如何找出索引中被删除（docs.deleted）的文档的原因？

Question

我正在尝试使用_bulk API并将文档插入到我的弹性搜索集群中。插入操作完成后，我看到一些文档最终保存在docs.deleted存储桶中。

_cat / indices返回以下内容

green  open   18f1d25144e2979dc93de6bcdbbfb151 BWXgCY3NRBCdK2rR-r7xeA   3   1          360475            27       386.6mb         188mb

我想找出为什么删除这27个文档。

以下是我索引上的_settings

{
    "18f1d25144e2979dc93de6bcdbbfb151": {
        "settings": {
            "index": {
                "refresh_interval": "1s",
                "number_of_shards": "3",
                "provided_name": "18f1d25144e2979dc93de6bcdbbfb151",
                "creation_date": "1558551213029",
                "analysis": {
                    "analyzer": {
                        "lowercase_analyzer": {
                            "filter": [
                                "lowercase"
                            ],
                            "type": "custom",
                            "tokenizer": "whole_text"
                        }
                    },
                    "tokenizer": {
                        "whole_text": {
                            "pattern": "  abcd  ",
                            "type": "pattern"
                        }
                    }
                },
                "number_of_replicas": "1",
                "uuid": "BWXgCY3NRBCdK2rR-r7xeA",
                "version": {
                    "created": "5010299"
                }
            }
        }
    }
}

以下是该索引上的映射

{
    "18f1d25144e2979dc93de6bcdbbfb151": {
        "mappings": {
            "result": {
                "properties": {
                    "begin_time_0": {
                        "type": "long"
                    },
                    "begin_time_dow": {
                        "type": "text",
                        "fields": {
                            "keyword": {
                                "type": "keyword"
                            },
                            "lowercase": {
                                "type": "text",
                                "analyzer": "lowercase_analyzer"
                            }
                        }
                    },
                    "concurrent_commit_total_time_percentile_95": {
                        "type": "double"
                    },
                    "es_id": {
                        "type": "keyword"
                    },
                    "instances_accessed_percentile_95": {
                        "type": "double"
                    },
                    "row_count_simple_sum": {
                        "type": "long"
                    },
                    "task_oms_time_percentile_95": {
                        "type": "double"
                    },
                    "total_time_percentile_95": {
                        "type": "double"
                    },
                    "u_task_running_time_percentile_95": {
                        "type": "double"
                    }
                }
            }
        }
    }
}

弹性搜索版本-5.1.2