直接熔断报错了,请问大佬有没有什么解决方案,急用谢谢了
查询语句:
查询语句:
{
"from": 0,
"size": 0,
"query": {
"bool": {
"adjust_pure_negative": true,
"boost": 1
}
},
"version": true,
"explain": false,
"aggregations": {
"effectStatusAgg": {
"terms": {
"field": "effectStatus.code.keyword",
"size": 999999,
"min_doc_count": 1,
"shard_min_doc_count": 0,
"show_term_doc_count_error": false,
"order": [
{
"_count": "desc"
},
{
"_key": "asc"
}
]
},
"aggregations": {
"lawIdList": {
"top_hits": {
"from": 0,
"size": 999999,
"version": false,
"seq_no_primary_term": false,
"explain": false
}
}
}
}
}
}
报错: "reason": {
"type": "circuit_breaking_exception",
"reason": "[parent] Data too large, data for [indices:data/read/search[phase/query]] would be [4094396904/3.8gb], which is larger than the limit of [4080218931/3.7gb], real usage: [4094395984/3.8
猜测是由于tophits返回的集合太大了而超时(大概最多有几千个元素,去掉这一项查询就不报错了),请问各位大佬有没有设么解决办法
1 个回复
zmc - ES PAAS、JuiceFS
赞同来自:
解决的话,有资源直接扩容,没资源的话减小返回的数量,或者牺牲一点准确性,调整 shard_size