@InterfaceAudience.LimitedPrivate(value="Configuration") public class AsyncFSWAL extends AbstractFSWAL<WALProvider.AsyncWriter>
Here 'waitingConsumePayloads' acts as the RingBuffer in FSHLog.
For append, we process it as follow:
shouldScheduleConsumer() for more details.
waitingConsumePayloads and insert it into
toWriteAppendstoWriteAppends, append it to the AsyncWriter, and insert it into
unackedAppendsbatchSize, or there is a sync request, then we call
sync on the AsyncWriter.unackedAppends and drop it.unackedAppends back to toWriteAppends and
wait for writing them again.
Here we only describe the logic of doReplaceWriter. The main logic of rollWriter is same with
FSHLog.
For a normal roll request(for example, we have reached the log roll size):
waitingRoll(int) to true and
readyForRolling to false, and then wait on readyForRolling(see
waitForSafePoint()).waitingConsumePayloads if
waitingRoll(int) is true, and also stop writing the entries in toWriteAppends out.
unackedAppends is empty,
signal the readyForRollingCond.writerBroken(int) and waitingRoll(int) to false.WAL.Entry, WAL.Readerblocksize, closed, conf, coprocessorHost, DEFAULT_SLOW_SYNC_TIME_MS, filenum, fs, highestSyncedTxid, highestUnsyncedTxid, implClassName, listeners, LOG_NAME_COMPARATOR, logrollsize, maxLogs, numEntries, ourFiles, prefixPathStr, rollRequested, rollWriterLock, sequenceIdAccounting, shutdown, slowSyncNs, totalLogSize, walArchiveDir, walDir, walFile2Props, walFilePrefix, walFileSuffix, writer| Constructor and Description |
|---|
AsyncFSWAL(org.apache.hadoop.fs.FileSystem fs,
org.apache.hadoop.fs.Path rootDir,
String logDir,
String archiveDir,
org.apache.hadoop.conf.Configuration conf,
List<WALActionsListener> listeners,
boolean failIfWALExists,
String prefix,
String suffix,
org.apache.hbase.thirdparty.io.netty.channel.EventLoopGroup eventLoopGroup,
Class<? extends org.apache.hbase.thirdparty.io.netty.channel.Channel> channelClass) |
| Modifier and Type | Method and Description |
|---|---|
private void |
addTimeAnnotation(SyncFuture future,
String annotation) |
long |
append(RegionInfo hri,
WALKeyImpl key,
WALEdit edits,
boolean inMemstore)
NOTE: This append, at a time that is usually after this call returns, starts an mvcc
transaction by calling 'begin' wherein which we assign this update a sequenceid.
|
private void |
appendAndSync() |
private long |
closeWriter() |
private void |
consume() |
protected WALProvider.AsyncWriter |
createWriterInstance(org.apache.hadoop.fs.Path path) |
protected void |
doAppend(WALProvider.AsyncWriter writer,
FSWALEntry entry) |
protected boolean |
doCheckLogLowReplication() |
protected void |
doReplaceWriter(org.apache.hadoop.fs.Path oldPath,
org.apache.hadoop.fs.Path newPath,
WALProvider.AsyncWriter nextWriter)
Notice that you need to clear the
AbstractFSWAL.rollRequested flag in this method, as the new writer
will begin to work before returning from this method. |
protected void |
doShutdown() |
private static int |
epoch(int epochAndState) |
private int |
finishSync(boolean addSyncTrace) |
private int |
finishSyncLowerThanTxid(long txid,
boolean addSyncTrace) |
(package private) int |
getLogReplication()
This method gets the datanode replication count for the current WAL.
|
(package private) org.apache.hadoop.hdfs.protocol.DatanodeInfo[] |
getPipeline()
This method gets the pipeline for the current WAL.
|
private boolean |
shouldScheduleConsumer() |
void |
sync()
Sync what we have in the WAL.
|
void |
sync(long txid)
Sync the WAL if the txId was not already sync'd.
