@InterfaceAudience.Private public class HStore extends Object implements Store
There's no reason to consider append-logging at this level; all logging and locking is handled at the HRegion level. Store just provides services to manage sets of StoreFiles. One of the most important of those services is compaction services where files are aggregated once they pass a configurable threshold.
The only thing having to do with logs that Store needs to deal with is the reconstructionLog. This is a segment of an HRegion's log that might NOT be present upon startup. If the param is NULL, there's nothing to do. If the param is non-NULL, we need to process the log to reconstruct a TreeMap that might not have been written to disk before the process died.
It's assumed that after this constructor returns, the reconstructionLog file will be deleted (by whoever has instantiated the Store).
Locking and transactions are handled at a higher level. This API should not be called directly but by an HRegion manager.
Modifier and Type | Field and Description |
---|---|
static String |
BLOCKING_STOREFILES_KEY |
static String |
COMPACTCHECKER_INTERVAL_MULTIPLIER_KEY |
static long |
DEEP_OVERHEAD |
static int |
DEFAULT_BLOCKING_STOREFILE_COUNT |
static int |
DEFAULT_COMPACTCHECKER_INTERVAL_MULTIPLIER |
static long |
FIXED_OVERHEAD |
protected MemStore |
memstore |
NO_PRIORITY, PRIORITY_USER
Modifier | Constructor and Description |
---|---|
protected |
HStore(HRegion region,
HColumnDescriptor family,
org.apache.hadoop.conf.Configuration confParam)
Constructor
|
Modifier and Type | Method and Description |
---|---|
long |
add(KeyValue kv)
Adds a value to the memstore
|
void |
addChangedReaderObserver(ChangedReadersObserver o) |
boolean |
areWritesEnabled() |
void |
assertBulkLoadHFileOk(org.apache.hadoop.fs.Path srcPath)
This throws a WrongRegionException if the HFile does not fit in this region, or an
InvalidHFileException if the HFile is not valid.
|
void |
bulkLoadHFile(String srcPathStr,
long seqNum)
This method should only be called from HRegion.
|
void |
cancelRequestedCompaction(CompactionContext compaction) |
boolean |
canSplit() |
com.google.common.collect.ImmutableCollection<StoreFile> |
close()
Close all the readers We don't need to worry about subsequent requests because the HRegion
holds a write lock that will prevent any more reads or writes.
|
List<StoreFile> |
compact(CompactionContext compaction)
Compact the StoreFiles.
|
void |
compactRecentForTestingAssumingDefaultPolicy(int N)
This method tries to compact N recent files for testing.
|
protected void |
completeCompaction(Collection<StoreFile> compactedFiles) |
void |
completeCompactionMarker(WALProtos.CompactionDescriptor compaction)
Call to complete a compaction.
|
org.apache.hadoop.hbase.regionserver.StoreFlushContext |
createFlushContext(long cacheFlushId) |
StoreFile.Writer |
createWriterInTmp(long maxKeyCount,
Compression.Algorithm compression,
boolean isCompaction,
boolean includeMVCCReadpoint,
boolean includesTag) |
protected long |
delete(KeyValue kv)
Adds a value to the memstore
|
void |
deleteChangedReaderObserver(ChangedReadersObserver o) |
protected List<org.apache.hadoop.fs.Path> |
flushCache(long logCacheFlushId,
SortedSet<KeyValue> snapshot,
TimeRangeTracker snapshotTimeRangeTracker,
AtomicLong flushedSize,
MonitoredTask status)
Write out current snapshot.
|
long |
getBlockingFileCount()
The number of files required before flushes for this store will be blocked.
|
static int |
getBytesPerChecksum(org.apache.hadoop.conf.Configuration conf)
Returns the configured bytesPerChecksum value.
|
CacheConfig |
getCacheConfig()
Used for tests.
|
static ChecksumType |
getChecksumType(org.apache.hadoop.conf.Configuration conf)
Returns the configured checksum algorithm.
|
static int |
getCloseCheckInterval() |
String |
getColumnFamilyName() |
long |
getCompactedCellsCount() |
long |
getCompactedCellsSize() |
long |
getCompactionCheckMultiplier() |
CompactionProgress |
getCompactionProgress()
getter for CompactionProgress object
|
int |
getCompactPriority() |
KeyValue.KVComparator |
getComparator() |
RegionCoprocessorHost |
getCoprocessorHost() |
HFileDataBlockEncoder |
getDataBlockEncoder() |
HColumnDescriptor |
getFamily() |
org.apache.hadoop.fs.FileSystem |
getFileSystem() |
long |
getFlushableSize() |
long |
getFlushedCellsCount() |
long |
getFlushedCellsSize() |
HRegion |
getHRegion() |
long |
getLastCompactSize() |
long |
getMajorCompactedCellsCount() |
long |
getMajorCompactedCellsSize() |
long |
getMaxMemstoreTS() |
long |
getMemstoreFlushSize() |
long |
getMemStoreSize() |
HRegionFileSystem |
getRegionFileSystem() |
HRegionInfo |
getRegionInfo() |
KeyValue |
getRowKeyAtOrBefore(byte[] row)
Find the key that matches row exactly, or the one that immediately precedes it.
