Ceph » rgw

I am using ceph 16.2.7 on Rocky Linux 8.5

Bucket index objects, such as .dir.<zoneid>.<bucketid>.N, where there are N objects based on the number of index shards the bucket had, still exist after an s3 bucket deletion. These are in the the index pool, of course. They still exist after a radosgw-admin gc process --include-all,
but I am not sure if that matters.

This is a multisite with two zones (1 at each site) that replicate in both directions. The index objects persist on both sides' index pool.

These clusters originally started at 15.2.x and have been upgraded. The bucket used for this test was created after the upgrade. But a similar bucket that was
created before the upgrade and was deleted before the upgrade still has its index objects in the index pool.

I see a really old ticket, https://tracker.ceph.com/issues/16412 that at least suggests it was 'fixed' in the past.

Please advise