Finding and deleting empty indexes

[gormanst (Hobo Joe)]

I am using the RestHighLevelClient to try and find then delete empty indexes. I can't find anything in the API that makes getting empty indexes simple. Is this something that is usually handled by Curator? Is dealing with empty indexes a common problem and if so I would be so grateful for a push in the right direction?

[Bernt_Rostad]

@gormanst

Is dealing with empty indexes a common problem

Not in my business, so I usually delete unwanted indices, whether empty or not, manually.

I did write a script a while back that lists all indices in a cluster with a doc count. To identify empty indices you can do something like this:

#!/bin/bash

srv="localhost:9200"

for idx in $(curl -XGET "http://$srv/_all/_settings?pretty" -s | jq 'keys' | sed 's/["|,]//g' | grep -E '[[:alpha:]]+'); do
    count=$(curl -X GET "http://$srv/$idx/_count" -s | jq '.count')
    if (( $count == 0 )); then
        echo "** Index $idx is empty ... delete it"
        curl -XDELETE "http://$srv/$idx"
    else·
        echo "Count $idx -> $count"
    fi
done

I've just extracted the host as a variable ($srv) in case running from a different server.