The elasticsearch create index name rule problem

[nilnon]

Now the elasticsearch index name suffix is based on execution_date. In the scenario of task rerun, many historical indexes will be generated. Can this rule be optimized?