A python script to detect duplicate documents in Elasticsearch.
The script creates a file containing a delete action for each duplicate document that is detected, which looks like this:
- Single index is provided:
{"delete": {"_type": "doc", "_id": "1", "_index": "test_index"}}
{"delete": {"_type": "doc", "_id": "2", "_index": "test_index"}}
{"delete": {"_type": "doc", "_id": "3", "_index": "test_index"}}
- Index pattern or alias is provided (test_index_*):
{"delete": {"_type": "doc", "_id": "1", "_index": "test_index_a"}}
{"delete": {"_type": "doc", "_id": "2", "_index": "test_index_b"}}
{"delete": {"_type": "doc", "_id": "3", "_index": "test_index_c"}}
Once all duplicates are found, you can use the bulk API to perform all of the delete operations in a single API call by either providing text file input to curl or attaching the file contents in the request body.
The curl command for the first option would be the following:
curl -s -H 'Content-Type: application/x-ndjson' -XPOST 'localhost:9200/_bulk' --data-binary "@bulk_deletions_file.txt"
$ python deduplicate-elasticsearch/deduplicate-elaticsearch.py -h
usage: deduplicate-elaticsearch.py [-h] [-es ES_HOST] -i INDEX -k KEYS
[KEYS ...]
optional arguments:
-h, --help show this help message and exit
-es ES_HOST, --es_host ES_HOST
Elasticsearch host
-i INDEX, --index INDEX
<Required> Index name or alias to search on
-k KEYS [KEYS ...], --keys KEYS [KEYS ...]
<Required> List of fields that will determine
duplicate docs
For a full description on how this script works including an analysis of the memory requirements, see: https://alexmarquardt.com/2018/07/23/deduplicating-documents-in-elasticsearch/