Quantcast
Channel: File Services and Storage forum
Viewing all articles
Browse latest Browse all 10672

Re-creating DFS Replication between two folders

$
0
0

I have two DFS servers: SERVER1 and SERVER2. I have a namespace called "COMPANY" and one of the replicated folders is called "USERDATA" and it contains around 880gb worth of data. The DFS path is \\Company\UserData

SERVER1 had some problems and I had to remove \\SERVER1\UserData as a DFS folder target, and I turned off replication with SERVER2.  When it came time to bring it back online, I followed some advice and used Robocopy to make sure the data in \\SERVER1\UserData matched the UserData folder on Server2 exactly. This is the command I used:


robocopy \\Server2\UserData \\Server1\UserData /e /b /copyall /r:6 /w:5 /MT:64 /tee /log+:d:\robocopy_dep.txt /v



To be sure, I then ran the command in reverse:


robocopy \\Server1\UserData \\Server2\UserData /e /b /copyall /r:6 /w:5 /MT:64 /tee /log+:d:\robocopy_dep.txt /v



The results show the two folders are now identical; there were no extra files, and no files were copied when I ran the second command because everything was identical already.

I then made sure to delete the "DFSRPrivate" folder from each server's share. In DFS manager, I deleted the replicated folder "UserData" in the replication group (but not the whole replication group). I made sure the changes had replicated to SERVER1 and SERVER2.

I created the replicated folders again and set SERVER2 as the primary. After about 15 minutes I checked to see if there was a backlog. There was NO Backlog going from SERVER2 to SERVER1, but there was a backlog of 10k files, which had grown to 50k files within 10 minutes going from SERVER1 to SERVER2. The backlog list of the first 100 files was listingevery subfolder in the UserData share, so it seems like it is queuing up the entire contents of the folder, even though they are identical.

It does not appear to be moving anything as far as I can tell since the size of the UserData share on SERVER2 remains the same, as does the UserData share on SERVER1. But having a backlog of hundreds of thousands of files is just not good.

Am I doing something wrong? Can this be avoided or is it just part of the process and I have to wait for it to work itself through?

Thanks


Viewing all articles
Browse latest Browse all 10672

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>