org.apache.cassandra.db.index

Class SecondaryIndexManager

java.lang.Object

org.apache.cassandra.db.index.SecondaryIndexManager

public class SecondaryIndexManager
extends java.lang.Object

Manages all the indexes associated with a given CFS Different types of indexes can be created across the same CF

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static interface	SecondaryIndexManager.Updater

Field Summary

Fields

Modifier and Type	Field and Description
ColumnFamilyStore	baseCfs The underlying column family containing the source data for these indexes
static SecondaryIndexManager.Updater	nullUpdater

Constructor Summary

Constructors

Constructor and Description
SecondaryIndexManager(ColumnFamilyStore baseCfs)

Method Summary

Methods

Modifier and Type	Method and Description
java.util.concurrent.Future<?>	addIndexedColumn(ColumnDefinition cdef) Adds and builds a index for a column
java.util.Set<java.lang.String>	allIndexesNames()
void	deleteFromIndexes(DecoratedKey key, java.util.List<Cell> indexedColumnsInRow, OpOrder.Group opGroup) Delete all columns from all indexes for this row.
void	flushIndexesBlocking() Flush all indexes to disk
SecondaryIndexManager.Updater	gcUpdaterFor(DecoratedKey key) Updated closure with only the modified row key.
java.util.List<java.lang.String>	getBuiltIndexes()
SecondaryIndexSearcher	getHighestSelectivityIndexSearcher(java.util.List<IndexExpression> clause)
SecondaryIndex	getIndexByName(java.lang.String name)
java.util.Collection<SecondaryIndex>	getIndexes()
java.util.Set<ColumnFamilyStore>	getIndexesBackedByCfs()
java.util.Set<SecondaryIndex>	getIndexesByNames(java.util.Set<java.lang.String> idxNames)
java.util.Set<SecondaryIndex>	getIndexesNotBackedByCfs()
SecondaryIndex	getIndexForColumn(java.nio.ByteBuffer column)
java.util.List<SecondaryIndexSearcher>	getIndexSearchersForQuery(java.util.List<IndexExpression> clause) Get a list of IndexSearchers from the union of expression index types
boolean	hasIndexes()
boolean	hasIndexFor(java.util.List<IndexExpression> clause)
boolean	indexes(Cell cell)
boolean	indexes(CellName name)
boolean	indexes(CellName name, java.util.Collection<SecondaryIndex> indexes)
java.util.Set<SecondaryIndex>	indexFor(CellName name)
java.util.Set<SecondaryIndex>	indexFor(CellName name, java.util.Collection<SecondaryIndex> indexes)
void	indexRow(java.nio.ByteBuffer key, ColumnFamily cf, OpOrder.Group opGroup) When building an index against existing data, add the given row to the index
void	invalidate() Remove the index
void	maybeBuildSecondaryIndexes(java.util.Collection<SSTableReader> sstables, java.util.Set<java.lang.String> idxNames) Does a full, blocking rebuild of the indexes specified by columns from the sstables.
void	reload() Drops and adds new indexes associated with the underlying CF
void	removeIndexedColumn(java.nio.ByteBuffer column) Removes a existing index
java.util.List<Row>	search(ExtendedFilter filter) Performs a search across a number of column indexes
void	setIndexBuilt(java.util.Set<java.lang.String> idxNames)
void	setIndexRemoved(java.util.Set<java.lang.String> idxNames)
SecondaryIndexManager.Updater	updaterFor(DecoratedKey key, ColumnFamily cf, OpOrder.Group opGroup) This helper acts as a closure around the indexManager and updated cf data to ensure that down in Memtable's ColumnFamily implementation, the index can get updated.
SecondaryIndex	validate(java.nio.ByteBuffer rowKey, Cell cell)
void	validateIndexSearchersForQuery(java.util.List<IndexExpression> clause) Validates an union of expression index types.
void	validateRowLevelIndexes(java.nio.ByteBuffer key, ColumnFamily cf)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

nullUpdater

public static final SecondaryIndexManager.Updater nullUpdater

baseCfs

public final ColumnFamilyStore baseCfs

The underlying column family containing the source data for these indexes

Constructor Detail

SecondaryIndexManager

public SecondaryIndexManager(ColumnFamilyStore baseCfs)

Method Detail

reload

public void reload()

Drops and adds new indexes associated with the underlying CF

allIndexesNames

public java.util.Set<java.lang.String> allIndexesNames()

maybeBuildSecondaryIndexes

public void maybeBuildSecondaryIndexes(java.util.Collection<SSTableReader> sstables,
                              java.util.Set<java.lang.String> idxNames)

Does a full, blocking rebuild of the indexes specified by columns from the sstables. Does nothing if columns is empty. Caller must acquire and release references to the sstables used here.

