org.apache.hadoop.hive.common

Class ValidReaderWriteIdList

java.lang.Object

org.apache.hadoop.hive.common.ValidReaderWriteIdList

All Implemented Interfaces:

ValidWriteIdList

Direct Known Subclasses:

ValidCompactorWriteIdList

public class ValidReaderWriteIdList
extends Object
implements ValidWriteIdList

An implementation of ValidWriteIdList for use by readers. This class will view a write id as valid only if it maps to committed transaction. Write ids of both open and aborted transactions will be seen as invalid.

Nested Class Summary

Nested classes/interfaces inherited from interface org.apache.hadoop.hive.common.ValidWriteIdList

ValidWriteIdList.RangeResponse

Field Summary

Fields

Modifier and Type	Field and Description
protected BitSet	abortedBits
protected long[]	exceptions
protected long	highWatermark

Fields inherited from interface org.apache.hadoop.hive.common.ValidWriteIdList

VALID_WRITEIDS_KEY

Constructor Summary

Constructors

Constructor and Description
ValidReaderWriteIdList() This seems like a bad c'tor.
ValidReaderWriteIdList(String value)
ValidReaderWriteIdList(String tableName, long[] exceptions, BitSet abortedBits, long highWatermark) Used if there are no open write ids in the snapshot.
ValidReaderWriteIdList(String tableName, long[] exceptions, BitSet abortedBits, long highWatermark, long minOpenWriteId)

Method Summary

Modifier and Type	Method and Description
long	getHighWatermark() Get the largest write id used.
long[]	getInvalidWriteIds() Get the list of write ids under the high water mark that are not valid.
Long	getMinOpenWriteId() The smallest open write id.
String	getTableName() Get the table for which the ValidWriteIdList is formed
boolean	isValidBase(long writeId) We cannot use a base file if its range contains an open write id.
boolean	isWriteIdAborted(long writeId) Indicates whether a given write maps to aborted transaction.
ValidWriteIdList.RangeResponse	isWriteIdRangeAborted(long minWriteId, long maxWriteId) Find out if a range of write ids are aborted.
ValidWriteIdList.RangeResponse	isWriteIdRangeValid(long minWriteId, long maxWriteId) Find out if a range of write ids are valid.
boolean	isWriteIdValid(long writeId) Indicates whether a given write ID is valid.
void	readFromString(String src) Populate this ValidWriteIdList from the string.
String	toString()
ValidReaderWriteIdList	updateHighWatermark(long value)
String	writeToString() Write this ValidWriteIdList into a string.

Methods inherited from class java.lang.Object

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

Field Detail

exceptions

protected long[] exceptions

abortedBits

protected BitSet abortedBits

highWatermark

protected long highWatermark

Constructor Detail

ValidReaderWriteIdList

public ValidReaderWriteIdList()

This seems like a bad c'tor. It doesn't even have a table name in it and it's used every time ValidWriteIdList.VALID_WRITEIDS_KEY is not found in Configuration. But, if anything, that would indicate a bug if was done for an acid read since it considers everything valid - this should not be assumed.

ValidReaderWriteIdList

public ValidReaderWriteIdList(String tableName,
                              long[] exceptions,
                              BitSet abortedBits,
                              long highWatermark)

Used if there are no open write ids in the snapshot.

ValidReaderWriteIdList

public ValidReaderWriteIdList(String tableName,
                              long[] exceptions,
                              BitSet abortedBits,
                              long highWatermark,
                              long minOpenWriteId)

ValidReaderWriteIdList

public ValidReaderWriteIdList(String value)

Method Detail

isWriteIdValid

public boolean isWriteIdValid(long writeId)

Description copied from interface: ValidWriteIdList

Indicates whether a given write ID is valid. Note that valid may have different meanings for different implementations, as some will only want to see committed transactions and some both committed and aborted.

Specified by:

isWriteIdValid in interface ValidWriteIdList

Parameters:

writeId - write ID of the table

Returns:

true if valid, false otherwise

isValidBase

public boolean isValidBase(long writeId)

We cannot use a base file if its range contains an open write id.

Specified by:

isValidBase in interface ValidWriteIdList

Parameters:

writeId - from base_xxxx

Returns:

true if the base file can be used

isWriteIdRangeValid

public ValidWriteIdList.RangeResponse isWriteIdRangeValid(long minWriteId,
                                                          long maxWriteId)

Description copied from interface: ValidWriteIdList

Find out if a range of write ids are valid. Note that valid may have different meanings for different implementations, as some will only want to see committed transactions and some both committed and aborted.

Specified by:

isWriteIdRangeValid in interface ValidWriteIdList

Parameters:

minWriteId - minimum write ID to look for, inclusive

maxWriteId - maximum write ID to look for, inclusive

Returns:

Indicate whether none, some, or all of these transactions are valid.

toString

public String toString()

Overrides:

toString in class Object

writeToString

public String writeToString()

Description copied from interface: ValidWriteIdList

Write this ValidWriteIdList into a string. This should produce a string that can be used by ValidWriteIdList.readFromString(String) to populate a ValidWriteIdList.

Specified by:

writeToString in interface ValidWriteIdList

Returns:

the list as a string

readFromString

public void readFromString(String src)

Description copied from interface: ValidWriteIdList

Populate this ValidWriteIdList from the string. It is assumed that the string was created via ValidWriteIdList.writeToString() and the exceptions list is sorted.

Specified by:

readFromString in interface ValidWriteIdList

Parameters:

src - source string.

getTableName

public String getTableName()

Description copied from interface: ValidWriteIdList

Get the table for which the ValidWriteIdList is formed

Specified by:

getTableName in interface ValidWriteIdList

Returns:

table name (<db_name>.<table_name>) associated with ValidWriteIdList.

getHighWatermark

public long getHighWatermark()

Description copied from interface: ValidWriteIdList

Get the largest write id used.

Specified by:

getHighWatermark in interface ValidWriteIdList

Returns:

largest write id used

getInvalidWriteIds

public long[] getInvalidWriteIds()

Description copied from interface: ValidWriteIdList

Get the list of write ids under the high water mark that are not valid. Note that invalid may have different meanings for different implementations, as some will only want to see open transactions and some both open and aborted.

Specified by:

getInvalidWriteIds in interface ValidWriteIdList

Returns:

a list of invalid write ids

getMinOpenWriteId

public Long getMinOpenWriteId()

Description copied from interface: ValidWriteIdList

The smallest open write id.

Specified by:

getMinOpenWriteId in interface ValidWriteIdList

Returns:

smallest Open write Id in this set, null if there is none.

isWriteIdAborted

public boolean isWriteIdAborted(long writeId)

Description copied from interface: ValidWriteIdList

Indicates whether a given write maps to aborted transaction.

Specified by:

isWriteIdAborted in interface ValidWriteIdList

Parameters:

writeId - write id to be validated

Returns:

true if aborted, false otherwise

isWriteIdRangeAborted

public ValidWriteIdList.RangeResponse isWriteIdRangeAborted(long minWriteId,
                                                            long maxWriteId)

Description copied from interface: ValidWriteIdList

Find out if a range of write ids are aborted.

Specified by:

isWriteIdRangeAborted in interface ValidWriteIdList

Parameters:

minWriteId - minimum write Id to look for, inclusive

maxWriteId - maximum write Id to look for, inclusive

Returns:

Indicate whether none, some, or all of these write ids are aborted.

updateHighWatermark

public ValidReaderWriteIdList updateHighWatermark(long value)