3

I have a DFS Namespace currently in auto-recovery due to an unexpected server crash. There's about 600GB of data and it's consisting of small files for the most part.

It's been in auto-recovery for over 72 hours now.

I can run Get-WmiObject -computername computername -Namespace "root\MicrosoftDFS" -Query "SELECT * FROM DfsrReplicatedFolderInfo"

To confirm that it is in State 3, which correspond to being in auto-recovery mode and also confirm that there's enough CPU, network and disk usage by the dsfrs.exe to know that it's doing "something".

Is there any way to get some sort of idea as to when it might complete and how much work there's still left to do either in time or a percentage of completion?

jfrobishow
  • 71
  • 10

2 Answers2

2

Have a look at the DFSR debug log at %windir%\debug\DFSRn.log (Where n will most likely be 01000, depending on how long DFSR has been running and what your maximum log files are configured to be. In any case it will be the highest number and the only file not GZipped.)

If recovery is still at the first stage, you will see many entries that say

2312 JRWP   763 JournalWrapTask::MarkTopDown LDB Updating ID Record:

after the timestamp.

If it's in the second stage, you will see

3936 DIRW  1986 DirWalkerTask::WalkImmediateChildren Validating:

followed by

3936 DIRW  2713 DirWalkerTask::Unmark LDB Updating ID Record:

As for how far it is through each stage, I don't think Microsoft has implemented any way of viewing this.

A J Wilson
  • 353
  • 2
  • 10
  • Thank you, I had found those log, but didn't really know what step they represented. It seems to be doing it in batches for us. For the last few days I caught mostly WalkImmediateChildren when having a look. Today I briefly seen JournalWrapTask and now it's back to WalkImmediateChildren. Good to know that there's progress being made at least. – jfrobishow May 28 '20 at 21:02
0

I believe that you are asking information about the DFS Replication backlog. If so, you can monitor it via dfsrdiag command. For additional information, I suggest checking the following serverfault question: How to monitor DFSR backlog more efficiently than dfsrdiag

Once you are able to retrieve DFS-R backlog counters, you will be able to verify how fast they are decreasing and estimate how long they will take to reach zero.

Anderson Medeiros Gomes
  • 2,252
  • 8
  • 20
  • There is a DFS-R backlog and we already monitor it. It's not going down since once of the member's database is in auto-recovery followed a crash. There is activity from the process, but seemingly no way to verify progression. – jfrobishow May 25 '20 at 11:54
  • If the backlog counter is not going down, I don't think that your DFS infrastructure is actually auto-recovering from the crash. Is the backlog increasing? Maybe you need to manually resume the DFS replication by following instructions from [KB2846759](https://support.microsoft.com/kb/2846759) or restart the DFS Replication service. Have you checked for relevant DFSR event logs? – Anderson Medeiros Gomes May 25 '20 at 22:45
  • DFS recovery is turned on and the backlog is growing since no replication is currently occurring. [State 3 from SELECT * FROM DfsrReplicatedFolderInfo indicate auto recovery](https://docs.microsoft.com/en-us/previous-versions/windows/desktop/dfsr/dfsrreplicatedfolderinfo). We've checked the process using ProcessMonitor and it's scanning all the files in the share and writing to the DFS database at regular interval (about 6MB/s) and our delta to our Disaster Recovery site is also reporting a lot of outgoing traffic. The data is changing, but there just isn't a way to see progress it would seem. – jfrobishow May 25 '20 at 23:30