MongoDB WiredTiger Cache issues

Question

We are using MongoDB ReplicaSet 4.2 in Kubernetes. There is one Master and two secondaries. Each of them deployed on a separate node.

MongoDB documentation says it used both WiredTiger cache and file system cache. But in what sequence does it use both?

Let's say we need to perform some query for which a collection needs to be brought into memory. Does it use wiredTiger cache or file system cache?

And I also I see WiredTigerLAS.wt large in size (1.6GB) while my wiredTiger cache is just 0.25GB.

Is WiredTiger cache persisted on disk so that between restarts, it doesn't lose the context? I temporarily let MongoDB decide size of WiredTigerCache by removing the config that I am providing. It became around 1.5GB and I am able to see improved performance for my queries. I reduced WiredTigerCache size and restarted MongoDB and confirmed reduced WiredTigerCache size via commands. Even though cache size is reduced, I still see improved performance. Is it because earlier cache is persisted on the disk and up on restart it is used despite me providing lower size.