|
private void |
sync(WALProvider.AsyncWriter writer) |
private void |
syncCompleted(WALProvider.AsyncWriter writer,
long processedTxid,
long startTimeNs) |
private void |
syncFailed(long epochWhenSync,
Throwable error) |
private boolean |
trySetReadyForRolling() |
private void |
waitForSafePoint() |
private static boolean |
waitingRoll(int epochAndState) |
private static boolean |
writerBroken(int epochAndState) |
abortCacheFlush, append, atHeadOfRingBufferEventHandlerAppend, blockOnSync, checkLogLowReplication, close, completeCacheFlush, computeFilename, findRegionsToForceFlush, getCoprocessorHost, getCurrentFileName, getEarliestMemStoreSeqNum, getEarliestMemStoreSeqNum, getFilenum, getFileNumFromFileName, getFiles, getLogFileSize, getLogFileSizeIfBeingWritten, getNumLogFiles, getNumRolledLogFiles, getOldPath, getPreallocatedEventCount, getSyncFuture, getUnflushedEntriesCount, getWALArchivePath, isLogRollRequested, isUnflushedEntries, logRollAndSetupWalProps, main, postSync, registerWALActionsListener, replaceWriter, requestLogRoll, requestLogRoll, rollWriter, rollWriter, shutdown, stampSequenceIdAndPublishToRingBuffer, startCacheFlush, startCacheFlush, toString, unregisterWALActionsListener, updateStoreprivate static final org.slf4j.Logger LOG
private static final Comparator<SyncFuture> SEQ_COMPARATOR
public static final String WAL_BATCH_SIZE
public static final long DEFAULT_WAL_BATCH_SIZE
public static final String ASYNC_WAL_USE_SHARED_EVENT_LOOP
public static final boolean DEFAULT_ASYNC_WAL_USE_SHARED_EVENT_LOOP
public static final String ASYNC_WAL_WAIT_ON_SHUTDOWN_IN_SECONDS
public static final int DEFAULT_ASYNC_WAL_WAIT_ON_SHUTDOWN_IN_SECONDS
private final org.apache.hbase.thirdparty.io.netty.channel.EventLoopGroup eventLoopGroup
private final ExecutorService consumeExecutor
private final Class<? extends org.apache.hbase.thirdparty.io.netty.channel.Channel> channelClass
private final Lock consumeLock
private final Supplier<Boolean> hasConsumerTask
private static final int MAX_EPOCH
private volatile int epochAndState
private boolean readyForRolling
private final Condition readyForRollingCond
private final com.lmax.disruptor.RingBuffer<RingBufferTruck> waitingConsumePayloads
private final com.lmax.disruptor.Sequence waitingConsumePayloadsGatingSequence
private final AtomicBoolean consumerScheduled
private final long batchSize
private final ExecutorService closeExecutor
private volatile AsyncFSOutput fsOut
private final Deque<FSWALEntry> toWriteAppends
private final Deque<FSWALEntry> unackedAppends
private final SortedSet<SyncFuture> syncFutures
private long highestProcessedAppendTxid
private long fileLengthAtLastSync
private long highestProcessedAppendTxidAtLastSync
private final int waitOnShutdownInSeconds
public AsyncFSWAL(org.apache.hadoop.fs.FileSystem fs, org.apache.hadoop.fs.Path rootDir, String logDir, String archiveDir, org.apache.hadoop.conf.Configuration conf, List<WALActionsListener> listeners, boolean failIfWALExists, String prefix, String suffix, org.apache.hbase.thirdparty.io.netty.channel.EventLoopGroup eventLoopGroup, Class<? extends org.apache.hbase.thirdparty.io.netty.channel.Channel> channelClass) throws FailedLogCloseException, IOException
FailedLogCloseExceptionIOExceptionprivate static boolean waitingRoll(int epochAndState)
private static boolean writerBroken(int epochAndState)
private static int epoch(int epochAndState)
private boolean trySetReadyForRolling()
private void syncFailed(long epochWhenSync, Throwable error)
private void syncCompleted(WALProvider.AsyncWriter writer, long processedTxid, long startTimeNs)
private void sync(WALProvider.AsyncWriter writer)
private void addTimeAnnotation(SyncFuture future, String annotation)
private int finishSyncLowerThanTxid(long txid, boolean addSyncTrace)
private int finishSync(boolean addSyncTrace)
private void appendAndSync()
private void consume()
private boolean shouldScheduleConsumer()
public long append(RegionInfo hri, WALKeyImpl key, WALEdit edits, boolean inMemstore) throws IOException
AbstractFSWALwalKey parameter. Be warned that the WriteEntry is not
immediately available on return from this method. It WILL be available subsequent to a sync of
this append; otherwise, you will just have to wait on the WriteEntry to get filled in.append in interface WALappend in class AbstractFSWAL<WALProvider.AsyncWriter>hri - the regioninfo associated with appendkey - Modified by this call; we add to it this edits region edit/sequence id.edits - Edits to append. MAY CONTAIN NO EDITS for case where we want to get an edit
sequence id that is after all currently appended edits.inMemstore - Always true except for case where we are writing a compaction completion
record into the WAL; in this case the entry is just so we can finish an unfinished compaction
-- it is not an edit for memstore.key will have the region edit/sequence id
in it.IOExceptionpublic void sync() throws IOException
WALIOExceptionpublic void sync(long txid) throws IOException
WALtxid - Transaction id to sync to.IOExceptionprotected WALProvider.AsyncWriter createWriterInstance(org.apache.hadoop.fs.Path path) throws IOException
createWriterInstance in class AbstractFSWAL<WALProvider.AsyncWriter>IOExceptionprivate void waitForSafePoint()
private long closeWriter()
protected void doReplaceWriter(org.apache.hadoop.fs.Path oldPath, org.apache.hadoop.fs.Path newPath, WALProvider.AsyncWriter nextWriter) throws IOException
AbstractFSWALAbstractFSWAL.rollRequested flag in this method, as the new writer
will begin to work before returning from this method. If we clear the flag after returning from
this call, we may miss a roll request. The implementation class should choose a proper place to
clear the AbstractFSWAL.rollRequested flag so we do not miss a roll request, typically before you
start writing to the new writer.doReplaceWriter in class AbstractFSWAL<WALProvider.AsyncWriter>IOExceptionprotected void doShutdown() throws IOException
doShutdown in class AbstractFSWAL<WALProvider.AsyncWriter>IOExceptionprotected void doAppend(WALProvider.AsyncWriter writer, FSWALEntry entry)
doAppend in class AbstractFSWAL<WALProvider.AsyncWriter>org.apache.hadoop.hdfs.protocol.DatanodeInfo[] getPipeline()
AbstractFSWALgetPipeline in class AbstractFSWAL<WALProvider.AsyncWriter>int getLogReplication()
AbstractFSWALgetLogReplication in class AbstractFSWAL<WALProvider.AsyncWriter>protected boolean doCheckLogLowReplication()
doCheckLogLowReplication in class AbstractFSWAL<WALProvider.AsyncWriter>Copyright © 2007–2019 The Apache Software Foundation. All rights reserved.