Per-field metadata vs index-wide metadata rationale


In 7.6 the following PR was merged:

The PR added the possibility to have per-field metadata limited to five string fields with 20 character keys and 50 character values. The rationale was that this was enough for any use case and that it would not "bloat the cluster state".

There have for some time been a per-index field, "_meta" which Elasticsearch does not use - it's only for application data.


How is the data in the "_meta" field stored which allows it to serve virtually any data? Why does it not have the same restrictions as the per-field "meta" field?

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.