Lucene Indexes

Introduction

In this example, two VMware GemFire servers host a single partitioned region with entries that represent employee information. The example indexes the first and last names of employees.

This example demonstrates the use of a simple Lucene index. Lucene provides a powerful text search and analysis.

In this example, two servers host a single partitioned region with entries that represent employee information. The example indexes the first and last names of employees.

This example assumes that JDK11 and GemFire are installed.

The minimum java version is jdk 11.

These example use the GemFire Search extension which requires GemFire 10 to work

Set up the Lucene index and region

Set directory gemfire-examples/lucene to be the current working directory. Each step in this example specifies paths relative to that directory.

Build the example (with EmployeeData class)

 $ ../gradlew build

Add VMware GemFire Search extension path to the GEMFIRE_EXTENSIONS_REPOSITORY_PATH environment variable. For example, if your vmware-gemfire-search-.gfm file is located in /gemfire-extensions, use the following command:

 $ export GEMFIRE_EXTENSIONS_REPOSITORY_PATH=/gemfire-extensions

Run a script that starts a locator and two servers, creates a Lucene index called simpleIndex, and then creates the example-region region. A Lucene index must be created before creating the region.

 $ gfsh run --file=scripts/start.gfsh

Run the example to populate both the Lucene index and example-region. The data will also be retrieved from the region and printed to the console.

 $ ../gradlew run

Try gfsh commands that interact with the region and do Lucene searches

Run a gfsh command to see the contents of the region

 $ gfsh
 ...
 gfsh>connect --locator=localhost[10334]
 gfsh>query --query="select * from /example-region"
 ...

Try different Lucene searches for data in example-region

 gfsh> list lucene indexes

Note that each server that holds partitioned data for this region has both the simpleIndex , analyzerIndex and the nestedObjectIndex. Each Lucene index is stored as a co-located region with the partitioned data region.

Search for an exact name match

gfsh>search lucene --name=simpleIndex --region=example-region --queryStrings="Jive" --defaultField=lastName

Search for last name using fuzzy logic: sounds like ‘chive’

gfsh>search lucene --name=simpleIndex --region=example-region --queryStrings="chive~" --defaultField=lastName

Do a compound search on first and last name using fuzzy sounds like logic

gfsh>search lucene --name=simpleIndex --region=example-region --queryStrings="firstName:cat~ OR lastName:chive~" --defaultField=lastName

Do a compound search on last name and email using analyzerIndex

gfsh>search lucene --name=analyzerIndex --region=example-region --queryStrings="lastName:hall~ AND email:Kris.Call\@example.com" --defaultField=lastName

Do a compound search on nested object with both 5035330001 AND 5036430001 in contacts Note: 5035330001 is the phone number of one of the contacts, 5036430001 is phone number of another contact. Since they are both contacts of this employee, it will lead to this employee.

gfsh>search lucene --name=nestedObjectIndex --region=/example-region --queryString="5035330001 AND 5036430001" --defaultField=contacts.phoneNumbers

If query on 5035330001 AND 5036430002, it will not find the person, because the 2 phone numbers belong to different people’s contacts.

gfsh>search lucene --name=nestedObjectIndex --region=/example-region --queryString="5035330001 AND 5036430002" --defaultField=contacts.phoneNumbers

If query on 5035330001 OR 5036430002, it will find 2 people’s entries

gfsh>search lucene --name=nestedObjectIndex --region=/example-region --queryString="5035330001 OR 5036430002" --defaultField=contacts.phoneNumbers

Examine the Lucene index statistics

 gfsh>describe lucene index --name=simpleIndex --region=example-region

Note the statistic show the fields that are indexed and the Lucene analyzer used for each field. In the next example we will specify a different Lucene analyzer for each field. Additional statistics listed are the number of documents (region entries) indexed, number of entries committed as well as the number of queries executed for each Lucene index.

Exit gfsh and shut down the cluster

gfsh>exit
 $ gfsh run --file=scripts/stop.gfsh

Clean up any generated directories and files so this example can be rerun.

 $ ../gradlew cleanServer