Summary
In Active Directory Domain Services (AD DS), abandoned objects are objects that exist on some domain controllers (DCs) in the forest but that are never replicated to all DCs. Abandoned objects typically appear in read-only naming contexts (also referred to as partitions) on global catalog servers (GCs) or read-only domain controllers (RODCs). Because standard lingering object removal tools can't remove abandoned objects, they can cause persistent replication errors and confusing search results that are difficult to diagnose.
This article discusses how to identify and remove abandoned objects. It covers the following topics:
- How to recognize symptoms of abandoned objects, including replication errors and transient or phantom objects.
- How abandoned objects differ from lingering objects.
- How to remove abandoned objects by using scripts and manual steps.
- How to use replication metadata, up-to-dateness vectors, and DC identifiers to investigate the root cause of the abandoned objects.
Prerequisites
-
Understanding of lingering objects and how to remove them, as described in the following articles:
-
Understanding of replication components, including:
- Directory Service Agent globally unique identifiers (DSA GUIDs). These agents are also known as DSAs.
- Invocation IDs.
- Update sequence numbers (USNs).
These components are described in the following articles:
Symptoms
You run a Windows Active Directory forest that has multiple domains or Read-Only Domain Controllers (RODCs) or both. You encounter behavior, events, or error messages that resemble any of the following examples.
Example 1: Transient or phantom objects
An application or user searches for an Active Directory object, but the search results are inconsistent.
- The user or application finds the object while connected to an RODC or a GC. However, if the user or application connects to a writable DC, the object appears not to exist.
- If the user or application searches at the forest level, they find the object. If they search at the domain level, the object doesn't appear to exist in any domain.
Similarly, a user or application tries to create an object, but receives an error message that indicates that the distinguished name (DN), user principal name (UPN), or service principal name (SPN) is already in use. However, a search can't find an object that has that specific DN.
Example 2: GC replication errors: "Couldn't update an object" and "There is no such object on the server"
Note:
Unlike GCs, RODCs might not generate events or errors when they contain abandoned objects. GCs can replicate to one another, but RODCs don't replicate to any other DCs.
The event log contains Event ID 1084. This entry indicates that an expected object is missing from the Active Directory instance on the destination GC.
Log Name: Directory Service
Event Source: Microsoft-Windows-ActiveDirectory_DomainService
Event ID: 1084
Task Category: Replication
Level: Error
Description:
The directory replication agent (DRA) couldn't update object <FQDN> (GUID <ObjectGUID>) on this system with changes which have been received from source server <SourceServerGUID>._msdcs.bc.com. An error occurred during the application of the changes to the directory database on this system.
If you run repadmin /showreps at a command prompt on the destination GC, the output includes a message that resembles the following example.
<Site_Name>\<GC> via RPC
DSA object GUID: 2521a874-d379-4281-8744-4bd34c792026
Last attempt @ 2002-01-21 16:10.54 failed, result 8240:
There is no such object on the server.
Note:
In this example, <Site_Name>/<GC> is the site name and NetBios name that identify the destination GC. The Directory System Agent (DSA) object GUID is the permanent identifier of the source GC.
Example 2: "Lingering objects" that you can't remove
In addition to Event ID 1084, you might see Event ID 1988 logged. Typically, this event indicates that Active Directory contains lingering objects. However, in this case, it might indicate abandoned objects.
Event Type: Error
Event Source: NTDS Replication
Event Category: Replication
Event ID: 1988
User: NT AUTHORITY\ANONYMOUS LOGON
Description:
Active Directory Replication encountered the existence of objects in the following partition that have been deleted from the local domain controllers (DCs) Active Directory database. Not all direct or transitive replication partners replicated in the deletion before the tombstone lifetime number of days passed. Objects that have been deleted and garbage collected from an Active Directory partition but still exist in the writable partitions of other DCs in the same domain, or read-only partitions of global catalog servers in other domains in the forest are known as "lingering objects".
This event is being logged because the source DC contains a lingering object which does not exist on the local DCs Active Directory database. This replication attempt has been blocked.
The best solution to this problem is to identify and remove all lingering objects in the forest.
