My organisation uses DFS to replicate three servers - a hub at one site, a spoke at one site, and a spoke at a remote site. They contain a number of folders that are separate shares, all within the same replication group. So the namespace is CompFileShare, and it contains IT, Public, etc as shares that users can map based on permissions.
Our remote site's drive recently filled up, and since the drive it was hosted on was MBR and maxed out at 2 TB, we created a new 4 TB GPT drive. We then robocopied the MBR drive to the GPT drive using the recommended xcopy command flags, and let it finish. After it finished, we unshared the original IT folder, shared the new IT folder on the GPT drive, and changed the replication target from MBR to GPT and let it replicate.
For some reason, possibly unrelated, we are now seeing the remote site's fileshare server throwing Event ID 516 every hour:
DFSN service has started performing complete refresh of metadata for namespace TTFileShare. This task can take time if the namespace has large number of folders and may delay namespace administration operations.
Because it is running, all replication to the remote site is halted, even for shares still on the old MBR drive. It has been sitting like this for a day before I really got a chance to look at it. Any tips or places I should look to resolve this issue?
So...
Part of the migration reason was that our MBR drive was completely full. 100 Kbs left.
We deleted a folder (that was already robocopied to the new drive) from the full drive, (freed up ~20 MB) and restarted the server.
The server then replicated the everything, no problems. So lesson learned - don't let drives get so full DFS can't even manage their data.
EDIT: Also, the namespace refresh is flagged as warning when it should be flagged as informational. If you look under informational messages you will see the namespace refresh finished message shortly after the namespace refresh started message. This is the actual cause of the title problem - a microsoft bug mislabeling namespace refresh as a warning. Its just informational