Guest OS file restore and DFS Replication Issue

[c0f33cup]

Case ID: 04633273

I manage about 20 sites using DFSR. On the manufacturing side I collect remote server data to the corporate office for backup via Veeam B&R. On the distribution side I replicate a folder structure to remote sites so all users have the same data set. For corporate I have a replication group as well. DFSR is configured such that a central collection server (same for all three replication groups) is the hub and remote site servers are the spokes.

I ran a restore for the distribution side this past Friday, then again on Saturday due to complications. During the restore I was unable to query the server for it's backlog. When the restore completed and the backup windows were dismissed, the server started responding as usual to my backlog query commands.

Code: Select all

C:\WINDOWS\system32>dfsrdiag backlog /rmem:cle-archive02 /smem:cle-fileprint01 /rgname:"remotebackup-corporate" /rfname:"corporate"
[ERROR] Failed to execute GetOutboundBacklogFileCount method. Err: -2147217406 (0x80041002)
Operation Failed

This morning I received many reports of missing data for the corporate replication group / folder. The missing data was from Friday only; the data is in Friday's backup but missing from the server. It seems as though the restore on Friday prohibited the synchronization of data from the corporate replication group to it's replication partners and somehow that data was removed when replication kicked back off after the restore.

Nervous that another restore (corporate files) could cause issues with the distribution replication that isn't quite finished, I've disabled all other member servers for the distribution and corporate replication groups. The plan is to re-enable the members one at a time and allow them to replicate (MS indicates a disabled member, when re-enabled, will perform a non-authoritative sync).

As noted above I have engaged support, and the gentlemen I spoke with has informed me that Veeam restore does not interact with DFSR (or really anything else on the server) as we're doing a 'dumb' guest OS file restore. What's strange is that during the restore DFSR is unresponsive via the command line and responsive after the restore finishes. In the DFS Management MMC none of the server members or connections showed as disabled or unavailable.

Have any of you experienced this or something similar?
Are you aware of Veeam interacting with the OS or DFSR service in such a way to cause a replication database reset or other replication problem?
Are there prescribed steps for restoring files in a situation like this that is heavily dependent on DFS replication?

Thanks.

[PetrM]

Hello,

I'd say that the statement provided by our support engineer is correct, DFSR activity should be transparent for file-level restore. I'd look for warning or error messages coming from DFSR in Windows event logs, perhaps it might give a better idea what's going on under the "hood" at the level of OS. Nevertheless, the described behavior must be investigated by our support team, it's hard to believe that we're talking about simple coincidence.

As far as I see, you opened the ticket today so I would let our engineers to get more time in order to properly analyze the debug logs and come up with a reply. Anyway, you always have an option to escalate the support case if you feel that it's not on the right track.

Thanks!