Source DC (Transport-specific network address):
a3a724aa-55cf-4a0f-8889-2b3fd37a72bd._msdcs.contoso.com
Object:
DC=175.3\0ACNF:160e575d-9205-493d-89b8-97f152af40a5,CN=LostAndFound,DC=contoso,DC=com
Object GUID:
160e575d-9205-493d-89b8-97f152af40a5
To distinguish whether you have lingering objects or abandoned objects, open a Command Prompt window, and then run the following command:
repadmin /removelingeringobjects
After the command stops running, run it a second time. If the objects are still present, they're likely to be abandoned objects instead of lingering objects.
When you check the objects by name or GUID, you don't find them on writable DCs.
Cause
The basic scenario is that an object is created on a hub DC that has RODCs or GCs as replication partners, in addition to other read-write replicas. A condition on the DC or in the infrastructure allows the object to replicate to a read-only replica, but not to a writeable replica. A firewall configuration issue or a networking issue could cause this behavior.
Before this replication issue is fixed, the hub DC is forcefully removed from the forest. Because the writeable DCs don't use GCs or RODCs as inbound replication partners, the object can't reach writeable DCs. Therefore, the object is considered to be an abandoned object on the RODCs or GCs.
A similar situation occurs if an object is deleted and then restored on a hub DC that has RODCs or GCs as replication partners, in addition to other read-write replicas. Again, the restoration replicates only to read-only replicas before the hub DC is forcibly removed. On the writeable DCs, the object remains in its tombstone state until its tombstone lifetime expires. Then, the garbage collection process removes the object completely. Again, the object is considered to be an abandoned object on the RODCs or GCs.
Abandoned objects resemble lingering objects. However, lingering objects are objects that at some point existed in all Active Directory replicas, and one or more replicas didn't get the "delete" change. Abandoned objects exist on some read-only replicas, but the remaining replicas never got the "create" change. This difference is one reason that lingering object removal tools might identify abandoned objects, but can't remove them. For more information about how abandoned objects and lingering objects occur and how they differ, see How abandoned objects and lingering objects occur.
Resolution
You might need assistance to follow the proper course of action for your particular issue. You can request Microsoft Technical Support to assist you with resolving this problem.
Important:
These methods focus on removing the abandoned objects instead of creating or restoring the objects on the DCs that they didn't replicate to. For reference information that could help you identify the replication issues that generated the abandoned objects in the first place, see More information.
The algorithm that repadmin /removelingeringobjects uses checks the scope of the up-to-dateness vector. It doesn't remove objects outside the up-to-dateness vector. Abandoned objects, by definition, don't exist in the up-to-dateness vectors of the writeable DCs. Therefore, you have to use tools other than repadmin to remove abandoned objects.
This article describes processes that use manual steps and scripts to remove abandoned objects. The precise details of the steps that you follow depend on the details of your forest and the type of abandoned objects that you have.
Step 1: Collect the GUIDs of the abandoned objects and the DCs that host them
To remove abandoned objects, you have to identify each object by its objectGUID value. You can get this information from events and error messages, and from log files that record operations, such as searches, that produce inconsistent results. You can also use scripts to identify objects that exist in some Active Directory replicas but not in others.
The following example PowerShell script detects abandoned objects. It creates a list of the objects in the affected GC naming context, and then it checks to see whether each object is present in the reference child domain. If an object isn't present, the script exports the object information to a .csv file and an .ldf file. The script doesn't change any of the objects or object attributes.
Note:
In this script, change the value of$GCdomainto the DN of the affected domain, and change$rwdomainto the DN of the reference DC.
Important: This sample script is not supported under any Microsoft standard support program or service.
The sample script is provided AS IS without warranty of any kind. Microsoft further disclaims all implied warranties including, without limitation, any implied warranties of merchantability or of fitness for a particular purpose.
The entire risk arising out of the use or performance of the sample scripts and documentation remains with you. In no event shall Microsoft, its authors, or anyone else involved in the creation, production, or delivery of the scripts be liable for any damages whatsoever (including, without limitation, damages for loss of business profits, business interruption, loss of business information, or other pecuniary loss) arising out of the use of or inability to use the sample scripts or documentation, even if Microsoft has been advised of the possibility of such damages.
