/*
* Read ClusterID, StorageID, StorageType, CTime from
* DataStorage VERSION file and verify them.
*/
@Override
protected void setFieldsFromProperties(Properties props, StorageDirectory sd) throws IOException {
setLayoutVersion(props, sd);
setcTime(props, sd);
setStorageType(props, sd);
setClusterId(props, layoutVersion, sd);
// Read NamespaceID in version before federation
if (!LayoutVersion.supports(Feature.FEDERATION, layoutVersion)) {
setNamespaceID(props, sd);
}
// valid storage id, storage id may be empty
String ssid = props.getProperty("storageID");
if (ssid == null) {
throw new InconsistentFSStateException(
sd.getRoot(), "file " + STORAGE_FILE_VERSION + " is invalid.");
}
String sid = getStorageID();
if (!(sid.equals("") || ssid.equals("") || sid.equals(ssid))) {
throw new InconsistentFSStateException(sd.getRoot(), "has incompatible storage Id.");
}
if (sid.equals("")) { // update id only if it was empty
setStorageID(ssid);
}
}
private void init() throws LogHeaderCorruptException, IOException {
Preconditions.checkState(state == State.UNINIT);
BufferedInputStream bin = null;
try {
fStream = log.getInputStream();
bin = new BufferedInputStream(fStream);
tracker = new FSEditLogLoader.PositionTrackingInputStream(bin);
dataIn = new DataInputStream(tracker);
try {
logVersion = readLogVersion(dataIn);
} catch (EOFException eofe) {
throw new LogHeaderCorruptException("No header found in log");
}
if (LayoutVersion.supports(Feature.ADD_LAYOUT_FLAGS, logVersion)) {
try {
LayoutFlags.read(dataIn);
} catch (EOFException eofe) {
throw new LogHeaderCorruptException("EOF while reading layout " + "flags from log");
}
}
reader = new FSEditLogOp.Reader(dataIn, tracker, logVersion);
reader.setMaxOpSize(maxOpSize);
state = State.OPEN;
} finally {
if (reader == null) {
IOUtils.cleanup(LOG, dataIn, tracker, bin, fStream);
state = State.CLOSED;
}
}
}
/**
* Hardlink all finalized and RBW blocks in fromDir to toDir
*
* @param fromDir The directory where the 'from' snapshot is stored
* @param fromBbwDir In HDFS 1.x, the directory where blocks that are under construction are
* stored.
* @param toDir The current data directory
* @throws IOException If error occurs during hardlink
*/
private void linkAllBlocks(File fromDir, File fromBbwDir, File toDir) throws IOException {
HardLink hardLink = new HardLink();
// do the link
int diskLayoutVersion = this.getLayoutVersion();
if (LayoutVersion.supports(Feature.APPEND_RBW_DIR, diskLayoutVersion)) {
// hardlink finalized blocks in tmpDir/finalized
linkBlocks(
new File(fromDir, STORAGE_DIR_FINALIZED),
new File(toDir, STORAGE_DIR_FINALIZED),
diskLayoutVersion,
hardLink);
// hardlink rbw blocks in tmpDir/rbw
linkBlocks(
new File(fromDir, STORAGE_DIR_RBW),
new File(toDir, STORAGE_DIR_RBW),
diskLayoutVersion,
hardLink);
} else { // pre-RBW version
// hardlink finalized blocks in tmpDir
linkBlocks(fromDir, new File(toDir, STORAGE_DIR_FINALIZED), diskLayoutVersion, hardLink);
if (fromBbwDir.exists()) {
/*
* We need to put the 'blocksBeingWritten' from HDFS 1.x into the rbw
* directory. It's a little messy, because the blocksBeingWriten was
* NOT underneath the 'current' directory in those releases. See
* HDFS-3731 for details.
*/
linkBlocks(fromBbwDir, new File(toDir, STORAGE_DIR_RBW), diskLayoutVersion, hardLink);
}
}
LOG.info(hardLink.linkStats.report());
}
/**
* Upgrade -- Move current storage into a backup directory, and hardlink all its blocks into the
* new current directory.
