Creating Elasticsearch rollover indexes

If you need to create a new Elasticsearch index, you can use the _rollover API. For example, your index might be too large or too old, or you might want to use a different number of shards.

Before you begin

 For 18.0.0  and  For 18.0.1 : First verify that the index template for the index that you want to rollover exists.

About this task

One example of using the _rollover API is deciding the number of shards of your index. Upon startup, IBM® Business Automation Insights creates indexes made of 5 shards and one replica. You can update the number of replicas of an existing index, but not the number of shards. By using the _rollover API, you can pass the number of shards you want as a payload to the POST request that regenerates the index.

  • Checking the index template
     For 18.0.0  and  For 18.0.1 : First make sure that the index template exists.
  • Creating indexes: General procedure
    Provided that the appropriate index template exists, you can create a new Elasticsearch index by running _rollover POST requests.
  • Rolling over indexes that store active summaries
    For BPMN and Case events, if you need to roll over an active index that store active summaries, before running _rollover POST requests, you must stop the Flink job, and after the new index is created, you must re-index data.
Parent topic: Operating Elasticsearch