Error in indexing polygon data in Elasticsearch 8.8

[amal_antony]

Greetings to the community!

I am experiencing issues while ingesting polygon data into Elasticsearch 8.8. An issue had been raised before in the same context, link. This was identified as a bug in Lucene, the fix for which was integrated to Elasticsearch version 8.7 and later. Even with ES 8.8, we are experiencing issues.
The shape has been validated in geojson.io.

Sharing the sample polygon data.
https://raw.githubusercontent.com/amalantonygit/polygon-data/main/polygon.geojson

Would appreciate any help in identifying the underlying issue.
Thanks in advance

[Ignacio_Vera]

hello,

yes, this is an issue on lucene side. The algorithm that breaks the polygon (Tessellator) in order to index it is not able to handle those polygons. I had a quick look and I could not find anything obvious.

Would you like to open an issue in the lucene repository (GitHub - apache/lucene: Apache Lucene open-source search software) so we can take a deeper look?

[Ignacio_Vera]

I open Polygons failing to tessellate · Issue #12352 · apache/lucene · GitHub

[amal_antony]

Thanks for creating that @Ignacio_Vera.

Let me know if you need any additional info for debugging.

[amal_antony]

Hi @Ignacio_Vera.
Can you provide a tentative timeline as to when the fixes will be integrated to Elasticsearch?

[Ignacio_Vera]

Hello @amal_antony ,

The fix will be released in the 8.x line whenever a new version of Lucene is released and the release is adopted by Elasticsearch.

There was actually a branch cut on Lucene last week and a new release is expected soon so I hope the fix will be in the next Elasticsearch minor version (8.9.0).

The fix was backported to the 7.x line too so I hope it will be included in the next bug fix release too.

[amal_antony]

Thanks for the update, @Ignacio_Vera !