unusual negative values and high spikes (22 years) communication lag observed in repmgr replication_status

[Pdevanath]

Hello,

I am seeing random occurrences of negative values for comm_lag in repmgr replication_status table. and very high communication lag values (22 years) during add node to a cluster. Any idea why this happens? 

Thanks