Manage data from the command lineedit

Learn how to index, update, retrieve, search, and delete documents in an Elasticsearch cluster from the command line.

If you are looking for a user interface for Elasticsearch and your data, head on over to Kibana! Not only are there amazing visualization and index management tools, Kibana includes realistic sample data sets to play with so that you can get to know what you could do with your data.

Before you beginedit

To find out what the ELASTICSEARCH_URL is for your Elasticsearch cluster, grep on the output of the heroku config command for your app:

heroku config --app MY_APP | grep ELASTICSEARCH_URL

These examples use the elastic user. If you didn’t copy down the password for the elastic user, you can reset the password.

To use these examples, you also need to have the curl command installed.


To index a document into Elasticsearch, POST your document:

curl -u USER:PASSWORD https://ELASTICSEARCH_URL/my_index/_doc -XPOST -H 'Content-Type: application/json' -d '{
    "title": "One", "tags": ["ruby"]

To show that the operation worked, Elasticsearch returns a JSON response that looks like {"_index":"my_index","_type":"_doc","_id":"0KNPhW4BnhCSymaq_3SI","_version":1,"result":"created","_shards":{"total":2,"successful":2,"failed":0},"_seq_no":0,"_primary_term":1}.

In this example, the index my_index is created dynamically when the first document is inserted into it. All documents in Elasticsearch have a type and an id, which is echoed as "_type":"_doc" and _id":"0KNPhW4BnhCSymaq_3SI in the JSON response. If no ID is specified during indexing, a random id is generated.

Bulk indexingedit

To achieve the best possible performance, use the bulk API.

To index some additional documents with the bulk API:

curl -u USER:PASSWORD https://ELASTICSEARCH_URL/my_index/_doc/_bulk -XPOST -H 'Content-Type: application/json' -d '
{"index": {}}
{"title": "Two", "tags": ["ruby", "python"] }
{"index": {}}
{"title": "Three", "tags": ["java"] }
{"index": {}}
{"title": "Four", "tags": ["ruby", "php"] }

Elasticsearch returns a JSON response similar to this one:



To update an existing document in Elasticsearch, POST the updated document to http://ELASTICSEARCH_URL/my_index/_doc/ID, where the ID is the _id of the document.

For example, to update the last document indexed from the previous example with "_id":"06NqhW4BnhCSymaqFHQn":

curl -u USER:PASSWORD https://ELASTICSEARCH_URL/my_index/_doc/06NqhW4BnhCSymaqFHQn -XPOST -H 'Content-Type: application/json' -d '{
    "title": "Four updated", "tags": ["ruby", "php", "python"]

The JSON response shows that the version counter for the document got incremented to _version":2 to reflect the update.

Retrieving documentsedit

To take a look at a specific document you indexed, here the last document we updated with the ID 0KNPhW4BnhCSymaq_3SI:

curl -u USER:PASSWORD https://ELASTICSEARCH_URL/my_index/_doc/06NqhW4BnhCSymaqFHQn

This request didn’t include GET, as the method is implied if you don’t specify anything else. If the document you are looking for exists, Elasticsearch returns found":true along with the document as part of the JSON response. Otherwise, the JSON response contains "found":false.


You issue search requests for documents with one of these Elasticsearch endpoints:


Either a GET or a POST request with some URI search parameters works, or omit the method to default to GET request:

curl -u USER:PASSWORD https://ELASTICSEARCH_URL/my_index/_doc/_search?q=title:T*

For an explanation of the allowed parameters, check URI Search.

To make Elasticsearch return a more human readable JSON response, add ?pretty=true to the request:

curl -u USER:PASSWORD https://ELASTICSEARCH_URL/my_index/_doc/_search?pretty=true -H 'Content-Type: application/json' -d '{
    "query": {
        "query_string": {"query": "*"}

For performance reasons, ?pretty=true is not recommended in production. You can verify the performance difference yourself by checking the took field in the JSON response which tells you how long Elasticsearch took to evaluate the search in milliseconds. When we tested these examples ourselves, the difference was "took" : 4 against "took" : 18, a substantial difference.

For a full explanation of how the request body is structured, check Elasticsearch Request Body documentation. You can also execute multiple queries in one request with the Multi Search API.


You delete documents from Elasticsearch by sending DELETE requests.

To delete a single document by ID from an earlier example:

curl -u USER:PASSWORD https://ELASTICSEARCH_URL/my_index/_doc/06NqhW4BnhCSymaqFHQn -XDELETE

To delete a whole index, here my_index:


The JSON response returns {"acknowledged":true} to indicate that the index deletion was a success.