Problem: FTDC shows oplog occupying ~96GB of a 125GB cache while oplog wt cache eviction walk target pages histogram is capped at 128 pages per walk (hardcoded). With large caches and heavy oplog churn, eviction may be unable to keep up, causing frequent oplog page eviction and rereads, leading to replication lag and Flow Control.Evidence: Cache miss spikes align with bytes read into cache; oplog bytes in cache remains very high; eviction target page cap stays at 128 regardless of cache size.Hypothesis: Fixed walk target (128 pages) is too small for modern cache sizes; cache fill can outpace eviction, causing churn. A dynamic target based on cache size or oplog footprint may reduce misses.

Scope / Questions:

• Would a cache-size-proportional target reduce churn without harming other workloads?