There is a project, PM-3561 for incremental backup improvements. During a meeting it occurred to me that we don't understand the impact of continuous background compaction on incremental backup. This ticket should investigate that.

etienne.petrel@mongodb.com I will let you decide whether or not this ticket gets added to PM-3091.

Questions to be investigated are:

• Does background compaction result in more, less, about the same amount of data needing to be copied during incremental backup? I.e. does frequent compaction result in many more modified bits getting set than it otherwise would?
• Is there any impact of shrinking the file size and the information returned by incremental backup? (i.e. does it work correctly - do we not return any modified blocks beyond the file size?)

An initial analysis and a test (possibly format, the backup in timestamp_abort or a new test) and some way to see results varying workload results should be the result. We will need to use this test in the future for PM-3561.