|
ScanInfo |
getScanInfo() |
KeyValueScanner |
getScanner(Scan scan,
NavigableSet<byte[]> targetCols,
long readPt)
Return a scanner for both the memstore and the HStore files.
|
List<KeyValueScanner> |
getScanners(boolean cacheBlocks,
boolean isGet,
boolean usePread,
boolean isCompaction,
ScanQueryMatcher matcher,
byte[] startRow,
byte[] stopRow,
long readPt)
Get all scanners with no filtering based on TTL (that happens further down
the line).
|
long |
getSize() |
long |
getSmallestReadPoint() |
byte[] |
getSplitPoint()
Determines if Store should be split
|
Collection<StoreFile> |
getStorefiles() |
int |
getStorefilesCount() |
long |
getStorefilesIndexSize() |
long |
getStorefilesSize() |
long |
getStoreFileTtl() |
static org.apache.hadoop.fs.Path |
getStoreHomedir(org.apache.hadoop.fs.Path tabledir,
HRegionInfo hri,
byte[] family)
Deprecated.
|
static org.apache.hadoop.fs.Path |
getStoreHomedir(org.apache.hadoop.fs.Path tabledir,
String encodedName,
byte[] family)
Deprecated.
|
long |
getStoreSizeUncompressed() |
TableName |
getTableName() |
long |
getTotalStaticBloomSize()
Returns the total byte size of all Bloom filter bit arrays.
|
long |
getTotalStaticIndexSize()
Returns the total size of all index blocks in the data block indexes, including the root level,
intermediate levels, and the leaf level for multi-level indexes, or just the root level for
single-level indexes.
|
boolean |
hasReferences() |
boolean |
hasTooManyStoreFiles() |
long |
heapSize() |
boolean |
isMajorCompaction() |
boolean |
needsCompaction()
See if there's too much store files in this store
|
CompactionContext |
requestCompaction() |
CompactionContext |
requestCompaction(int priority,
CompactionRequest baseRequest) |
void |
rollback(KeyValue kv)
Removes a kv from the memstore.
|
boolean |
throttleCompaction(long compactionSize) |
long |
timeOfOldestEdit()
When was the last edit done in the memstore
|
String |
toString() |
void |
triggerMajorCompaction() |
long |
updateColumnValue(byte[] row,
byte[] f,
byte[] qualifier,
long newValue)
Used in tests.
|
long |
upsert(Iterable<Cell> cells,
long readpoint)
Adds or replaces the specified KeyValues.
|
public static final String COMPACTCHECKER_INTERVAL_MULTIPLIER_KEY
public static final String BLOCKING_STOREFILES_KEY
public static final int DEFAULT_COMPACTCHECKER_INTERVAL_MULTIPLIER
public static final int DEFAULT_BLOCKING_STOREFILE_COUNT
protected final MemStore memstore
public static final long FIXED_OVERHEAD
public static final long DEEP_OVERHEAD
protected HStore(HRegion region, HColumnDescriptor family, org.apache.hadoop.conf.Configuration confParam) throws IOException
region
- family
- HColumnDescriptor for this columnconfParam
- configuration object
failed. Can be null.IOException
public String getColumnFamilyName()
getColumnFamilyName
in interface Store
public TableName getTableName()
getTableName
in interface Store
public org.apache.hadoop.fs.FileSystem getFileSystem()
getFileSystem
in interface Store
public HRegionFileSystem getRegionFileSystem()
public long getStoreFileTtl()
getStoreFileTtl
in interface StoreConfigInformation
public long getMemstoreFlushSize()
getMemstoreFlushSize
in interface StoreConfigInformation
public long getFlushableSize()
getFlushableSize
in interface Store
Store.getMemStoreSize()
unless we are carrying snapshots and then it will be the size of
outstanding snapshots.public long getCompactionCheckMultiplier()
getCompactionCheckMultiplier
in interface StoreConfigInformation
public long getBlockingFileCount()
StoreConfigInformation
getBlockingFileCount
in interface StoreConfigInformation
public static int getBytesPerChecksum(org.apache.hadoop.conf.Configuration conf)
conf
- The configurationpublic static ChecksumType getChecksumType(org.apache.hadoop.conf.Configuration conf)
conf
- The configurationpublic static int getCloseCheckInterval()
public HColumnDescriptor getFamily()
public long getMaxMemstoreTS()
getMaxMemstoreTS
in interface Store
@Deprecated public static org.apache.hadoop.fs.Path getStoreHomedir(org.apache.hadoop.fs.Path tabledir, HRegionInfo hri, byte[] family)
tabledir
- Path
to where the table is being storedhri
- HRegionInfo
for the region.family
- HColumnDescriptor
