org.elasticsearch.index.fielddata.plain

Class PagedBytesIndexFieldData.PagedBytesEstimator

java.lang.Object

org.elasticsearch.index.fielddata.plain.PagedBytesIndexFieldData.PagedBytesEstimator

All Implemented Interfaces:

AbstractIndexFieldData.PerValueEstimator

Enclosing class:

PagedBytesIndexFieldData

public class PagedBytesIndexFieldData.PagedBytesEstimator
extends java.lang.Object
implements AbstractIndexFieldData.PerValueEstimator

Estimator that wraps string field data by either using BlockTreeTermsReader, or wrapping the data in a RamAccountingTermsEnum if the BlockTreeTermsReader cannot be used.

Method Summary

Modifier and Type	Method and Description
void	adjustForNoTerms(long actualUsed) Adjust the breaker when no terms were actually loaded, but the field data takes up space regardless.
void	afterLoad(org.apache.lucene.index.TermsEnum termsEnum, long actualUsed) Adjust the circuit breaker now that terms have been loaded, getting the actual used either from the parameter (if estimation worked for the entire set), or from the TermsEnum if it has been wrapped in a RamAccountingTermsEnum.
org.apache.lucene.index.TermsEnum	beforeLoad(org.apache.lucene.index.Terms terms) Determine whether the BlockTreeTermsReader.FieldReader can be used for estimating the field data, adding the estimate to the circuit breaker if it can, otherwise wrapping the terms in a RamAccountingTermsEnum to be estimated on a per-term basis.
long	bytesPerValue(org.apache.lucene.util.BytesRef term)
long	estimateStringFieldData()

Methods inherited from class java.lang.Object

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

Method Detail

bytesPerValue

public long bytesPerValue(org.apache.lucene.util.BytesRef term)

Specified by:

bytesPerValue in interface AbstractIndexFieldData.PerValueEstimator

Returns:

the number of bytes for the term based on the length and ordinal overhead

estimateStringFieldData

public long estimateStringFieldData()

Returns:

the estimate for loading the entire term set into field data, or 0 if unavailable

beforeLoad

public org.apache.lucene.index.TermsEnum beforeLoad(org.apache.lucene.index.Terms terms)
                                             throws java.io.IOException

Determine whether the BlockTreeTermsReader.FieldReader can be used for estimating the field data, adding the estimate to the circuit breaker if it can, otherwise wrapping the terms in a RamAccountingTermsEnum to be estimated on a per-term basis.

Specified by:

beforeLoad in interface AbstractIndexFieldData.PerValueEstimator

Parameters:

terms - terms to be estimated

Returns:

A possibly wrapped TermsEnum for the terms

Throws:

java.io.IOException

afterLoad

public void afterLoad(org.apache.lucene.index.TermsEnum termsEnum,
                      long actualUsed)

Adjust the circuit breaker now that terms have been loaded, getting the actual used either from the parameter (if estimation worked for the entire set), or from the TermsEnum if it has been wrapped in a RamAccountingTermsEnum.

Specified by:

afterLoad in interface AbstractIndexFieldData.PerValueEstimator

Parameters:

termsEnum - terms that were loaded

actualUsed - actual field data memory usage

adjustForNoTerms

public void adjustForNoTerms(long actualUsed)

Adjust the breaker when no terms were actually loaded, but the field data takes up space regardless. For instance, when ordinals are used.

Parameters:

actualUsed - bytes actually used