Today the problem occurred that the FileServer replication via DFS over a site-to-site VPN was extremely slow. Normally, the WAN connection has a bandwidth of 6 Mbit/s, which was also achieved when manually copying data. However, the DFS replication was bobbing along at a pitiful 20 kbit/s.
Since this was an initial synchronization of the file servers, it would have taken forever for the file servers to even begin to synchronize with this bandwidth.
This message appeared in the event log on both servers every few minutes:
The DFS replication service terminates communication with partner SERVER for replication group domain.local\dfs\freigabe due to an error. The service will regularly try to re-establish the connection.
Further information:
Error: 1726 (The remote procedure call failed.)
Verbindungs-ID: 68DFFA96-99DA-4472-97B7-XXXXXXXXX
Replication group ID: CD8FB635-BB80-42D5-8DC9-XXXXXXXXX
The first approach was to deactivate TCP offloading (on both file servers):
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters
Value =DisableTaskOffload
Type = DWORD
Data = 1
Value =EnableTCPChimney
Type = DWORD
Data = 0
Value =EnableTCPA
Type = DWORD
Data = 0
Value =EnableRSS
Type = DWORD
Data = 0
This has already increased the speed by a factor of 10, but 200 kbit/s on a 6 Mbit/s line is still not satisfactory.
The solution was to change the MTU size, in this case to 1300
netsh interface ipv4 set subinterface "LAN connection" mtu=1300 store=persistent
After this change was made on both file servers and the servers were restarted, the replication speed increased to the maximum available bandwidth.
