The documented procedure to "Compact oplog.rs to reclaim disk space" might be improved on a couple points.
1. It implies, but does not make crystal-clear, that the compact operation must be performed only on SECONDARY members. I am working now with a customer who has been confused by this. (00551858)
2. The recommended step rs.stepDown() may not always work as expected in replica sets with asymmetric priorities. In a P-S-S with priorities 5, 2, and 1, when you `stepDown()` the primary, it will quickly initiate a priority-takeover, and "step back up".
- Specify in first paragraph that you must run compact against secondaries only
- Revise rs.stepdown to specify a stepdown timeout of 120 seconds (to ensure a high-priority primary does not immediately re-elect itself)