This request is to provide more detailed memory usage statistics for forestdb. Right now, buffercache memory usage is the only memory stat provided. GSI opens up in-memory snapshots(every 200ms) and with long running scans there can be lot of those open at any given point of time. Also I think iterators cause some temporary allocations inside forestdb. It seems there can be sizable amount of memory usage outside of buffercache.
The goal is for Indexer to know the memory usage better so it can throttle appropriately to avoid running into OOM scenarios(e.g. MB-19883). Also it will help us provide better sizing estimate to customers.