About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.
If a domain controller that held one or more FSMO (operations master) roles has failed and will not come back, the roles must be seized by a healthy domain controller. Do that with ntdsutil, then remove the dead server's metadata from Active Directory so the rest of the domain stops trying to replicate with it.
Seizing is a one-way decision. Microsoft recommends it only when the previous role holder is not returning to the domain, and the RID master role carries a side effect: a jump in the next RID value that cannot be undone. Work out first which roles were lost, because several of them can wait.
What the error means
Active Directory has five operations master roles. Two are forest-wide (schema master, domain naming master) and three are domain-wide (PDC emulator, RID master, infrastructure master). Each is held by exactly one domain controller at a time. Microsoft's description of the roles says each exists because certain tasks can only be done by one domain controller.
When the holder is down, nothing says "FSMO" on screen. You find out when a role-dependent task fails: a new domain controller cannot get a RID pool, a schema update will not run, a domain cannot be added or removed, or the PDC emulator, which Microsoft says receives password updates from replica domain controllers and is the default target for Group Policy updates, is unreachable. Microsoft's placement page states the availability needs this way:
- The schema master must be online for schema updates.
- The domain naming master must be online when domains or application partitions are added or removed.
- The PDC emulator must be online and accessible 24 hours a day, seven days a week.
- The RID master must be online when a newly promoted domain controller needs a local RID pool.
Roles on an offline domain controller only have to be transferred or seized when you need a task that depends on them. That is a judgement call, not a rule you must hurry past.
Transfer or seize. A transfer is a clean hand-over between two working domain controllers. A seize takes the role without the old holder's agreement. Microsoft recommends seizing when the current holder has an operational error that prevents the task completing and you cannot transfer, when a domain controller that owned a role was force-removed, or when the operating system of the original holder no longer exists or has been reinstalled. The same page advises seizing all roles only when the previous role holder is not returning to the domain.
What happens to the new holder. After a transfer or seizure, the new role holder behaves like a restarted holder. It waits for a successful inbound replication cycle of the relevant partition before acting. If the old holder later starts up, it also waits for inbound replication, learns that another domain controller now holds the role and relinquishes it. Microsoft says this reduces the chance of duplicate holders, but notes that conflicts can still arise in some cases. Do not rely on it as a plan.
Likely causes
The question here is which situation you are in, because it decides the path.
| Cause (situation) | What you see | How to confirm |
|---|---|---|
| Holder is dead and will not return (failed hardware, lost VM, no usable backup) | Role-dependent tasks fail; the server does not answer | Server is off or unreachable and you have decided it will not be recovered |
| Holder was force-demoted | Server still exists as a member server; roles left in an invalid state | Microsoft says the force-demote command can leave roles invalid until reassigned |
| Holder reinstalled or rebuilt with the same name | New operating system, old Active Directory objects still present | The server has no domain controller role, but its NTDS Settings object still exists |
| Holder is only offline for now | Server will return after repair or maintenance | You can bring it online before a role-dependent task is needed |
Only the first three call for a seize. For the fourth, Microsoft says you may delay seizing if the holder can be made operational before the role is needed.
Prove the cause before changing anything
All of these read only. Run them from a healthy domain controller in an elevated prompt.
- Find the current role holders:
netdom query fsmo. Microsoft documents this command for confirming role holders after a seizure. - Note which roles name the dead server. The result is your seize list.
- Check whether the surviving domain controller you plan to use as the new holder is a global catalog. Microsoft's transfer page shows this test, and its placement page ties it to where the infrastructure master may sit. For a surviving server named
DC02:(Get-ADDomainController -Filter { Name -Eq 'DC02' }).IsGlobalCatalogreturnsTrueorFalse. - In Active Directory Sites and Services, expand the site, then Servers, and look for the dead server and its NTDS Settings object. If the object is there, the metadata is still in the directory.
- Decide the new holder. Microsoft's guidance is a domain controller in the same domain as the previous holder that has the most recent replicated writable copy of the role's partition.
Be honest at step 1 about whether the server can return. If any chance remains that it is recovered from a backup or a disk image, stop here. Restoring a seized-over holder is the case Microsoft warns about.
Fix
Sign in to the domain controller that will take the roles. Microsoft's forest recovery page lists the credentials: Enterprise Admins for the domain naming master, Schema Admins for the schema master, and Domain Admins for the PDC emulator, RID master and infrastructure master. Its transfer page says Enterprise Administrators for both forest-wide roles. The two pages differ, so for the schema master hold both Schema Admins and Enterprise Admins.
Seize the roles with Ntdsutil.
Disruptive step: seizing a role changes which domain controller owns it, and the old holder cannot simply be returned to service afterwards. There is no clean way back to the old holder. A later transfer to another working domain controller is possible. Microsoft says the RID master seizure raises the next RID value (see the RID master note below), and its pages describe no way to lower it again.
From an elevated prompt, as Microsoft's forest recovery guide lists them:
ntdsutil
roles
connections
connect to server <ServerFQDN>
quit
seize schema master
seize naming master
seize infrastructure master
seize pdc
seize rid master
quit
quit
<ServerFQDN> is the fully qualified name of the domain controller you are signed in to; if that fails, Microsoft says to use the NetBIOS name. Issue only the seize lines for roles the dead server held. After you confirm each request, Active Directory tries a transfer first, shows error information if that fails, and proceeds with the seizure. A list of roles and holders then appears.
