This is expected, since 7.x indexes are created by index lifecycle management (ILM). ILM creates a single alias (metricbeat-<version>) to where metricbeat sends the data. This alias is configured to point to an index where all the data is written. ILM starts a new index when the current index is older than a month or bigger than 50GB. This is intended, to avoid having too many small indexes, but still have some .
This behaviour is controlled by an index lifecycle policy maintained by Metricbeat, and can be configured.
This wouldn't update current index, but you can manually rollover it, and the next one will have the new policy. To force a rollover:
POST metricbeat-7.4.0/_rollover
You can find what policies are being used by the indexes managed by metricbeat-7.4.0 rollover alias with:
GET metricbeat-7.4.0/_ilm/explain
Take into account that you will need to repeat the process when you upgrade Metricbeat, because the version is included in the policy name. Since 7.6.0 (elastic/beats#14745) the version is not included, so Metricbeat continues using custom policies if defined.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.