describing the column family@Deprecated public static org.apache.hadoop.fs.Path getStoreHomedir(org.apache.hadoop.fs.Path tabledir, String encodedName, byte[] family)
tabledir
- Path
to where the table is being storedencodedName
- Encoded region name.family
- HColumnDescriptor
describing the column familypublic HFileDataBlockEncoder getDataBlockEncoder()
getDataBlockEncoder
in interface Store
public long add(KeyValue kv)
Store
public long timeOfOldestEdit()
Store
timeOfOldestEdit
in interface Store
protected long delete(KeyValue kv)
kv
- public void rollback(KeyValue kv)
Store
public Collection<StoreFile> getStorefiles()
getStorefiles
in interface Store
public void assertBulkLoadHFileOk(org.apache.hadoop.fs.Path srcPath) throws IOException
Store
assertBulkLoadHFileOk
in interface Store
IOException
public void bulkLoadHFile(String srcPathStr, long seqNum) throws IOException
Store
bulkLoadHFile
in interface Store
seqNum
- sequence Id associated with the HFileIOException
public com.google.common.collect.ImmutableCollection<StoreFile> close() throws IOException
Store
close
in interface Store
StoreFiles
that were previously being used.IOException
- on failureprotected List<org.apache.hadoop.fs.Path> flushCache(long logCacheFlushId, SortedSet<KeyValue> snapshot, TimeRangeTracker snapshotTimeRangeTracker, AtomicLong flushedSize, MonitoredTask status) throws IOException
snapshot()
has been called
previously.logCacheFlushId
- flush sequence numbersnapshot
- snapshotTimeRangeTracker
- flushedSize
- The number of bytes flushedstatus
- IOException
public StoreFile.Writer createWriterInTmp(long maxKeyCount, Compression.Algorithm compression, boolean isCompaction, boolean includeMVCCReadpoint, boolean includesTag) throws IOException
createWriterInTmp
in interface Store
IOException
public List<KeyValueScanner> getScanners(boolean cacheBlocks, boolean isGet, boolean usePread, boolean isCompaction, ScanQueryMatcher matcher, byte[] startRow, byte[] stopRow, long readPt) throws IOException
getScanners
in interface Store
IOException
public void addChangedReaderObserver(ChangedReadersObserver o)
addChangedReaderObserver
in interface Store
public void deleteChangedReaderObserver(ChangedReadersObserver o)
deleteChangedReaderObserver
in interface Store
public List<StoreFile> compact(CompactionContext compaction) throws IOException
During this time, the Store can work as usual, getting values from StoreFiles and writing new StoreFiles from the memstore. Existing StoreFiles are not destroyed until the new compacted StoreFile is completely written-out to disk.
The compactLock prevents multiple simultaneous compactions. The structureLock prevents us from interfering with other write operations.
We don't want to hold the structureLock for the whole time, as a compact() can be lengthy and we want to allow cache-flushes during this period.
Compaction event should be idempotent, since there is no IO Fencing for the region directory in hdfs. A region server might still try to complete the compaction after it lost the region. That is why the following events are carefully ordered for a compaction: 1. Compaction writes new files under region/.tmp directory (compaction output) 2. Compaction atomically moves the temporary file under region directory 3. Compaction appends a WAL edit containing the compaction input and output files. Forces sync on WAL. 4. Compaction deletes the input files from the region directory. Failure conditions are handled like this: - If RS fails before 2, compaction wont complete. Even if RS lives on and finishes the compaction later, it will only write the new data file to the region directory. Since we already have this data, this will be idempotent but we will have a redundant copy of the data. - If RS fails between 2 and 3, the region will have a redundant copy of the data. The RS that failed won't be able to finish snyc() for WAL because of lease recovery in WAL. - If RS fails after 3, the region region server who opens the region will pick up the the compaction marker from the WAL and replay it by removing the compaction input files. Failed RS can also attempt to delete those files, but the operation will be idempotent See HBASE-2231 for details.
