There are few issues with vacuuming in the current code:

1. Multiple vacuuming child processes can trigger at the same time which can be quite expensive on a loaded system.

2. Vacuuming process uses blocking locks while checking database records, which can get in the way of regular database processing.

3. Recovery daemon processes vacuum fetch records one by one which can take longer than the actual vacuuming interval and cause subsequent vacuum fetch requests to be dropped.