(filing this on stennie 's recommendation)

Something that's come up a few times has been that we don't have a tutorial explaining the different ways to recover disk space when running a replica set. These include:

• bringing down each node in rolling fashion and running repairDatabase
• nuking secondaries and reconstructing clean ones with an initialsync

the issues to be touched on here would be:

• how to do it
• differing requirements (network bandwidth, free disk space, time)