Mechanisms behind async_search

[alexpanas (Alexey Panasenko)]

Hello,

Could you please elaborate more about the mechanisms behind the current implementation of async search in Elasticsearch 7.8?
As far as I can see - in our environment the .async_search index has grew up to 40 Gb of total storage:

curl -s -XGET localhost:9200/_cat/indices/.async-search?v
health status index         uuid                   pri rep docs.count docs.deleted store.size pri.store.size
green  open   .async-search XI7AfidQSKuyzQrKKW7FtQ   1   1      40436         8495       46gb         21.3gb

And contains a number of deleted documents.

        "docs" : {
          "count" : 80679,
          "deleted" : 25540
        },

Is there a way (without resorting to searches against mentioned index) to get a list of currently stored results for async searches?

How long those results are supposed to be stored in mentioned index? Is there a way for us to change the storage time?

How often old results are being deleted from mentioned index? Is there a way to configure the frequency of this cleanup?

How often mentioned index is supposed to be force merged (as I am a bit concerned in regards to the current size of the index - and also - about the number of deleted documents listed in the stats)? Is this frequency configurable? Is it safe to force merge this index manually to free up the space?

Best regards,
Alex