elasticsearch runtime_error ai_generated true

QueryPhaseExecutionException: Result window is too large, from + size must be less than or equal to: [10000] but was [50000]

ID: elasticsearch/max-clause-count-exceeded

Also available as: JSON · Markdown · 中文

85%Fix Rate

88%Confidence

1Evidence

2024-03-15First Seen

Version Compatibility

Version	Status	Introduced	Deprecated	Notes
elasticsearch 7.17	active	—	—	—
elasticsearch 8.11	active	—	—	—
elasticsearch 8.12	active	—	—	—

Root Cause

The combination of 'from' and 'size' parameters in a search request exceeds the 'index.max_result_window' setting, causing query execution to fail.

generic

中文

搜索请求中 'from' 和 'size' 参数的组合超过了 'index.max_result_window' 设置，导致查询执行失败。

Official Documentation

https://www.elastic.co/guide/en/elasticsearch/reference/current/search-request-from-size.html

Workarounds

90% success Use search_after with sort to implement cursor-based pagination, avoiding from+size limits. Example: POST /my_index/_search { "size": 10, "sort": [ { "date": "asc" } ], "search_after": [ "2024-01-01T00:00:00" ] }
```
Use search_after with sort to implement cursor-based pagination, avoiding from+size limits. Example: POST /my_index/_search { "size": 10, "sort": [ { "date": "asc" } ], "search_after": [ "2024-01-01T00:00:00" ] }
```
80% success Temporarily increase index.max_result_window on the specific index: PUT /my_index/_settings { "index.max_result_window": 50000 }
```
Temporarily increase index.max_result_window on the specific index: PUT /my_index/_settings { "index.max_result_window": 50000 }
```
95% success Use point-in-time (PIT) with search_after for stable deep pagination in ES 7.10+: POST /my_index/_pit?keep_alive=1m then POST /_search { "pit": { "id": "...", "keep_alive": "1m" }, "size": 100, "sort": [ { "_shard_doc": "asc" } ], "search_after": [ ... ] }
```
Use point-in-time (PIT) with search_after for stable deep pagination in ES 7.10+: POST /my_index/_pit?keep_alive=1m then POST /_search { "pit": { "id": "...", "keep_alive": "1m" }, "size": 100, "sort": [ { "_shard_doc": "asc" } ], "search_after": [ ... ] }
```

中文步骤

使用 search_after 配合排序实现基于游标的分页，避免 from+size 限制。示例：POST /my_index/_search { "size": 10, "sort": [ { "date": "asc" } ], "search_after": [ "2024-01-01T00:00:00" ] }

临时提高特定索引的 index.max_result_window：PUT /my_index/_settings { "index.max_result_window": 50000 }

在 ES 7.10+ 中使用 point-in-time (PIT) 和 search_after 实现稳定的深度分页：POST /my_index/_pit?keep_alive=1m 然后 POST /_search { "pit": { "id": "...", "keep_alive": "1m" }, "size": 100, "sort": [ { "_shard_doc": "asc" } ], "search_after": [ ... ] }

Dead Ends

Common approaches that don't work:

60% fail
Excessive increase can cause severe memory pressure and OOM on coordinating nodes, especially with deep pagination.
70% fail
Scroll is designed for deep pagination but not for random access; it requires maintaining context and doesn't support from+size directly.
80% fail
This suppresses total hit count but doesn't resolve the from+size limit; the error persists if from+size still exceeds max_result_window.