Our project team found that even though we were discarding builds that the number of builds we were keeping times the frequency that our jobs were building was creating a lot of build folders.

After a while when the job would complete the discard builds clean up process had a pretty dramatic effect on the IO performance on the server which slowed down other applications running on the same server.

On top of it people were confused about why the server was running slow and why the spinner in the console logs was spinning and blocking the job from running again.

It would be nice to:
1) Not mark the job as a success until clean up of old builds had completed
2) Log to the console that cleanup of old builds is occurring
3) Add a helpful hint to the discard old builds section that mentions performance impacts from using it, much like there is for the Building and Polling SCM sections.