Elasticsearch 2.3 Geo Shape 类型映射

Question

我在 Elasticsearch 索引中有一个具有以下映射的字段

   "cluster_boundary": {
        "type": "geo_shape",
        "tree": "quadtree",
        "precision": "1mm"
    }

但是当我将此映射放入弹性搜索索引时，我收到以下错误

java.lang.IllegalArgumentException: maxLevels of 36 exceeds limit of 29
     at org.apache.lucene.spatial.prefix.tree.PackedQuadPrefixTree.<init>(PackedQuadPrefixTree.java:70)
     at org.elasticsearch.index.mapper.geo.GeoShapeFieldMapper$GeoShapeFieldType.freeze(GeoShapeFieldMapper.java:276)
     at org.elasticsearch.index.mapper.FieldMapper.<init>(FieldMapper.java:289)
     at org.elasticsearch.index.mapper.geo.GeoShapeFieldMapper.<init>(GeoShapeFieldMapper.java:428)
     at org.elasticsearch.index.mapper.geo.GeoShapeFieldMapper$Builder.build(GeoShapeFieldMapper.java:160)
     at org.elasticsearch.index.mapper.geo.GeoShapeFieldMapper$Builder.build(GeoShapeFieldMapper.java:119)
     at org.elasticsearch.index.mapper.object.ObjectMapper$Builder.build(ObjectMapper.java:167)
     at org.elasticsearch.index.mapper.DocumentMapper$Builder.<init>(DocumentMapper.java:88)
     at org.elasticsearch.index.mapper.MapperBuilders.doc(MapperBuilders.java:44)
     at org.elasticsearch.index.mapper.DocumentMapperParser.parse(DocumentMapperParser.java:118)
     at org.elasticsearch.index.mapper.DocumentMapperParser.parse(DocumentMapperParser.java:99)
     at org.elasticsearch.index.mapper.MapperService.parse(MapperService.java:498)
     at org.elasticsearch.cluster.metadata.MetaDataMappingService$PutMappingExecutor.applyRequest(MetaDataMappingService.java:257)
     at org.elasticsearch.cluster.metadata.MetaDataMappingService$PutMappingExecutor.execute(MetaDataMappingService.java:230)
     at org.elasticsearch.cluster.service.InternalClusterService.runTasksForExecutor(InternalClusterService.java:468)
     at org.elasticsearch.cluster.service.InternalClusterService$UpdateTask.run(InternalClusterService.java:772)
     at org.elasticsearch.common.util.concurrent.PrioritizedEsThreadPoolExecutor$TieBreakingPrioritizedRunnable.runAndClean(PrioritizedEsThreadPoolExecutor.java:231)
     at org.elasticsearch.common.util.concurrent.PrioritizedEsThreadPoolExecutor$TieBreakingPrioritizedRunnable.run(PrioritizedEsThreadPoolExecutor.java:194)
     at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
     at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
     at java.lang.Thread.run(Thread.java:745)

如果我将quadtree 更改为geohash 它可以工作。并且，根据elasticsearch，quadtree 中允许的最大树级别是50。

Answer 1

PackedQuadPrefixTree 的最高等级为 29，而QuadPrefixTree 的最高等级为 50。您可以检查here 和here。

我查看了 ES source code。 quadtree 映射到 lucene PackedQuadPrefixTree，因此您必须将精度值降低到 1m，即转换为 26 个 tree_levels，或者您可以使用最多 50 个级别的 legacyquadtree。

我注意到的一件事是，如果您使用 legacyquadtree 指定非常多的级别而不是说最大允许级别是 50，则 ES 会抛出 out of memory error。

PUT test_index/_mapping/test_type
{
  "properties": {
    "test": {
      "type": "geo_shape",
      "tree": "legacyquadtree",
      "tree_levels": "54354552"
    }
  }
}

理想情况下，上面的代码应该说最大级别为 50，但它会耗尽堆空间并引发 OOM 错误。

我也认为 1mm 太精确了，即使使用 leagcyquadtree 也会占用大量内存。 More 对此。我还在 github 上开了一个issue，你可以关注。