Indexes are not replicated between the replicas of a Redis Enterprise Active-Active database. Indexes must be managed independently in the different database replicas. One reason is that some regions could have different query needs, requiring different index setups. In the same way as for index creation, FT.DROPINDEX isn’t supported in the CRDT protocol.

So, following the index creation (FT.CREATE) on each of the participants in each cluster, deleting the index follows the same logic, meaning that the FT.DROPINDEX command must be executed on all the participant clusters individually.

References

Learn about the Active-Active geo-distributed Redis Enterprise feature.