DFSR Backlog

[Mo]

What usually causes replicating servers to generate a backlog? Is it usually
network related or some other reason like staging quotas set wrong? I have a
server that is constantly generating a backlog to and from the server with
all replication partners. The overall performance on the server is fine.
The server is attached to NAS using ISCI through a couple of network switches.

Any information would be greatly appreciated.

Thanks

[Robert A Post, Jr (MSFT)]

If you generate files or make modifications to files faster than:

1) network can process
2) disk can process
3) dfsr can process

Almost always, the disk is slow, try running perfmon and monitoring
network, disk and dfsr usage.

A typical example is a hub machine with many spokes. The spokes can
generate a few modifications a piece, equating to "# of spokes" *
modifications for the Hub to process at once.

Solutions... Get another hub, get better hardware, reduce the spokes...

Thanks,

Robert A Post, Jr [MSFT]
Distributed File Service Replication Test
This posting is provided "AS IS" with no warranties, and confers no rights.

[Brent]

I have found a few servers which don't clear the back log for some reason.

If you are 100% sure it isn't a network or security issue, then I would
suggest deleting the member from DFS-R and re-enabling it.

90% of the time I find that it is a network or secuirty issue (our security
team have locked our environment down a lot).

But I have found that on very rare occasions that even though there is no
security or network issues, a server will not complete it's initial
replication.

If this occurs, I have found deleting the member and re-creating the member
will fix everything. This is probably only occured to me 1% of the time, but
it is worth a thought