For ElasticSearch (provisioned through AppSearch), is there a way to get a detailed breakdown of consumed memory within the jvm or elastic process?
We are currently suffering from a lot of circuit breaks but there is no obvious cache or index that seems to be consuming a large amount of memory.
The request_cache
, query_cache
, and fielddata
index are all relatively small.
When we use the _cache/clear
endpoint, we get a brief respite, but its not permanent
Edit: Example of one of our circuit breaker errors
{
"error": {
"root_cause": [
{
"type": "circuit_breaking_exception",
"reason": "[parent] Data too large, data for [<http_request>] would be [15357980864/14.3gb], which is larger than the limit of [15125499084/14gb], real usage: [15357970248/14.3gb], new bytes reserved: [10616/10.3kb], usages [model_inference=0/0b, inflight_requests=10616/10.3kb, request=0/0b, fielddata=77884/76kb, eql_sequence=0/0b]",
"bytes_wanted": 15357980864,
"bytes_limit": 15125499084,
"durability": "PERMANENT"
}
]
}
}