*
* <p>Upgrade from pre-0.22 to 0.22 or later release e.g. 0.19/0.20/ => 0.22/0.23
*
* <ul>
* <li>If <SD>/previous exists then delete it
* <li>Rename <SD>/current to <SD>/previous.tmp
* <li>Create new <SD>/current/<bpid>/current directory
* <li>
* <ul>
* <li>Hard links for block files are created from <SD>/previous.tmp to
* <SD>/current/<bpid>/current
* <li>Saves new version file in <SD>/current/<bpid>/current directory
* </ul>
* <li>Rename <SD>/previous.tmp to <SD>/previous
* </ul>
*
* There should be only ONE namenode in the cluster for first time upgrade to 0.22
*
* @param sd storage directory
* @throws IOException on error
*/
void doUpgrade(StorageDirectory sd, NamespaceInfo nsInfo) throws IOException {
if (LayoutVersion.supports(Feature.FEDERATION, layoutVersion)) {
clusterID = nsInfo.getClusterID();
layoutVersion = nsInfo.getLayoutVersion();
writeProperties(sd);
return;
}
LOG.info(
"Upgrading storage directory "
+ sd.getRoot()
+ ".\n old LV = "
+ this.getLayoutVersion()
+ "; old CTime = "
+ this.getCTime()
+ ".\n new LV = "
+ nsInfo.getLayoutVersion()
+ "; new CTime = "
+ nsInfo.getCTime());
File curDir = sd.getCurrentDir();
File prevDir = sd.getPreviousDir();
File bbwDir = new File(sd.getRoot(), Storage.STORAGE_1_BBW);
assert curDir.exists() : "Data node current directory must exist.";
// Cleanup directory "detach"
cleanupDetachDir(new File(curDir, STORAGE_DIR_DETACHED));
// 1. delete <SD>/previous dir before upgrading
if (prevDir.exists()) deleteDir(prevDir);
// get previous.tmp directory, <SD>/previous.tmp
File tmpDir = sd.getPreviousTmp();
assert !tmpDir.exists() : "Data node previous.tmp directory must not exist.";
// 2. Rename <SD>/current to <SD>/previous.tmp
rename(curDir, tmpDir);
// 3. Format BP and hard link blocks from previous directory
File curBpDir = BlockPoolSliceStorage.getBpRoot(nsInfo.getBlockPoolID(), curDir);
BlockPoolSliceStorage bpStorage =
new BlockPoolSliceStorage(
nsInfo.getNamespaceID(),
nsInfo.getBlockPoolID(),
nsInfo.getCTime(),
nsInfo.getClusterID());
bpStorage.format(curDir, nsInfo);
linkAllBlocks(tmpDir, bbwDir, new File(curBpDir, STORAGE_DIR_CURRENT));
// 4. Write version file under <SD>/current
layoutVersion = HdfsConstants.LAYOUT_VERSION;
clusterID = nsInfo.getClusterID();
writeProperties(sd);
// 5. Rename <SD>/previous.tmp to <SD>/previous
rename(tmpDir, prevDir);
LOG.info("Upgrade of " + sd.getRoot() + " is complete");
addBlockPoolStorage(nsInfo.getBlockPoolID(), bpStorage);
}
/**
* Analize which and whether a transition of the fs state is required and perform it if necessary.
