When a config server process starts up it starts the distlock pinger. The distlock pinger, however, doesn't check if the node is primary before trying to ping the distributed lock. This is fine from a correctness perspective, but it fills the logs on config server secondaries with unnecessary messages like

2016-08-22T18:36:17.025-0400 W SHARDING [replSetDistLockPinger] pinging failed for distributed lock pinger :: caused by :: NotMaster: not master