A: Hyper-V Replica replicates changes to the storage of the replicated virtual machine (VM) every five minutes. If the network connection between the Hyper-V servers is unavailable, then initially the connection attempt will be retried after one minute, then at continually increasing intervals until a retry interval of 30 minutes is reached.

At this point, the connection attempt will retry indefinitely every 30 minutes. When the connection is reestablished, all log information is sent over automatically, no manual action is required.

If during the disconnected time the log file size is more than 50 percent of the base virtual hard disk (VHD) size, then the replication changes to a "Resynchronization required" state and future changes are no longer tracked. When connection is reestablished, a block-by-block comparison between the primary and replica VM is performed to ascertain what blocks need to be copied.

This block-by-block check can be fairly resource intensive, so one option if you experience this state is to remove the replica and just restart replication.

There is a nice Microsoft document "Understand and Troubleshoot Hyper-V Replica in Windows Server "8" Beta,"  which helps troubleshoot Hyper-V Replica.