*
* <p>Rollback if previousLV >= LAYOUT_VERSION && prevCTime <= namenode.cTime Upgrade if this.LV >
* LAYOUT_VERSION || this.cTime < namenode.cTime Regular startup if this.LV = LAYOUT_VERSION &&
* this.cTime = namenode.cTime
*
* @param datanode Datanode to which this storage belongs to
* @param sd storage directory
* @param nsInfo namespace info
* @param startOpt startup option
* @throws IOException
*/
private void doTransition(
DataNode datanode, StorageDirectory sd, NamespaceInfo nsInfo, StartupOption startOpt)
throws IOException {
if (startOpt == StartupOption.ROLLBACK) {
doRollback(sd, nsInfo); // rollback if applicable
}
readProperties(sd);
checkVersionUpgradable(this.layoutVersion);
assert this.layoutVersion >= HdfsConstants.LAYOUT_VERSION : "Future version is not allowed";
boolean federationSupported = LayoutVersion.supports(Feature.FEDERATION, layoutVersion);
// For pre-federation version - validate the namespaceID
if (!federationSupported && getNamespaceID() != nsInfo.getNamespaceID()) {
throw new IOException(
"Incompatible namespaceIDs in "
+ sd.getRoot().getCanonicalPath()
+ ": namenode namespaceID = "
+ nsInfo.getNamespaceID()
+ "; datanode namespaceID = "
+ getNamespaceID());
}
// For version that supports federation, validate clusterID
if (federationSupported && !getClusterID().equals(nsInfo.getClusterID())) {
throw new IOException(
"Incompatible clusterIDs in "
+ sd.getRoot().getCanonicalPath()
+ ": namenode clusterID = "
+ nsInfo.getClusterID()
+ "; datanode clusterID = "
+ getClusterID());
}
// regular start up
if (this.layoutVersion == HdfsConstants.LAYOUT_VERSION && this.cTime == nsInfo.getCTime())
return; // regular startup
// verify necessity of a distributed upgrade
UpgradeManagerDatanode um = datanode.getUpgradeManagerDatanode(nsInfo.getBlockPoolID());
verifyDistributedUpgradeProgress(um, nsInfo);
// do upgrade
if (this.layoutVersion > HdfsConstants.LAYOUT_VERSION || this.cTime < nsInfo.getCTime()) {
doUpgrade(sd, nsInfo); // upgrade
return;
}
// layoutVersion == LAYOUT_VERSION && this.cTime > nsInfo.cTime
// must shutdown
throw new IOException(
"Datanode state: LV = "
+ this.getLayoutVersion()
+ " CTime = "
+ this.getCTime()
+ " is newer than the namespace state: LV = "
+ nsInfo.getLayoutVersion()
+ " CTime = "
+ nsInfo.getCTime());
}
/*
* Set ClusterID, StorageID, StorageType, CTime into
* DataStorage VERSION file
*/
@Override
protected void setPropertiesFromFields(Properties props, StorageDirectory sd) throws IOException {
props.setProperty("storageType", storageType.toString());
props.setProperty("clusterID", clusterID);
props.setProperty("cTime", String.valueOf(cTime));
props.setProperty("layoutVersion", String.valueOf(layoutVersion));
props.setProperty("storageID", getStorageID());
// Set NamespaceID in version before federation
if (!LayoutVersion.supports(Feature.FEDERATION, layoutVersion)) {
props.setProperty("namespaceID", String.valueOf(namespaceID));
}
}
/**
* Cleanup the detachDir.
*
* <p>If the directory is not empty report an error; Otherwise remove the directory.
*
* @param detachDir detach directory
* @throws IOException if the directory is not empty or it can not be removed
*/
private void cleanupDetachDir(File detachDir) throws IOException {
if (!LayoutVersion.supports(Feature.APPEND_RBW_DIR, layoutVersion)
&& detachDir.exists()
&& detachDir.isDirectory()) {
if (FileUtil.list(detachDir).length != 0) {
throw new IOException(
"Detached directory "
+ detachDir
+ " is not empty. Please manually move each file under this "
+ "directory to the finalized directory if the finalized "
+ "directory tree does not have the file.");
} else if (!detachDir.delete()) {
throw new IOException("Cannot remove directory " + detachDir);
}
}
}
/** modified by tony */
@SuppressWarnings("deprecation")
int loadEditRecords(int logVersion, DataInputStream in, boolean closeOnExit) throws IOException {
FSNamesystem.LOG.info("logversion: " + logVersion);
FSDirectory fsDir = fsNamesys.dir;
int numEdits = 0;
String clientName = null;
String clientMachine = null;
String path = null;
int numOpAdd = 0,
numOpClose = 0,
numOpDelete = 0,
numOpRenameOld = 0,
numOpSetRepl = 0,
numOpMkDir = 0,
numOpSetPerm = 0,
numOpSetOwner = 0,
numOpSetGenStamp = 0,
numOpTimes = 0,
numOpRename = 0,
numOpConcatDelete = 0,
numOpSymlink = 0,
numOpGetDelegationToken = 0,
numOpRenewDelegationToken = 0,
numOpCancelDelegationToken = 0,
numOpUpdateMasterKey = 0,
numOpOther = 0;
try {
while (true) {
long timestamp = 0;
long mtime = 0;
long atime = 0;
long blockSize = 0;
byte opcode = -1;
try {
in.mark(1);
opcode = in.readByte();
if (opcode == Ops.OP_INVALID) {
in.reset(); // reset back to end of file if somebody reads it again
break; // no more transactions
}
} catch (EOFException e) {
break; // no more transactions
}
numEdits++;
switch (opcode) {
case Ops.OP_ADD:
case Ops.OP_CLOSE:
{
// versions > 0 support per file replication
// get name and replication
int length = in.readInt();
// modified by tony
if (-7 == logVersion && length != 3
|| -17 < logVersion && logVersion < -7 && length != 4
|| logVersion <= -17 && length != 7) {
throw new IOException(
"Incorrect data format."
