A map/reduce job running on the Primary is causing a mongodump operation, running on the Secondary, to fail.
mongodump reports that the a temporary collection does not exist anymore by the time it tries to list its indexes.

1) should mongodump consider the temporary map/reduce collections when performing a dump of the databases?
2) the command to drop the collection was issued before the mongodump operation started (but may not have been completed by the time mongodump started), however the error in looking up the collection appears way later (30 minutes) than the drop database command.