Risk: the RID master. To avoid duplicate SIDs, a RID master seizure raises the next available RID in the pool. Microsoft states that ntdsutil raises it by 10,000 and Move-ADDirectoryServerOperationMasterRole by 30,000 from what it finds in Active Directory, and warns that this can consume RID ranges quickly. Seize the RID master only when you are sure the current one cannot return. If this server was not the RID master before the failure, Microsoft says a dialog may ask whether to continue because it cannot synchronise with a partner; the guide says to click Yes. Back-out: Microsoft's pages describe no way to lower the next available RID value again, so the raise cannot be backed out.
Remove the dead server's metadata. Microsoft states that cleanup is required after a forced removal of AD DS. It removes the data that identifies the domain controller to replication and removes DFS Replication connections.
Microsoft's metadata cleanup page says cleanup also attempts to transfer or seize any operations master roles the removed server still holds, so seize first, as above, and choose the new holder yourself.
In Active Directory Users and Computers, open Domain Controllers, right-click the dead server and choose Delete. In the Deleting Domain Controller dialog, select the option saying the domain controller is permanently offline and can no longer be demoted, then Delete. If the server is a global catalog, a dialog asks you to click Yes to continue. If it still holds operations master roles, a dialog moves them to the domain controller it shows, which you cannot change. Metadata cleanup runs automatically.
Disruptive step: deleting the domain controller object removes its identity from the directory. The only way back is to rebuild the server and promote it again; do not restore an old copy of it.
If you get "Access is denied", Microsoft says to clear Protect object from accidental deletion on the computer object and the NTDS Settings object. In Active Directory Users and Computers this appears under View, Advanced Features, then the object's Object tab.
Risk: clearing that protection removes a safeguard on the object. Tick the box again on any object you are keeping.
Microsoft adds a caution on the final check: if the server object in Active Directory Sites and Services has no NTDS Settings child, you can delete it; if a child object is still there, leave the server object alone because another application is using it.
If the old server is rebuilt. Microsoft's advice is to rebuild a repaired holder from scratch (format and reinstall, or forcibly demote), clean up its metadata on another domain controller, then promote it again. Restoring the old domain controller from a backup rebuilds it as a role holder again.
Confirm the fix
netdom query fsmolists only healthy domain controllers.- The dead server no longer appears under Domain Controllers in Active Directory Users and Computers, and has no NTDS Settings object in Active Directory Sites and Services.
- A role-dependent task completes: for example, promoting a test domain controller after a RID master seizure, once the new holder has finished an inbound replication cycle.
Example (illustrative): in CONTOSO, dc01 held the PDC emulator, RID master and infrastructure master and has failed with no usable backup. dc02 and dc03 are healthy. The administrator sees all three roles on dc01 from netdom query fsmo, signs in to dc02 as a Domain Admin and seizes the three roles. If the RID master's next available RID was 12,300 when Active Directory was read, ntdsutil seizure sets it to 22,300 (12,300 + 10,000). The PowerShell route would have set it to 42,300 (12,300 + 30,000). The administrator then deletes the dc01 computer object with the permanently-offline option and confirms dc01 no longer appears.
What to watch afterwards
- The roles stay with the intended holders. Record them somewhere your team reads.
- Replication between the remaining domain controllers keeps working. Event logs on the new holders show whether a role-dependent operation stalls.
- Leftover references: Microsoft's event 4015 page documents that demoting a DNS-hosting domain controller does not clean up its DNS objects, which can leave orphaned records. That page concerns demotion, not a forced removal, so treat it as a prompt to look for name server and SRV records for the dead server in your zones.
- The infrastructure master: Microsoft says it is not needed when every domain controller is a global catalog or the Recycle Bin is enabled, but a valid holder avoids monitoring errors. Microsoft's placement page says that in a multi-domain forest where not every domain controller in the domain is a global catalog, the infrastructure master must sit on a domain controller that is not a global catalog.
Read next
Related articles you may want next.
- Backup & RecoveryActive Directory Forest Recovery: Authoritative Restore, RID Pool and DC OrderWhich domain controller to restore first, what authoritative means for the directory and SYSVOL, why the RID pool is raised, and health checks.Read article
- Troubleshooting"The directory service was unable to allocate a relative identifier": why you cannot create users or computersCreating a user, group or computer fails with a relative identifier (RID) error. Find out whether one domain controller cannot reach the RID master or the whole domain is running low on RIDs, and fix it safely.Read article
- Backup & RecoveryRecovering deleted Active Directory objects: the Recycle Bin first, backups secondUnderstand Active Directory object recovery, Recycle Bin prerequisites, parent order, retention limits and backup-based recovery decisions.Read article
Risks and limits
These commands have not been run or lab-tested. Menu names and prompts follow Microsoft's pages for Windows Server 2008 and later; check them on your version. This page does not cover recovering a whole forest, which Microsoft's forest recovery guide handles.
Never seize a role from a domain controller you can still reach and transfer from. Do not bring a seized server back onto the network with its old state. Microsoft says a restarted holder relinquishes a role it learns has moved, but also that conflicts between duplicate holders can arise for the schema master, domain naming master and RID master, and recommends rebuilding the server from scratch instead of restoring it.
Sources
- Transfer or seize Operation Master roles — Microsoft Learn, page dated 2025-04-07, accessed 18 November 2026
- AD Forest Recovery - Seizing an Operations Master Role — Microsoft Learn, page dated 2025-05-12, accessed 18 November 2026
- Clean up AD DS server metadata — Microsoft Learn, page dated 2025-12-01, accessed 18 November 2026
- Flexible Single-Master Operation (FSMO) placement and optimization on AD DCs — Microsoft Learn, page dated 2026-02-12, accessed 18 November 2026
- View and transfer FSMO roles — Microsoft Learn, page dated 2026-02-12, accessed 18 November 2026
- Event ID 4015 - The DNS server encountered a critical error — Microsoft Learn, page dated 2026-02-17, accessed 18 November 2026
This article was originally published on .