@Override
public int read() throws IOException {
mReadByte++;
if (mReadByte > mBlockInfo.length) {
doneRecache();
return -1;
}
if (mCurrentBuffer != null) {
if (mCurrentBuffer.remaining() == 0) {
mBufferStartPosition = mReadByte - 1;
updateCurrentBuffer();
}
if (mCurrentBuffer != null) {
int ret = mCurrentBuffer.get() & 0xFF;
if (mRecache) {
mBlockOutStream.write(ret);
}
return ret;
}
setupStreamFromUnderFs(mBlockInfo.offset + mReadByte - 1, mUFSConf);
}
int ret = mCheckpointInputStream.read() & 0xFF;
if (mRecache) {
mBlockOutStream.write(ret);
}
return ret;
}
@Override
public int read(byte b[], int off, int len) throws IOException {
if (b == null) {
throw new NullPointerException();
} else if (off < 0 || len < 0 || len > b.length - off) {
throw new IndexOutOfBoundsException();
} else if (len == 0) {
return 0;
}
long ret = mBlockInfo.length - mReadByte;
if (ret < len) {
len = (int) ret;
}
if (ret == 0) {
return -1;
}
if (mCurrentBuffer != null) {
if (mCurrentBuffer.remaining() == 0) {
mBufferStartPosition = mReadByte;
updateCurrentBuffer();
}
if (mCurrentBuffer != null) {
ret = Math.min(ret, mCurrentBuffer.remaining());
ret = Math.min(ret, len);
mCurrentBuffer.get(b, off, (int) ret);
mReadByte += ret;
if (mRecache) {
mBlockOutStream.write(b, off, (int) ret);
if (mReadByte == mBlockInfo.length) {
doneRecache();
}
}
return (int) ret;
}
setupStreamFromUnderFs(mBlockInfo.offset + mReadByte, mUFSConf);
}
ret = mCheckpointInputStream.read(b, off, len);
mReadByte += ret;
if (mRecache) {
mBlockOutStream.write(b, off, (int) ret);
if (mReadByte == mBlockInfo.length) {
doneRecache();
}
}
return (int) ret;
}
/**
* Cancels the re-caching attempt
*
* @throws IOException when the underlying stream cannot be canceled
*/
private void cancelRecache() throws IOException {
if (mRecache) {
mRecache = false;
if (mBlockOutStream != null) {
mBlockOutStream.cancel();
}
}
}
@Override
public void close() throws IOException {
if (!mClosed) {
if (mRecache) {
mBlockOutStream.cancel();
}
if (mCheckpointInputStream != null) {
mCheckpointInputStream.close();
}
}
mClosed = true;
}
@Override
public long skip(long n) throws IOException {
if (n <= 0) {
return 0;
}
long ret = mBlockInfo.length - mReadByte;
if (ret > n) {
ret = n;
}
if (mCurrentBuffer != null) {
if (mCurrentBuffer.remaining() < ret) {
mBufferStartPosition = mReadByte + ret;
updateCurrentBuffer();
}
if (mCurrentBuffer != null) {
if (ret > 0) {
if (mRecache) {
mBlockOutStream.cancel();
}
mRecache = false;
}
return (int) ret;
}
setupStreamFromUnderFs(mBlockInfo.offset + mReadByte, mUFSConf);
}
long tmp = mCheckpointInputStream.skip(ret);
ret = Math.min(ret, tmp);
mReadByte += ret;
if (ret > 0) {
if (mRecache) {
mBlockOutStream.cancel();
}
mRecache = false;
}
return ret;
}
@Override
public void close() throws IOException {
if (mClosed) {
return;
}
if (mRecache && mBlockOutStream != null) {
// We only finish re-caching if we've gotten to the end of the file
if (mBlockPos == mBlockInfo.length) {
mBlockOutStream.close();
} else {
mBlockOutStream.cancel();
}
}
if (mCheckpointInputStream != null) {
mCheckpointInputStream.close();
}
if (mBytesReadRemote > 0) {
mTachyonFS.getClientMetrics().incBlocksReadRemote(1);
}
closeReader();
mClosed = true;
}
/**
* Re-cache the block into memory
*
* @param blockIndex The block index of the current file.
