A repetitive 5 seconds slow queries - and I'm out of ideas

[scaraliu (Sam)]

On search queries sent via POST method I'm getting from time to time a no-less than 5 seconds, repetitive, delay until the response is received from Elasticsearch.

Had this issue in the past, solved but now on 8.1.0 it seems impossible to resolve.

Something is missing - but I don't know what.

On get requests It seems that this behavior is not present.
Bellow are settings that I tried.

Does anyone have a clue...

{
      "index.mapping.total_fields.limit": "10000",
      "index.refresh_interval": "-1",
      "index.translog.sync_interval": "180s",
      "index.translog.flush_threshold_size": "2024mb",
      "index.blocks.read_only_allow_delete": "true",
      "index.query.default_field": "search",
      "index.priority": "1",
      "index.number_of_replicas": "3",
      "index.routing.allocation.include._tier_preference": "data_hot"
}

[warkolm (Mark Walkom)]

Welcome to our community!

It'd be useful if you could show the query.

[scaraliu (Sam)]

All queries. Either basic ones, either complex. No difference, all get a repetitive, from time to time, 5 seconds delay / stand-by. I tried like 5-10 different queries, same 5 seconds of inexplicable from time to time delay / stand-by.