/**
* Returns the local filename for the block if that file exists on the local file system. This is
* an alpha power-api feature for applications that want short-circuit-read files directly. There
* is no guarantee that the file still exists after this call returns, as Tachyon may evict blocks
* from memory at any time.
*
* @param blockIndex The index of the block in the file.
* @return filename on local file system or null if file not present on local file system.
* @throws IOException
*/
public String getLocalFilename(int blockIndex) throws IOException {
ClientBlockInfo blockInfo = getClientBlockInfo(blockIndex);
long blockId = blockInfo.getBlockId();
int blockLockId = mTachyonFS.getBlockLockId();
String filename = mTachyonFS.lockBlock(blockId, blockLockId);
if (filename != null) {
mTachyonFS.unlockBlock(blockId, blockLockId);
}
return filename;
}
private ByteBuffer readRemoteByteBuffer(ClientBlockInfo blockInfo, long offset, long len) {
ByteBuffer buf = null;
try {
List<NetAddress> blockLocations = blockInfo.getLocations();
LOG.info("Block locations:" + blockLocations);
for (int k = 0; k < blockLocations.size(); k++) {
String host = blockLocations.get(k).mHost;
int port = blockLocations.get(k).mSecondaryPort;
// The data is not in remote machine's memory if port == -1.
if (port == -1) {
continue;
}
if (host.equals(InetAddress.getLocalHost().getHostName())
|| host.equals(InetAddress.getLocalHost().getHostAddress())) {
String localFileName = CommonUtils.concat(TFS.getRootFolder(), blockInfo.blockId);
LOG.warn("Master thinks the local machine has data " + localFileName + "! But not!");
}
LOG.info(
host
+ ":"
+ port
+ " current host is "
+ InetAddress.getLocalHost().getHostName()
+ " "
+ InetAddress.getLocalHost().getHostAddress());
try {
buf =
retrieveByteBufferFromRemoteMachine(
new InetSocketAddress(host, port), blockInfo.blockId, offset, len);
if (buf != null) {
break;
}
} catch (IOException e) {
LOG.error(e.getMessage());
buf = null;
}
}
} catch (IOException e) {
LOG.error("Failed to get read data from remote " + e.getMessage());
buf = null;
}
return buf;
}
/**
* Reads from a remote byte buffer.
*
* @param tachyonFS a TachyonFS
* @param blockInfo the block information
* @param offset offset to start the read at
* @param len number of bytes to read
* @param conf Tachyon configuration
* @return <code>ByteBuffer</code> containing the bytes read
*/
public ByteBuffer readRemoteByteBuffer(
TachyonFS tachyonFS, ClientBlockInfo blockInfo, long offset, long len, TachyonConf conf) {
ByteBuffer buf = null;
try {
List<NetAddress> blockLocations = blockInfo.getLocations();
LOG.info("Block locations:" + blockLocations);
String localhost = NetworkAddressUtils.getConnectHost(ServiceType.WORKER_RPC, conf);
for (NetAddress blockLocation : blockLocations) {
String host = blockLocation.mHost;
int port = blockLocation.mSecondaryPort;
// The data is not in remote machine's memory if primary port == -1. We check primary port
// because if the data is in the under storage, the secondary port (data transfer port)
// will be set.
if (blockLocation.mPort == -1) {
continue;
}
if (host.equals(InetAddress.getLocalHost().getHostName())
|| host.equals(InetAddress.getLocalHost().getHostAddress())
|| host.equals(localhost)) {
LOG.warn(
"Master thinks the local machine has data, but not!"
+ "(or local read is disabled) blockId:{}",
blockInfo.blockId);
}
LOG.info(
host
+ ":"
+ port
+ " current host is "
+ localhost
+ " "
+ NetworkAddressUtils.getLocalIpAddress(conf));
try {
buf =
retrieveByteBufferFromRemoteMachine(
new InetSocketAddress(host, port), blockInfo.blockId, offset, len, conf);
if (buf != null) {
break;
}
} catch (IOException e) {
LOG.error(
"Fail to retrieve byte buffer for block "
+ blockInfo.blockId
+ " from remote "
+ host
+ ":"
+ port
+ " with offset "
+ offset
+ " and length "
+ len,
e);
buf = null;
}
}
} catch (IOException e) {
LOG.error("Failed to get read data from remote ", e);
buf = null;
}
return buf;
}
@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;
}
/**
* Promote block back to top layer after access
*
* @param blockIndex the index of the block
* @return true if success, false otherwise
* @throws IOException
*/
public boolean promoteBlock(int blockIndex) throws IOException {
ClientBlockInfo blockInfo = getClientBlockInfo(blockIndex);
return mTachyonFS.promoteBlock(blockInfo.getBlockId());
}