* @return true if succeed, false otherwise
* @throws IOException
*/
boolean recache(int blockIndex) throws IOException {
String path = getUfsPath();
UnderFileSystem underFsClient = UnderFileSystem.get(path, mTachyonConf);
InputStream inputStream = null;
BlockOutStream bos = null;
try {
inputStream = underFsClient.open(path);
long length = getBlockSizeByte();
long offset = blockIndex * length;
inputStream.skip(offset);
int bufferBytes = (int) mTachyonConf.getBytes(Constants.USER_FILE_BUFFER_BYTES, Constants.MB);
byte[] buffer = new byte[bufferBytes];
bos = BlockOutStream.get(this, WriteType.TRY_CACHE, blockIndex, mTachyonConf);
int limit;
while (length > 0 && ((limit = inputStream.read(buffer)) >= 0)) {
if (limit != 0) {
if (length >= limit) {
bos.write(buffer, 0, limit);
length -= limit;
} else {
bos.write(buffer, 0, (int) length);
length = 0;
}
}
}
bos.close();
} catch (IOException e) {
LOG.warn(e.getMessage(), e);
if (bos != null) {
bos.cancel();
}
return false;
} finally {
if (inputStream != null) {
inputStream.close();
}
}
return true;
}
private void doneRecache() throws IOException {
if (mRecache) {
mBlockOutStream.close();
}
}
@Override
public int read(byte[] b, int off, int len) throws IOException {
if (b == null) {
throw new NullPointerException();
} else if (off < 0 || len < 0 || len > b.length - off) {
throw new IndexOutOfBoundsException();
} else if (len == 0) {
return 0;
} else if (mBlockPos == mBlockInfo.length) {
return -1;
}
// We read at most len bytes, but if mBlockPos + len exceeds the length of the file, we only
// read up to the end of the file
len = (int) Math.min(len, mBlockInfo.length - mBlockPos);
int bytesLeft = len;
// Lazy initialization of the out stream for caching to avoid collisions with other caching
// attempts that are invalidated later due to seek/skips
if (bytesLeft > 0 && mBlockOutStream == null && mRecache) {
try {
mBlockOutStream = BlockOutStream.get(mFile, WriteType.TRY_CACHE, mBlockIndex, mTachyonConf);
// We should only cache when we are writing to a local worker
if (mBlockOutStream instanceof RemoteBlockOutStream) {
LOG.info("Cannot find a local worker to write to, recache attempt cancelled.");
cancelRecache();
}
} catch (IOException ioe) {
LOG.warn("Recache attempt failed.", ioe);
cancelRecache();
}
}
// While we still have bytes to read, make sure the buffer is set to read the byte at mBlockPos.
// If we fail to set mCurrentBuffer, we stream the rest from the underfs
while (bytesLeft > 0 && mAttemptReadFromWorkers && updateCurrentBuffer()) {
int bytesToRead = Math.min(bytesLeft, mCurrentBuffer.remaining());
mCurrentBuffer.get(b, off, bytesToRead);
if (mRecache) {
mBlockOutStream.write(b, off, bytesToRead);
}
off += bytesToRead;
bytesLeft -= bytesToRead;
mBlockPos += bytesToRead;
}
mBytesReadRemote += len - bytesLeft;
mTachyonFS.getClientMetrics().incBytesReadRemote(len - bytesLeft);
if (bytesLeft > 0) {
// Unable to read from worker memory, reading this block from underfs in the future.
mAttemptReadFromWorkers = false;
// We failed to read everything from mCurrentBuffer, so we need to stream the rest from the
// underfs
if (!setupStreamFromUnderFs()) {
LOG.error(
"Failed to read at position "
+ mBlockPos
+ " in block "
+ mBlockInfo.getBlockId()
+ " from workers or underfs");
// Return the number of bytes we managed to read
return len - bytesLeft;
}
while (bytesLeft > 0) {
int readBytes = mCheckpointInputStream.read(b, off, bytesLeft);
if (readBytes <= 0) {
LOG.error("Checkpoint stream read 0 bytes, which shouldn't ever happen");
return len - bytesLeft;
}
if (mRecache) {
mBlockOutStream.write(b, off, readBytes);
}
off += readBytes;
bytesLeft -= readBytes;
mBlockPos += readBytes;
mCheckpointPos += readBytes;
mTachyonFS.getClientMetrics().incBytesReadUfs(readBytes);
}
}
return len;
}
