Christopher Peck

> Good catch! The root cause is that `NotDocIdIterator` breaks the contract that no more calls to `next()` or `advance()` are allowed after they return `EOF`. We should move modify...

> Is it possible to share some numbers on how this improvement has helped with freshness and at the same time the corresponding impact on CPU / mem / IO...

> I don't fully follow the change. When there is no partition assigned to the given instance, the loop should all be skipped, and it should return GOOD right? Not...

> My bad, now I see the problem. Should we consider first looping over the ideal state and create a map from serving partition to state, then only loop over...

I think this was covered in https://github.com/apache/pinot/pull/14029 (but uses latency penalty instead of query failure count)