Parameters:

sstables - the data to build from

idxNames - the list of columns to index, ordered by comparator

indexes

public boolean indexes(CellName name,
              java.util.Collection<SecondaryIndex> indexes)

indexFor

public java.util.Set<SecondaryIndex> indexFor(CellName name,
                                     java.util.Collection<SecondaryIndex> indexes)

indexes

public boolean indexes(Cell cell)

indexes

public boolean indexes(CellName name)

indexFor

public java.util.Set<SecondaryIndex> indexFor(CellName name)

hasIndexFor

public boolean hasIndexFor(java.util.List<IndexExpression> clause)

Returns:

true if at least one of the indexes can handle the clause.

removeIndexedColumn

public void removeIndexedColumn(java.nio.ByteBuffer column)

Removes a existing index

Parameters:

column - the indexed column to remove

addIndexedColumn

public java.util.concurrent.Future<?> addIndexedColumn(ColumnDefinition cdef)

Adds and builds a index for a column

Parameters:

cdef - the column definition holding the index data

Returns:

a future which the caller can optionally block on signaling the index is built

getIndexForColumn

public SecondaryIndex getIndexForColumn(java.nio.ByteBuffer column)

Parameters:

column - the name of indexes column

Returns:

the index

invalidate

public void invalidate()

Remove the index

flushIndexesBlocking

public void flushIndexesBlocking()

Flush all indexes to disk

getBuiltIndexes

public java.util.List<java.lang.String> getBuiltIndexes()

Returns:

all built indexes (ready to use)

getIndexesBackedByCfs

public java.util.Set<ColumnFamilyStore> getIndexesBackedByCfs()

Returns:

all CFS from indexes which use a backing CFS internally (KEYS)

getIndexesNotBackedByCfs

public java.util.Set<SecondaryIndex> getIndexesNotBackedByCfs()

Returns:

all indexes which do *not* use a backing CFS internally

getIndexes

public java.util.Collection<SecondaryIndex> getIndexes()

Returns:

all of the secondary indexes without distinction to the (non-)backed by secondary ColumnFamilyStore.

getIndexByName

public SecondaryIndex getIndexByName(java.lang.String name)

hasIndexes

public boolean hasIndexes()

Returns:

if there are ANY indexes for this table..

indexRow

public void indexRow(java.nio.ByteBuffer key,
            ColumnFamily cf,
            OpOrder.Group opGroup)

When building an index against existing data, add the given row to the index

Parameters:

key - the row key

cf - the current rows data

deleteFromIndexes

public void deleteFromIndexes(DecoratedKey key,
                     java.util.List<Cell> indexedColumnsInRow,
                     OpOrder.Group opGroup)

Delete all columns from all indexes for this row. For when cleanup rips a row out entirely.

Parameters:

key - the row key

indexedColumnsInRow - all column names in row

updaterFor

public SecondaryIndexManager.Updater updaterFor(DecoratedKey key,
                                       ColumnFamily cf,
                                       OpOrder.Group opGroup)

This helper acts as a closure around the indexManager and updated cf data to ensure that down in Memtable's ColumnFamily implementation, the index can get updated. Note: only a CF backed by AtomicSortedColumns implements this behaviour fully, other types simply ignore the index updater.

gcUpdaterFor

public SecondaryIndexManager.Updater gcUpdaterFor(DecoratedKey key)

Updated closure with only the modified row key.

getIndexSearchersForQuery

public java.util.List<SecondaryIndexSearcher> getIndexSearchersForQuery(java.util.List<IndexExpression> clause)

Get a list of IndexSearchers from the union of expression index types

Parameters:

clause - the query clause

Returns:

the searchers needed to query the index

validateIndexSearchersForQuery

public void validateIndexSearchersForQuery(java.util.List<IndexExpression> clause)
                                    throws InvalidRequestException

Validates an union of expression index types. It will throw a RuntimeException if any of the expressions in the provided clause is not valid for its index implementation.

Parameters:

clause - the query clause

Throws:

InvalidRequestException - in case of validation errors

search

public java.util.List<Row> search(ExtendedFilter filter)

Performs a search across a number of column indexes

Parameters:

filter - the column range to restrict to

Returns:

found indexed rows

getIndexesByNames

public java.util.Set<SecondaryIndex> getIndexesByNames(java.util.Set<java.lang.String> idxNames)

setIndexBuilt

public void setIndexBuilt(java.util.Set<java.lang.String> idxNames)

setIndexRemoved

public void setIndexRemoved(java.util.Set<java.lang.String> idxNames)

validate

public SecondaryIndex validate(java.nio.ByteBuffer rowKey,
                      Cell cell)

validateRowLevelIndexes

public void validateRowLevelIndexes(java.nio.ByteBuffer key,
                           ColumnFamily cf)
                             throws InvalidRequestException

Throws:

InvalidRequestException

getHighestSelectivityIndexSearcher

public SecondaryIndexSearcher getHighestSelectivityIndexSearcher(java.util.List<IndexExpression> clause)