+ " logVersion is "
+ logVersion
+ " but writables.length is "
+ length
+ ". ");
}
path = FSImageSerialization.readString(in);
short replication = fsNamesys.adjustReplication(readShort(in));
mtime = readLong(in);
if (LayoutVersion.supports(Feature.FILE_ACCESS_TIME, logVersion)) {
atime = readLong(in);
}
if (logVersion < -7) {
blockSize = readLong(in);
}
long fileSize = readLong(in);
byte type = (byte) readLong(in);
// get blocks
boolean isFileUnderConstruction = (opcode == Ops.OP_ADD);
BlockInfo blocks[] = readBlocks(in, logVersion, isFileUnderConstruction, replication);
// Older versions of HDFS does not store the block size in inode.
// If the file has more than one block, use the size of the
// first block as the blocksize. Otherwise use the default
// block size.
if (-8 <= logVersion && blockSize == 0) {
if (blocks.length > 1) {
blockSize = blocks[0].getNumBytes();
} else {
long first = ((blocks.length == 1) ? blocks[0].getNumBytes() : 0);
blockSize = Math.max(fsNamesys.getDefaultBlockSize(), first);
}
}
PermissionStatus permissions = fsNamesys.getUpgradePermission();
if (logVersion <= -11) {
permissions = PermissionStatus.read(in);
}
CodingMatrix codingMatrix = CodingMatrix.getMatrixofCertainType(type);
codingMatrix.readFields(in);
/** added by tony* */
LongWritable offset = new LongWritable();
offset.readFields(in);
long headeroffset = offset.get();
// clientname, clientMachine and block locations of last block.
if (opcode == Ops.OP_ADD && logVersion <= -12) {
clientName = FSImageSerialization.readString(in);
clientMachine = FSImageSerialization.readString(in);
if (-13 <= logVersion) {
readDatanodeDescriptorArray(in);
}
} else {
clientName = "";
clientMachine = "";
}
// The open lease transaction re-creates a file if necessary.
// Delete the file if it already exists.
if (FSNamesystem.LOG.isDebugEnabled()) {
FSNamesystem.LOG.debug(
opcode
+ ": "
+ path
+ " numblocks : "
+ blocks.length
+ " clientHolder "
+ clientName
+ " clientMachine "
+ clientMachine);
}
fsDir.unprotectedDelete(path, mtime);
/** modified by tony add to the file tree */
INodeFile node =
(INodeFile)
fsDir.unprotectedAddFile(
path,
permissions,
codingMatrix,
headeroffset,
fileSize,
blocks,
replication,
mtime,
atime,
blockSize);
if (isFileUnderConstruction) {
numOpAdd++;
//
// Replace current node with a INodeUnderConstruction.
// Recreate in-memory lease record.
