CODEWITHSUNDEEP
elasticsearch
Thomas
Thomas

Reputation: 4330

Does value of docs.count in elasticsearch also include value of docs.deleted?

I have an index with the following docs.count and docs.deleted. 

docs.count   = 2762694
docs.deleted = 2508162

Is docs.count the total number of documents in disk including the ones marked as deleted? Or is the total number of documents in disk is docs.count + docs.deleted?

UPDATE: To clarify my concern, if docs.count also include the docs.deleted in it, then in the above example about 90% of the documents are deleted. But on the other hand the docs.deleted is not counted as part of docs.count and they are additional then only 47% of the documents are deleted. Based on this I can decide how important it is to use force merge.

Upvotes: 5

Views: 3585

Answers (2)

deep
deep

Reputation: 1676

No. The docs.count does not include the docs.deleted.

docs.deleted are the documents that are marked for deletion by elasticsearch.

I don't know the exact operation you are performing but the docs.deleted can be this high even when you update a lot of your documents. As elasticsearch doesn't update the document in-place but deletes the older one and replaces it with the newer one.

Upvotes: 4

Val
Val

Reputation: 217274

The counts you see in the _cat/indices API response are counts at the Lucene level, so docs.count also includes hidden nested documents.

docs.deleted are indeed documents marked as deleted but not yet expunged.

If you want a real count of your documents you should use the _cat/count API which provides a clean count that indicates the number of live documents and does not include deleted documents which have not yet been cleaned up by the merge process.

Upvotes: 8

Related Questions