Cluster link: https://cloud.mongodb.com/v2/66fa9468b7f41e3bb6affe62#/overview

Log download here : https://cloud.mongodb.com/log/collection/request/66fa9468b7f41e3bb6affe62/68b183545d3ff44353ff7711/download

Config nodes are moving from dedicated to embedded

Customer reports high cache update ratio on secondary nodes only which is increasing latency and impacting the application

Please help to identify why the cache update ratio is increasing and causing eviction to go into aggressive mode,

Currently there is a shard draining process so we expect only deletes to dirty the cache during this time, 

Question: do deletes from range deleter contribute to the update ratio?

Please also help us with what we can do to mitigate here:

• Restart nodes
• Stop the balancer and scale the cluster (currently we cant scale until the shard has drained)
• Change the snapshot history window or any other wiredTiger params