If, suppose, the primary node steps down before inserting data to "afterClientKills" collection, the insert command fails. Since, currently we don't check for write command error for that insert op, assert.soonNoExcept keeps retrying the insert op for 10 mins without spitting any kind of write command error message.