I was assuming the problem is with indexes which are still managed, but contain stale entities - on top of valid indexed entities. So in this case the goal would not be to delete the whole index but to purge the individual entries which don't match anymore.

LOL, I guess we need a clarification here what the actual problem is Marc Schipperheyn!?

If I misunderstood and the goal is to completely delete an index which is no longer used, then we wouldn't even have an IndexManager for it so adding such a purge(indexName) method wouldn't be of use.

True, true - we would not have one created by the configuration. However, we could create one on the fly (at least that was my thought here).

In this case I don't see why the user can't just `rm -R` the index from filesystem.

Except, it might not be on a local file system. But true, maybe this is just something which there is no point providing anything and you either have to try to delete the state index manually or you blow everything away and re-index from scratch.