Use Repadmin to identify which domain controller, replication partner and directory partition is failing before attempting a repair. A replication warning is a symptom; restarting services or forcing synchronization without understanding the error can make an Active Directory incident harder to recover from.
Start with an evidence baseline
Run the checks from an elevated administrative session on a domain controller or a management computer with the AD DS administration tools. Record the forest, affected sites, complaint time and recent changes to DNS, firewalls or domain controllers. Use an account delegated to inspect the environment; do not assume every operator needs permanent Enterprise Admin membership.
Take a second observation later. An isolated failure followed by successful replication has a different meaning from an error that repeats for hours. Protect exported reports because controller names and directory structure can reveal internal infrastructure.
Check the forest summary
Begin with a summary to find controllers with failures or unusually old successful replication. Compare the largest replication delta with the intended site-link schedule; a scheduled intersite delay is different from an unexplained failure.
repadmin /replsummary
repadmin /showrepl * /csv > C:\Temp\ad-replication.csvInspect one failing controller
Review the failing partner and naming context on the destination controller. The report describes inbound replication; the source named in the error is the partner from which changes could not be received. Inspect the domain, Configuration, Schema and DNS application partitions separately rather than assuming one healthy partition proves the whole controller is healthy.
repadmin /showrepl DC02.corp.example
dcdiag /test:replications /s:DC02Trace the dependency behind the error
Name resolution, time, authentication and RPC connectivity must work between the affected partners. Confirm that the controller uses AD-aware DNS servers and can resolve its partner. Check the Directory Service, DNS Server and System logs around the first failed attempt. A network-path test to one port does not validate every dynamic RPC connection required for replication.
Error 1722 often requires investigation of RPC reachability; an access-denied result points toward authentication, permissions or security context. Treat a tombstone-lifetime or lingering-object warning as a recovery case: do not force synchronization, delete directory objects or rebuild a controller as a routine first step.
Confirm recovery and avoid unnecessary changes
After correcting the documented cause, rerun the same summary and partner-level checks. Verify that each affected naming context records a new successful replication and observe a complete scheduled interval. Confirm SYSVOL and application behaviour separately where the original incident affected Group Policy or logon.
These diagnostic commands do not change the directory, so they need no configuration rollback. Any repair does: retain the before-and-after output and a record of DNS, firewall or account changes. Escalate to the directory recovery owner if errors persist or the controller has been disconnected for an extended period.
Microsoft references
See Microsoft’s Active Directory replication troubleshooting guidance and replication verification after forest recovery. Example controller names must be replaced with your own environment.