//
// INodeFileUnderConstruction cons = new INodeFileUnderConstruction(
// node.getLocalNameBytes(),
// node.getReplication(),
// node.getModificationTime(),
// node.getPreferredBlockSize(),
// node.getBlocks(),
// node.getPermissionStatus(),
// clientName,
// clientMachine,
// null);
// TODO:
INodeFileUnderConstruction cons = null;
fsDir.replaceNode(path, node, cons);
fsNamesys.leaseManager.addLease(cons.getClientName(), path);
}
break;
}
case Ops.OP_SET_REPLICATION:
{
numOpSetRepl++;
path = FSImageSerialization.readString(in);
short replication = fsNamesys.adjustReplication(readShort(in));
fsDir.unprotectedSetReplication(path, replication, null);
break;
}
case Ops.OP_CONCAT_DELETE:
{
numOpConcatDelete++;
int length = in.readInt();
if (length < 3) { // trg, srcs.., timestam
throw new IOException("Incorrect data format. " + "Mkdir operation.");
}
String trg = FSImageSerialization.readString(in);
int srcSize = length - 1 - 1; // trg and timestamp
String[] srcs = new String[srcSize];
for (int i = 0; i < srcSize; i++) {
srcs[i] = FSImageSerialization.readString(in);
}
timestamp = readLong(in);
fsDir.unprotectedConcat(trg, srcs);
break;
}
case Ops.OP_RENAME_OLD:
{
numOpRenameOld++;
int length = in.readInt();
if (length != 3) {
throw new IOException("Incorrect data format. " + "Mkdir operation.");
}
String s = FSImageSerialization.readString(in);
String d = FSImageSerialization.readString(in);
timestamp = readLong(in);
HdfsFileStatus dinfo = fsDir.getFileInfo(d, false);
fsDir.unprotectedRenameTo(s, d, timestamp);
fsNamesys.changeLease(s, d, dinfo);
break;
}
case Ops.OP_DELETE:
{
numOpDelete++;
int length = in.readInt();
if (length != 2) {
throw new IOException("Incorrect data format. " + "delete operation.");
}
path = FSImageSerialization.readString(in);
timestamp = readLong(in);
fsDir.unprotectedDelete(path, timestamp);
break;
}
case Ops.OP_MKDIR:
{
numOpMkDir++;
PermissionStatus permissions = fsNamesys.getUpgradePermission();
int length = in.readInt();
if (-17 < logVersion && length != 2 || logVersion <= -17 && length != 3) {
throw new IOException("Incorrect data format. " + "Mkdir operation.");
}
path = FSImageSerialization.readString(in);
timestamp = readLong(in);
// The disk format stores atimes for directories as well.
// However, currently this is not being updated/used because of
// performance reasons.
if (LayoutVersion.supports(Feature.FILE_ACCESS_TIME, logVersion)) {
atime = readLong(in);
}
if (logVersion <= -11) {
permissions = PermissionStatus.read(in);
}
fsDir.unprotectedMkdir(path, permissions, timestamp);
break;
}
case Ops.OP_SET_GENSTAMP:
{
numOpSetGenStamp++;
long lw = in.readLong();
fsNamesys.setGenerationStamp(lw);
break;
}
case Ops.OP_DATANODE_ADD:
{
numOpOther++;
// Datanodes are not persistent any more.
FSImageSerialization.DatanodeImage.skipOne(in);
break;
}
case Ops.OP_DATANODE_REMOVE:
{
numOpOther++;
DatanodeID nodeID = new DatanodeID();
nodeID.readFields(in);
// Datanodes are not persistent any more.
break;
}
case Ops.OP_SET_PERMISSIONS:
{
numOpSetPerm++;
fsDir.unprotectedSetPermission(
FSImageSerialization.readString(in), FsPermission.read(in));
break;
}
case Ops.OP_SET_OWNER:
{
numOpSetOwner++;
fsDir.unprotectedSetOwner(
FSImageSerialization.readString(in),
FSImageSerialization.readString_EmptyAsNull(in),
FSImageSerialization.readString_EmptyAsNull(in));
break;
}
case Ops.OP_SET_NS_QUOTA:
{
fsDir.unprotectedSetQuota(
FSImageSerialization.readString(in),
readLongWritable(in),
FSConstants.QUOTA_DONT_SET);
break;
}
case Ops.OP_CLEAR_NS_QUOTA:
{
fsDir.unprotectedSetQuota(
FSImageSerialization.readString(in),
FSConstants.QUOTA_RESET,
