3.1.6 Other Local Events

Volume dismounts and errors. The per-replicated folder state gets reset when encountering errors that prevent processing files on the volume on which they reside. At the protocol level, such events result in discontinuing per-replicated folder replication activity. On the server side, this amounts to failing RPC calls that are specific to the particular replicated folders. Locally, volume state is rechecked based on a periodic timer. Remotely, a client attempts to reestablish per-replicated folder sessions by using the EstablishSession RPC call. The call is retried using an exponential backoff scheme with a maximal time-out of 5 minutes. That is, retries are first attempted with a delay of 1 second, then a delay of 2 seconds, then delays of 4, 8, 16, 32, 64, 128, and 256 seconds, and are then followed by delays of 300 seconds.<18>