When dropping an index, it's possible for the operation to fail due to a "disk space quota exceeded" error. Seems like removal operations that reduce the size of the database shouldn't ever cause this kind of error.

command failed [command
failed [drop]

{ "nIndexesWas" : 1.0 , "msg" : "indexes dropped for collection" , "ns" : "xxx.yyy" , "errmsg" : "exception: db disk space quota exceeded xxx" , "code" : 12501 , "ok" : 0.0}