Issue Summary

WT-16912 introduced a stat that prints the number of dirty bytes in use for each btree during checkpoint progress messages. Currently, there is no test verifying the correctness or expected behavior of this stat.

Context

• The stat should either decrement during the checkpoint (unless parallel writes occur), or the bytes written by checkpoint should accumulate up to the dirty bytes value.
• There is uncertainty about how the stat behaves in the presence of parallel writes.
• The goal is to ensure that the stat is either internally consistent or matches other checkpoint-related stats.

Proposed Solution

• Implement a test that verifies the behavior of the dirty bytes stat during checkpoint.
• The test should:

1. Check if the stat decrements as expected during checkpoint (if no parallel writes).
2. Alternatively, confirm that bytes written by checkpoint match the dirty bytes stat.

• The test should also validate the relationship between this stat and other checkpoint stats.

Original Slack thread: Slack Thread
This ticket was generated by AI from a Slack thread.