compact
in interface Store
compaction
- compaction details obtained from requestCompaction()IOException
public void completeCompactionMarker(WALProtos.CompactionDescriptor compaction) throws IOException
completeCompactionMarker
in interface Store
compaction
- IOException
public void compactRecentForTestingAssumingDefaultPolicy(int N) throws IOException
N
- Number of files.IOException
public boolean hasReferences()
hasReferences
in interface Store
public CompactionProgress getCompactionProgress()
Store
getCompactionProgress
in interface Store
public boolean isMajorCompaction() throws IOException
isMajorCompaction
in interface Store
IOException
public CompactionContext requestCompaction() throws IOException
requestCompaction
in interface Store
IOException
public CompactionContext requestCompaction(int priority, CompactionRequest baseRequest) throws IOException
requestCompaction
in interface Store
IOException
public void cancelRequestedCompaction(CompactionContext compaction)
cancelRequestedCompaction
in interface Store
protected void completeCompaction(Collection<StoreFile> compactedFiles) throws IOException
IOException
public KeyValue getRowKeyAtOrBefore(byte[] row) throws IOException
Store
getRowKeyAtOrBefore
in interface Store
row
- The row key of the targeted row.IOException
public byte[] getSplitPoint()
Store
getSplitPoint
in interface Store
public long getLastCompactSize()
getLastCompactSize
in interface Store
public long getSize()
public void triggerMajorCompaction()
triggerMajorCompaction
in interface Store
public KeyValueScanner getScanner(Scan scan, NavigableSet<byte[]> targetCols, long readPt) throws IOException
Store
getScanner
in interface Store
scan
- Scan to apply when scanning the storestargetCols
- columns to scanIOException
- on failurepublic int getStorefilesCount()
getStorefilesCount
in interface Store
public long getStoreSizeUncompressed()
getStoreSizeUncompressed
in interface Store
public long getStorefilesSize()
getStorefilesSize
in interface Store
public long getStorefilesIndexSize()
getStorefilesIndexSize
in interface Store
public long getTotalStaticIndexSize()
Store
getTotalStaticIndexSize
in interface Store
public long getTotalStaticBloomSize()
Store
getTotalStaticBloomSize
in interface Store
public long getMemStoreSize()
getMemStoreSize
in interface Store
public int getCompactPriority()
getCompactPriority
in interface Store
public boolean throttleCompaction(long compactionSize)
throttleCompaction
in interface Store
public HRegion getHRegion()
public RegionCoprocessorHost getCoprocessorHost()
getCoprocessorHost
in interface Store
public HRegionInfo getRegionInfo()
getRegionInfo
in interface Store
public boolean areWritesEnabled()
areWritesEnabled
in interface Store
public long getSmallestReadPoint()
getSmallestReadPoint
in interface Store
public long updateColumnValue(byte[] row, byte[] f, byte[] qualifier, long newValue) throws IOException
row
- row to updatef
- family to updatequalifier
- qualifier to updatenewValue
- the new value to set into memstoreIOException
public long upsert(Iterable<Cell> cells, long readpoint) throws IOException
Store
For each KeyValue specified, if a cell with the same row, family, and qualifier exists in MemStore, it will be replaced. Otherwise, it will just be inserted to MemStore.
This operation is atomic on each KeyValue (row/family/qualifier) but not necessarily atomic across all of them.
upsert
in interface Store
readpoint
- readpoint below which we can safely remove duplicate KVsIOException
public org.apache.hadoop.hbase.regionserver.StoreFlushContext createFlushContext(long cacheFlushId)
createFlushContext
in interface Store
public boolean needsCompaction()
Store
needsCompaction
in interface Store
public CacheConfig getCacheConfig()
Store
getCacheConfig
in interface Store
public long heapSize()
public KeyValue.KVComparator getComparator()
getComparator
in interface Store
public ScanInfo getScanInfo()
getScanInfo
in interface Store
public boolean hasTooManyStoreFiles()
hasTooManyStoreFiles
in interface Store
public long getFlushedCellsCount()
getFlushedCellsCount
in interface Store
public long getFlushedCellsSize()
getFlushedCellsSize
in interface Store
public long getCompactedCellsCount()
getCompactedCellsCount
in interface Store
public long getCompactedCellsSize()
getCompactedCellsSize
in interface Store
public long getMajorCompactedCellsCount()
getMajorCompactedCellsCount
in interface Store
public long getMajorCompactedCellsSize()
getMajorCompactedCellsSize
in interface Store
Copyright © 2014 The Apache Software Foundation. All rights reserved.