncate(long toSize)
{
try
{
fchannel.truncate(toSize);
lastFlushOffset = toSize;
}
catch (IOException e)
{
throw new FSWriteError(e, getPath());
}
}
public boolean isOpen()
{
return channel.isOpen();
}
public final void prepareToCommit()
{
txnProxy.prepareToCommit();
}
public final Throwable commit(Throwable accumulate)
{
return txnProxy.commit(accumulate);
}
public final Throwable abort(Throwable accumulate)
{
return txnProxy.abort(accumulate);
}
@Override
public final void close()
{
if (option.finishOnClose())
txnProxy.finish();
else
txnProxy.close();
}
public int writeDirectlyToChannel(ByteBuffer buf) throws IOException
{
if (strictFlushing)
throw new UnsupportedOperationException();
// Don't allow writes to the underlying channel while data is buffered
flush();
return channel.write(buf);
}
public final void finish()
{
txnProxy.finish();
}
protected TransactionalProxy txnProxy()
{
return new TransactionalProxy();
}
/**
* Class to hold a mark to the position of the file
*/
protected static class BufferedFileWriterMark implements DataPosition
{
final long pointer;
public BufferedFileWriterMark(long pointer)
{
this.pointer = pointer;
}
}
}
© 2015 - 2024 Weber Informatics LLC | Privacy Policy