$GCdomain = "contoso.com"
$rwdomain = "child.contoso.com"
$rwdomainNC = get-addomain $rwdomain
$GC = "dc2.contoso.com:3268"
$gcdc = "dc2.contoso.com"
$allrwdcs = Get-ADDomainController -filter * -server $rwdomain | select-object hostname
get-adobject -filter '*' -searchbase $rwdomainnc .distinguishedname -server$GC|select-propertyname,distinguishedname,objectguid|export-csv-path.\root.csv-NoTypeInformation
$objects = import-csv .\root.csv
Foreach($objectin$objects)
{
try
{
foreach ($dcin$allrwdcs)
{get-adobject$object.objectguid -server$dc.hostname >$Null}
}
catch {$object|export-csv-path.\abandoned.csv-NoTypeInformation-Append
$name = $object .name
ldifde -f .\ $name .ldf -d $object .distinguishedname -s$gcdc-t3268-lname,objectclass,objectguid
}
}
Step 2: Resolve abandoned objects that were restored on the read-only replica but are still tombstones on the writeable replicas
You can use one or both of the following methods for this step, depending on the details of the abandoned objects.
- If you use method 1, you probably don't have to use step 3. However, method 1 is feasible only under limited circumstances, and it might not resolve all the abandoned objects.
- You can use method 2 in most cases. However, you have to continue to step 3 to complete the removal process.
Method 1: Use the tombstones to reset the replication metadata for the objects across the replication topology
Important:
This method isn't appropriate for every type of object. For example, using this approach on a Domain Controller Computer object that has a child Rid Set object would cause further issues.
You can use this method under the following conditions:
- The abandoned object is restored from a deleted state.
- The object still exists as a tombstone in the other replicas in the forest (
tombstoneLifetimehasn't expired). - The GC or RODC that has the abandoned object is replicating correctly.
In this scenario, you can connect to one of the writeable replicas, undelete the object, and then delete it again. The changes replicate to the other DCs and to the original RODC or GC, and the object's state becomes consistent throughout the forest. The up-to-dateness vectors also synchronize.
Method 2: Clear the tombstones from the replication topology
In many cases, the reference DC has tombstones (or objects in the Recycle Bin) for the abandoned objects that you want to remove, and Method 1 isn't feasible. In such cases, it might not be practical to wait for the tombstones to expire naturally. For example, if you have Recycle Bin enabled, objects remain in the Recycle Bin for msDS-deletedObjectLifetime (180 days by default). At that point, they become recycled objects. Active Directory treats recycled objects as tombstones. Therefore, the objects remain in this state for tombstoneLifetime (180 days by default).
Note:
For GC naming contexts in which a GC initially replicates from a writeable DC and later switches to replicate from a read-only replica on a GC, you might receive the replication errors beforetombstoneLifetimeexpires. Therefore, to let these objects expire and still allow other replication to occur, you might have to disable strict replication consistency, and also allow replication that involves corrupt or divergent partners. For information about how to use these settings, see Troubleshoot Active Directory replication error 8614.
To more quickly clean out the tombstones or deleted objects from the reference DC, use the following methods to shorten the amount of time that these objects remain before the garbage collection process removes them.
-
Remove objects from the Recycle Bin. You can use Windows PowerShell to force deleted objects to become recycled objects before
msDS-deletedObjectLifetimeexpires. To run this process, open an administrative PowerShell Command Prompt window on the reference DC, and then run a command that resembles the following example:Get-ADObject -Filter 'isDeleted -eq $True -and -not (isRecycled -eq $true) -and name -ne "Deleted Objects" -and lastKnownParent -eq "<LastKnownParent_DN>"' -IncludeDeletedObjects | Remove-ADObject[!NOTE]
In this command, <LastKnownParent_DN> represent the DN of the container that held the object before the object was deleted.After you run this command,
tombstoneLifetimecontrols how long the recycled objects remain in the replica. -
Shorten
tombstoneLifetime. If you reducetombstoneLifetimefrom 180 days to only a few days, you can speed up the garbage collection process.[!IMPORTANT]
If your forest has ongoing replication issues, and some DCs are significantly out of sync, setting a lowertombstoneLifetimevalue increases the risk that lingering objects will accumulate in the forest.
