DFS Replication questions

[maybeimaleo]

Can anyone help me find some answers to the below?

1) If there is a failure when using DFS Replication, the length of time before it discovers a failure is 30 seconds. Is that for the server as a whole or for each file?

2) If a failure occurs and users are redirected to the backup server how does DFS know to failback when the primary server is back online?

3) How does it replicate back to the server if changes are made on the backup server while the primary is out of commission?

 

[ITPangaeaArchitect]

1) If there is a failure when using DFS Replication, the length of time before it discovers a failure is 30 seconds. Is that for the server as a whole or for each file?
A: That is for the server as a whole. I believe what you are referring to here is the interval it takes to get a new dfs referral on the client.

2) If a failure occurs and users are redirected to the backup server how does DFS know to failback when the primary server is back online?
A: I believe this will only occur at the next connection attempt to the DFS target, or after logoff/logon..cant remember off top of head, too early . Ots whenever the next DFS referral attempt occurs and the recovered server is considered the least cost. Remember, DFS is a multimaster topology, so when multiple targets are enabled, they are all considered to be "primary" (yes even if you set costs...that just changes the order theyll be contacted in if contact is possible basically).

3) How does it replicate back to the server if changes are made on the backup server while the primary is out of commission?
A: The USN table is updated for a file with each change made. A failed, then recovered, DFS server would have a lower USN than the "backup" DFS server, which tells it that it needs to pull the changes in those files.

-Brandon Wilson
MCSE00/03, MCSA:Messaging00, MCSA03, A+
Manager - Global AD Operations
ACS, Inc.