Fix ERROR_DS_DRA_NO_REPLICA (0X00002104) in 3 Steps

What the Error Means

You're seeing ERROR_DS_DRA_NO_REPLICA (0X00002104) in dcdiag or event logs. Plain English: the source domain controller is trying to replicate a partition (like Domain or Configuration NC) that the destination DC doesn't know about. Common scenarios: you removed a DC improperly, or a lingering object from a decommissioned domain is still trying to sync. I've seen this most often after force-demoting a DC without cleaning up metadata, or when a GC is pointing to a removed domain.

Quick Fix (30 Seconds) – Check Replication Topology

Don't jump to reinstall yet. Run this from an elevated command prompt on the failing DC:

repadmin /showrepl * /csv | findstr /i "0X00002104"

This lists every replication attempt failing with this code. Look at the source and destination DC names. If you see a DC that no longer exists (like an old child domain controller), that's your culprit. The fix: remove that DC from the topology.

If the failing DC is still a valid server, move to the moderate fix.

Moderate Fix (5 Minutes) – Remove Orphaned References with ADSI Edit

This handles the most common cause: a lingering server reference in the Configuration partition.

1. Open ADSI Edit – Run adsiedit.msc as Domain Admin.
2. Connect to Configuration NC – Right-click ADSI Edit → Connect to → Select Configuration from the dropdown. OK.
3. Navigate to CN=Sites,CN=Configuration,DC=yourdomain,DC=com.
4. Drill into CN=Servers under the specific site. Find the orphaned DC's server object (e.g., CN=OLD-DC). Right-click → Delete.
5. Repeat for CN=NTDS Settings under that server object – delete the whole server object.

After deletion, run repadmin /syncall /AdeP on the affected DC. If the error's gone, you're done.

Advanced Fix (15+ Minutes) – Force Replication with Repadmin and Remove Lingering Objects

If the moderate fix didn't work, or the error persists across multiple DCs, you've got corrupted metadata or lingering objects. This is the nuclear option but it works.

Step 1: Identify the Stale Partition

Run this on the source DC (the one trying to push replication):

repadmin /showrepl  /nc:

Look for the Naming Context line that fails. Write it down (e.g., DC=DomainDnsZones,DC=yourdomain,DC=com).

Step 2: Force Removal of the Stale Partition

On the destination DC (the one getting the error), open an elevated Command Prompt. Run:

ntdsutil
metadata cleanup
remove selected server 
quit
quit

This wipes the source DC's metadata from AD. After this, run repadmin /replicate from the destination back to a healthy DC. The error should vanish.

Step 3: Clean Up Lingering Objects (If Still Broken)

Sometimes the replication partner itself has leftover references. On the healthy DC, run:

repadmin /removelingeringobjects   /advisory_mode

Check the output – it lists objects to be removed. If it looks safe, re-run without /advisory_mode:

repadmin /removelingeringobjects  

Then force replication with:

repadmin /syncall /AdeP

When to Rebuild the DC

If all three steps fail (rare, but happens), the DC's AD database is too corrupt to salvage. I've only had to re-promote a DC twice in 14 years for this error – and both times it was because someone manually deleted objects in the NTDS folder. If you're at that point, demote the DC (use dcpromo /forceremoval if needed), clean up metadata with ntdsutil, and promote it fresh. It's 20 minutes of work vs. days of chasing ghosts.

Pro tip: Always run repadmin /showrepl weekly on all DCs. Catching this error early means a 30-second fix instead of an hour-long cleanup.