FSConstants.QUOTA_DONT_SET);
break;
}
case Ops.OP_SET_QUOTA:
fsDir.unprotectedSetQuota(
FSImageSerialization.readString(in), readLongWritable(in), readLongWritable(in));
break;
case Ops.OP_TIMES:
{
numOpTimes++;
int length = in.readInt();
if (length != 3) {
throw new IOException("Incorrect data format. " + "times operation.");
}
path = FSImageSerialization.readString(in);
mtime = readLong(in);
atime = readLong(in);
fsDir.unprotectedSetTimes(path, mtime, atime, true);
break;
}
case Ops.OP_SYMLINK:
{
numOpSymlink++;
int length = in.readInt();
if (length != 4) {
throw new IOException("Incorrect data format. " + "symlink operation.");
}
path = FSImageSerialization.readString(in);
String value = FSImageSerialization.readString(in);
mtime = readLong(in);
atime = readLong(in);
PermissionStatus perm = PermissionStatus.read(in);
fsDir.unprotectedSymlink(path, value, mtime, atime, perm);
break;
}
case Ops.OP_RENAME:
{
numOpRename++;
int length = in.readInt();
if (length != 3) {
throw new IOException("Incorrect data format. " + "Mkdir operation.");
}
String s = FSImageSerialization.readString(in);
String d = FSImageSerialization.readString(in);
timestamp = readLong(in);
Rename[] options = readRenameOptions(in);
HdfsFileStatus dinfo = fsDir.getFileInfo(d, false);
fsDir.unprotectedRenameTo(s, d, timestamp, options);
fsNamesys.changeLease(s, d, dinfo);
break;
}
case Ops.OP_GET_DELEGATION_TOKEN:
{
numOpGetDelegationToken++;
DelegationTokenIdentifier delegationTokenId = new DelegationTokenIdentifier();
delegationTokenId.readFields(in);
long expiryTime = readLong(in);
fsNamesys
.getDelegationTokenSecretManager()
.addPersistedDelegationToken(delegationTokenId, expiryTime);
break;
}
case Ops.OP_RENEW_DELEGATION_TOKEN:
{
numOpRenewDelegationToken++;
DelegationTokenIdentifier delegationTokenId = new DelegationTokenIdentifier();
delegationTokenId.readFields(in);
long expiryTime = readLong(in);
fsNamesys
.getDelegationTokenSecretManager()
.updatePersistedTokenRenewal(delegationTokenId, expiryTime);
break;
}
case Ops.OP_CANCEL_DELEGATION_TOKEN:
{
numOpCancelDelegationToken++;
DelegationTokenIdentifier delegationTokenId = new DelegationTokenIdentifier();
delegationTokenId.readFields(in);
fsNamesys
.getDelegationTokenSecretManager()
.updatePersistedTokenCancellation(delegationTokenId);
break;
}
case Ops.OP_UPDATE_MASTER_KEY:
{
numOpUpdateMasterKey++;
DelegationKey delegationKey = new DelegationKey();
delegationKey.readFields(in);
fsNamesys.getDelegationTokenSecretManager().updatePersistedMasterKey(delegationKey);
break;
}
default:
{
throw new IOException("Never seen opcode " + opcode);
}
}
}
} catch (IOException ex) {
check203UpgradeFailure(logVersion, ex);
} finally {
if (closeOnExit) in.close();
}
if (FSImage.LOG.isDebugEnabled()) {
FSImage.LOG.debug(
"numOpAdd = "
+ numOpAdd
+ " numOpClose = "
+ numOpClose
+ " numOpDelete = "
+ numOpDelete
+ " numOpRenameOld = "
+ numOpRenameOld
+ " numOpSetRepl = "
+ numOpSetRepl
+ " numOpMkDir = "
+ numOpMkDir
+ " numOpSetPerm = "
+ numOpSetPerm
+ " numOpSetOwner = "
+ numOpSetOwner
+ " numOpSetGenStamp = "
+ numOpSetGenStamp
+ " numOpTimes = "
+ numOpTimes
+ " numOpConcatDelete = "
+ numOpConcatDelete
+ " numOpRename = "
+ numOpRename
+ " numOpGetDelegationToken = "
+ numOpGetDelegationToken
+ " numOpRenewDelegationToken = "
+ numOpRenewDelegationToken
+ " numOpCancelDelegationToken = "
+ numOpCancelDelegationToken
+ " numOpUpdateMasterKey = "
+ numOpUpdateMasterKey
+ " numOpOther = "
+ numOpOther);
}
return numEdits;
}
public boolean versionSupportsFederation() {
return LayoutVersion.supports(Feature.FEDERATION, layoutVersion);
}