[OAK-9091] oak-search-elastic: review query code

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.32.0
• Component/s: elastic-search, indexing
• Assignee: Fabrizio Fortino
• Reporter: Fabrizio Fortino
• Created: 25/May/20 08:15
• Updated: 07/Sep/20 10:22
• Resolved: 12/Jun/20 05:43

Description

The code to execute queries in the elastic module needs to be improved.

It has been inherited from the first prototype but has never been reviewed/refactored.

Here is what needs to be done:

• ElasticsearchRowIterator [0] is the central point. We need to break it up and separate the query building phase from the interpretation of the response
• Centralize how we resolve property names. This affects also mapping and indexing phases
• We need to review how the queries are composed to increase performance. As a rule of thumb, we should always create a BoolQuery and leverage as much as possible the use of filters over matching queries. This allows us to use ES caching efficiently
• Increase test coverage. Most of the testing in this area requires ES integration. We should add more unit tests checking that input query plans generate the expected ES requests.

[0] https://github.com/oak-indexing/jackrabbit-oak/blob/8cc86c4d142004bf2550cf2ac66400b54093209d/oak-search-elastic/src/main/java/org/apache/jackrabbit/oak/plugins/index/elasticsearch/query/ElasticsearchResultRowIterator.java

Issue Links

• links to: GitHub Pull Request #232