在Elasticsearch中对大量文档进行排序

Question

当我想从elasticsearch索引中检索大量文档时，我总是使用elasticsearch的扫描和滚动技术（ http://www.elasticsearch.org/guide/en/elasticsearch/guide/current/scan-scroll.html ） 如下：

conn = Elasticsearch( hosts = HOSTS )

the_query = { 'query': { 'match_all': {  } }, 'sort': { 'created_at': { 'order': 'asc' } } } # would like sort the documents according to the 'created_at' date

scanResp = conn.search( index=TARGET_INDEX, doc_type=TARGET_DOC_TYPE, body=the_query, search_type='scan', scroll='10m' )
scrollId = scanResp['_scroll_id']
doc_num = 1

response = conn.scroll( scroll_id = scrollId, scroll='10m' )

while ( len( response['hits']['hits'] ) > 0 ):
    for item in response['hits']['hits']:
        print '\tDocument ' + str(doc_num) + ' of ' + str( response['hits']['total'] )
        doc_num += 1

        # ====================
        #   Process the item
        # ====================
        the_doc = item['_source']


    # end for item
    scrollId = response['_scroll_id']
    if doc_num >= response['hits']['total']:
        break
    response = conn.scroll( scroll_id = scrollId, scroll='10m' )
# end of while

但是，正如elasticsearch文档中提到的那样，检索到的文档将不会排序，因此结果不是我想要的。

我的问题：如何在Elasticsearch中排序大量文档？

谢谢 ：）

Answer 1

遍历排序列表时滚动非常昂贵，但是如果您坚持要从查询中删除“ scan” search_type。 滚动时，scan禁用排序。