Hi,
We have 2 Windows 2012 R2 file servers that use DFS-R to repicate files. Initially about a year and a half ago when this was set up, it appeared to be working correctly. We have 4 DFS replication groups each represents a business data area, data is replicated
from the live site to a DR site, and the connection between the DR site as a sending member back to production has been disabled; the theory being data gets replicated to DR but not back, and if anyone made an error or deleted something at DR it would not
delete live data.
We've since done some investigative work (my colleage does not entirely trust DFS as a product) using Beyond Compare to check the numbers of files and folders on these 4 data areas, the theory being the numbers of files and folders should be the same at both
sites if DFSR is working properly.
We found:
•
Files present on Prod at not at DR
•
Files on both sides have newer time stamp at Live
•
Files on both sides have newer time stamp at DR
•
Files present at DR not Live
•
TMP and ~ files present at DR not Live
I think we can account for files having newer times stamps in live, and why a file would not replicate to live; we do have some PDFs / XPS / JPEGs that have a temp attribute set that are not replicated - this is a known issue and I've used poweshell to re set
the temp attribute and get some of these to replicate.
What I don't understand is why files would be showing as newer on the DR site (users do NOT point directly to this file server, we use a DNS CNAME that points to an (A) record for the live file server and as far as I know it has never been changed to point
to DR). I did run some PS commands to compare the file hashes on the file at live and DR with different timestamps and they had the same hash - when I put the actual files in to beyond compare the content was the same on the ones I've checked. They would appear
to be the same file.
However - I don't understand why there would be a file present on the DR site and not live. I identified such a file orphaned in DR and spoke to the user who thinks she deleted it on live, I did run some WMIC queries against this file that exists only at DR
that shows the file as having a flag of "5" which is "visible & present" e.g. not deleted. I got her to test another deletion and this went through ok no orphans - the WMIC query showing the file was deleted.
I also don't understand why we've got TMP files at DR when DFSR is not supposed to replicate these by default, and again I'm assured no users are pointing at the DR file server directly. Why would this not work?
I've been trying to look for some info on why DFSR would not process deletions, or why time stamps on files would be different and also why tmp files would replicate.
The files affected are a very small proportion of the total number of files, and I suppose you could say we actually have more files in DR than in Live(!!) but I'd still like to get to the bottom of some of these questions, so any advice or experience of similar
issues would be appreciated
Thanks
ND