Merging the index with lower max_number_segement causing increasing index size

Hello Team,

We have performed force merging on one index size of 74GB with 2 shards (primary + replica) and 36 segments (primary + replica) to remove deleted docs.

curl -s -X POST ""

The index size kept on increasing - in 20 mins 74 to 88GB. Is it normal?

Why are you using so many segments?

Already there are 36 segments. I wish to merge it to 18 so to remove the deleted docs.
/_forcemerge?only_expunge_deletes=true alone is not working

You typically merge to a single segment.

Is it fine to rerun the same command as existing task going on?

No, you need to wait unfortunately. You cannot cancel it either.

Thanks @warkolm
Will wait. question out of curiosity. How much space it would take overall given above numbers? any idea?

What sort of data is it? What are the mapping and index settings?
You'd normally expect it to drop, but all of those things above impact what happens.

alright. will wait till this settles then execute it with single segment

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.