java.lang.OutOfMemoryError: Java heap space

[koala775]

Describe the feature:
When I use nested aggregation of dsl return large amount of data,then elasticsearch error: java.lang.OutOfMemoryError: Java heap space.

Elasticsearch version (bin/elasticsearch --version):
6.3.0

Plugins installed: []

JVM version (java -version):
jdk_1.8.152

OS version (uname -a if on a Unix-like system):
linux
Description of the problem including expected versus actual behavior:
1.Without any operating,why ramCurrent is full.

2.Then query dsl frequently,elasticsearch happened error :java.lang.OutOfMemoryError: Java heap space
3.ramCurrent include what?
4.aggregation used which ram，filedata cache or query cache or request cache?

Steps to reproduce:

Please include a minimal but complete recreation of the problem, including
(e.g.) index creation, mappings, settings, query etc. The easier you make for
us to reproduce it, the more likely that somebody will take the time to look at it.

1.query node state
2.query dsl frequently
3.elasticsearch error

Provide logs (if relevant):

[elasticmachine]

Pinging @elastic/es-search-aggs

[jpountz]

When I use nested aggregation of dsl return large amount of data

Can you please share what the request looks like?

[koala775]

When I use nested aggregation of dsl return large amount of data

Can you please share what the request looks like?

GET index/_search
{
  "query": {
                "bool": {
                    "must_not": {
                        "term": {"xxx.keyword": "xxx"}
                    },
                    "filter": [
                        {
                            "range": {
                                "crt_dt": {"gt": "2018-01-14T09:00:00.000Z", "lte": "2018-01-14T10:00:00.000Z"}
                            }
                        }
                    ]
                }
            },
            "size" :0,
            "aggregations": {
                "time": {
                    "date_histogram" : {
                        "field" : "date" ,
                        "format": "yyyy-MM-dd kk:mm:ss",
                        "interval": "1m",
                        "time_zone": "Asia/Shanghai",
                        "order" : { "_key" : "desc" },
                        "offset":    "+8h",
                        "min_doc_count": 1

                    },
                    "aggregations": {
                        "name": {
                            "terms": {
                                "script" : {
                                    "source": "doc['xxx.keyword'].value + ';'+doc['xxx.keyword'].value + ';'+doc['xxx.keyword'].value+';'+doc['xxx.keyword'].value+';'+doc['xxx'].value",

                                    "lang": "painless"
                                } ,"size":1000000
                            },

                            "aggregations": {
                                "ava": {
                                    "avg": {
                                        "field": "ava"
                                    }
                                },
                                "success": {
                                    "sum": {
                                        "field": "success"
                                    }
                                },
                                "fail": {
                                    "sum": {
                                        "field": "fail"
                                    }
                                },
                                "max": {
                                    "max": {
                                        "field": "max"
                                    }
                                },
                                "min": {
                                    "min": {
                                        "field": "min"
                                    }
                                }
                            }
                        }
                    }
                }
            }

}

[elasticmachine]

Pinging @elastic/es-analytics-geo

[imotov]

That's a term aggregation that is trying to get back 1 million buckets within a date histogram aggregation. So, it's seems like an aggs-related circuit-breaker issue to me. @dliappis was it relabeled as :Analytics/Geo by mistake? I cannot really find anything geo-related here. Could you clarify?

[dliappis]

@imotov thanks for the clarification, indeed had a chat with @mayya-sharipova about the right team, but I think geo came by mistake. I'll relabel it as :Analytics/Aggregations.

[polyfractal]

I suspect this is just a case of #28220, where our agg framework does a poor job estimating the memory cost of each agg type. It underestimates some aggs, and over-estimates others.

The newish search.max_buckets setting will help, but I agree we should spend some time making the agg CB estimates better.