We'd like to replicate a mail drop folder between two servers; however, as files are written to the source folder, a timer job on that server is also regularly processing and deleting them, meaning they may not last very long.
My question is: does DFS guarantee that everything written to the source folder is, at some point, also written to the target folder, independent of how quickly it is deleted in the source folder after the initial write?
I should add that we're not replicating deletes, just writes (and there are no update operations).
References:
Best Answer
How are you planning on "not replicating deletes"? If your Member2 will be the Read Only member, will you be deleting the files outside of the DFS namespace on that server? Otherwise the Read Only function of the namespace will prevent deletions from Member2.
No, DFSR cannot guarantee that every write will be replicated to a member of the replication group if the writes are followed by delete actions. An example:
You add 1000 files totaling 1 GB to Member1. Member1 will begin replicating these files immediately, with a queue of 16 files at a time (if I remember correctly for Server 2008). These files will still take some time to be staged and transferred across the network, especially if this is over a WAN.
If during that staging operation, some of the files are deleted from Member1, they will stop staging and never make it to Member2.