Summary
Currently, the checkpoint cleanup stops processing whenever it detects that eviction is needed as per the below code.

if (__wt_cache_aggressive(session) || __wt_cache_full(session) || __wt_cache_stuck(session) ||      __wt_eviction_needed(session, false, false, NULL)) {
      *skipp = true;
        return (0);    
} 

As observed in an internal cluster, checkpoint-cleanup gave up whenever the eviction is required leading to an increase in disk size of the history store.

Motivation

• Does this affect any team outside of WT?
  No

• How likely is it that this use case or problem will occur?
  Very high operations scenario.

• If the problem does occur, what are the consequences and how severe are they?
  The disk space increases.

• Is this issue urgent?
  No

• Suggested Solution*

Instead of checking the eviction needed approach that checks for clean/update/dirty, checking only the clean page eviction could be beneficial as checkpoint-cleanup lead to more clean pages in the cache except when the page is marked to delete.