Step 3: Remove abandoned objects that don't have tombstones
You can use this method if the abandoned object doesn't exist in any form (such as a tombstone object or an object in the Recycle bin) on the writable DC that you use as a reference DC. Verify that the objectGUID of the abandoned object doesn't exist anywhere in the replica on the reference DC.
To remove abandoned objects, follow the process in Manually Remove Lingering Objects on Outdated Replication Partners. If you used a script to create a list of abandoned objects, you can use that list as input for the scripts that this process uses.
The linked article provides detailed instructions. Although it's presented as a manual process for removing lingering objects, it works for abandoned objects, also. The article includes scripts that run the ldp.exe tool's removelingeringobject operational attribute at the RootDSE level of the reference DC's replica.
After you complete that process, run the following command to make sure that the abandoned object doesn't exist in the forest.
repadmin /showobjmeta \* "\<guid=orphaned objectGUID\>"
This command checks the existence of the abandoned object on all DCs in the forest. If the object is successfully removed, the command generates output that resembles the following example:
DsReplicaGetInfo() failed with status 8439 (0x20f7):
The distinguished name specified for this replication operation is invalid.
More information
This section provides detailed background information that can help you diagnose how your forest generated abandoned objects.
How abandoned objects and lingering objects occur
The following table provides a simplified view of how lingering objects and abandoned objects occur. It highlights the similarities and differences between the three types of objects.
Important:
The following table describes sequences of events. These sequences aren't troubleshooting instructions.
| Sequence | Abandoned object type 1 | Abandoned object type 2 | Lingering object |
|---|---|---|---|
| State 1 | The object doesn't exist. | The object exists on all DCs. | The object exists on all DCs. |
| State 2 | The object doesn't exist. | The object exists on all DCs. | The object exists on all DCs.One DC (DC-1) disconnects from the network. |
| State 3 | The object doesn't exist. | The object is deleted. The "delete" change replicates to all DCs. | The object is deleted. The "delete" change replicates to all DCs except DC-1. |
| State 4 | The object is created on DC-2.The "create" change replicates to GCs or RODCs.Before the "create" change replicates to regular DCs, DC-2 disconnects from the network. | Before the tombstone lifetime expires, the object is restored (undeleted) on DC-3.The "restore" change replicates to GCs or RODCs.Before the "restore" change replicates to regular DCs, DC-3 disconnects from the network. | All DCs except DC-1 have the object's tombstone. DC-1 has the original object. |
| State 5 | Some GCs or RODCs have the object. Writeable DCs don't have the object. | Some GCs or RODCs have the object. Writeable DCs have the object's tombstone.When the tombstone lifetime expires, the tombstones are removed. | All DCs except DC-1 have the object's tombstone. DC-1 has the original object.When the tombstone lifetime expires, the tombstones are removed.Finally, DC-1 reconnects to the network. |
| Result | Some GCs or RODCs have the object, and the "create" change. Regular DCs don't have the object or the "create" change, and don't receive inbound replication from GCs or RODCs. | Some GCs or RODCs have the object, and the "restore" change. Regular DCs don't have the object or the "restore" change, and don't receive inbound replication from GCs or RODCs. | Only DC-1 has the object. DC-1 doesn't have the "delete" change. The other DCs don't have any reference to the object beyond the change history. |
Investigating abandoned objects: How to interpret information from replication partners and up-to-dateness vectors
Finding replication issues can be a complicated effort for many reasons. One reason is that replication reports and records typically use multiple identifiers to track a single DC, and those identifiers change over time. Each DC tracks inbound replication in the following terms:
-
The identity of the replication partner
-
The state of the Active Directory replica on that replication partner, including whether (and when) it was restored from backup at any point
-
Whether the replication partner has changes that haven't yet replicated
-
All changes that the replication partner sent previously, both before and after each restoration
Identifying replication partners
To trace abandoned objects, you need two GUIDs for each of the affected DCs and their inbound replication partners. The following table describes these GUIDs.
