File Replication Service Broken? Of course!

Every so often, one of your domain controllers (or other servers, if you're making use of FRS shares) will stop replicating Group Policies, NETLOGON scripts, or other data. Oh yes, it will happen. And if it has already happened, oh yes, it will happen again. Here's the not-to-miss quick fix.

If you have multiple Windows 2000/2003 Server domain controllers and you have received complaints that group policies are only intermittently applied to users or computers, you may be a victim of a File Replication Services burnout. This can affect the replication of policies, scripts, files in the SYSVOL/NETLOGON directories, and anything else you are replicating with the FRS service. FRS burnout can happen as a result of network drops, file corruption, or maybe it's just Tuesday – sometimes it just happens on Tuesday.

The best fix that just doesn't get nearly enough press on the Internets (but actually appears as a recommendation in the event viewer if you happen to be the lucky recipient of Event ID: 13568, a.k.a. the JRNL_WRAP_ERROR) is to go through the following steps:

1. Log in to the server with the suspected replication problem (Replmon, with its Show Group Policy Object Status can be a fantastic resource in the case of FRS burnout)

2. Navigate to this key in Regedit: HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\NtFrs\Parameters

3. Edit (or add as a DWORD value) "Enable Journal Wrap Automatic Restore", and set to 1

4. Wait for at least 5 minutes. While you’re at Starbucks, the NTFRS service will remove the system from the replication set, remove the local copy of the replicated data, and place the data in a backup folder called, “NtFrs_PreExisting___See_EventLog”, which you can delete when all of this is over.

5. When you’ve returned, set the value back to 0.

In theory, you should see positive event viewer items such as Event ID: 13553, which signals that the replication relationships have been rebuilt. Replication has only just started so depending on how much data you have in the replicated directories, it could already be finished, or it may be a while before the fresh replication is complete.

I’ve seen some pretty odd replication problems, but going through this process on the right replica members usually sorts things out.

May you never again be a victim of FRS burnout. But, oh yes, you will be.