| GUID type | Also known as | Description |
|---|---|---|
| DSA object GUID | DSA GUID | This GUID is the DC's permanent identifier. Windows creates this GUID when you promote a Windows-based server to a DC. This GUID doesn't change. |
| DSA invocationID | invocationID | This GUID identifies the current copy of Active Directory on the DC. Windows configures this GUID when you add the DC to a new or existing domain, and changes it when the DC is restored from a backup. When you promote a DC and create a new forest, that DC uses its DSA GUID as its invocation ID until the first time the DC is restored from a backup. |
Note:
If you stop and deallocate a virtual DC that runs in Azure, the next time that you start the DC, Windows behaves as if Active Directory were restored from a backup. The DC's invocation ID changes automatically. Because of this behavior, a virtual DC might change its invocation ID at a different rate than a physical DC does.If you have to stop an Azure virtual DC, you can avoid this behavior by shutting down the VM from within the guest operating system instead of by using the Azure portal. After the virtual DC shuts itself down, you can deallocate it. For more information, see Manageability in "Deploy AD DS in an Azure virtual network."
For more general best practices to follow when you restore virtualized DCs, see Virtualization safeguards in "Virtualized Domain Controller Deployment and Configuration."
Event ID 1084 and event ID 1988 both identify a source server. This server is typically a GC that acts as an inbound replication partner to the DC that generated the events (the destination server, typically also a GC). The abandoned object exists on the source server, but not on the destination server.
Take a snapshot of the current replication status of your forest, including the current DC identifiers (DSA GUIDS and invocation IDs). To get the snapshot, run a command that resembles the following command at a command prompt on one of the DCs:
repadmin /showrepl *
Note:
For large topologies, you can constrain the command to run on a particular DC. For more information, see Repadmin. You can also runrepadmin /?to view the online help for this command.
For a forest that has two GCs, this command generates output that resembles the following example:
Repadmin: running command /showrepl against full DC GC-1.contosotest.com
Default-First-Site-Name\GC-1
DSA Options: IS_GC
Site Options: (none)
DSA object GUID: 8e294159-140d-41f8-b041-23896e86cb00
DSA invocationID: 9789f50b-6b9f-4632-ba93-dccf617325a2
==== INBOUND NEIGHBORS ======================================
DC=contosotest,DC=com
Default-First-Site-Name\GC-2 via RPC
DSA object GUID: 0dd82239-b6e1-4f03-9899-76a80ad9f88a
Last attempt @ 2026-02-10 15:52:38 was successful.
. . .
Repadmin: running command /showrepl against full DC GC-2.contosotest.com
Default-First-Site-Name\GC-2
DSA Options: IS_GC
Site Options: (none)
DSA object GUID: 0dd82239-b6e1-4f03-9899-76a80ad9f88a
DSA invocationID: 3e04086d-da99-465f-b82d-e74f31a1caf1
==== INBOUND NEIGHBORS ======================================
DC=contosotest,DC=com
Default-First-Site-Name\GC-1 via RPC
DSA object GUID: 8e294159-140d-41f8-b041-23896e86cb00
Last attempt @ 2026-02-10 16:07:51 was successful.
This example identifies the following DCs:
| DC name | DSA GUID | Current invocation ID |
|---|---|---|
| GC-1 | 8e294159-140d-41f8-b041-23896e86cb00 | 3e04086d-da99-465f-b82d-e74f31a1caf1 |
| GC-2 | 0dd82239-b6e1-4f03-9899-76a80ad9f88a | 9789f50b-6b9f-4632-ba93-dccf617325a2 |
Deriving the replication history from up-to-dateness vectors
Each DC maintains an up-to-dateness vector as a record of its own state and the state of each of its replication partners. To see a DC's up-to-dateness vector, run a command that resembles the following command at a command prompt on the DC:
repadmin /showutdvec GC-1 dc=contosotest,dc=com
For a forest that has two GCs, this command generates output that resembles the following example:
Caching GUIDs.
..
Default-First-Site-Name\GC-2 @ USN 24987 @ Time 2026-02-10 16:56:31
Default-First-Site-Name\GC-1 @ USN 28677 @ Time 2026-02-10 15:07:21
Default-First-Site-Name\GC-1 @ USN 28957 @ Time 2026-02-10 17:03:15
Default-First-Site-Name\GC-2 (retired) @ USN 24580 @ Time 2026-02-10 15:10:48
This output is the up-to-dateness vector of GC-1. This vector records both the changes that originated on GC-1 and the changes that replicated in from GC-2. Each GC uses its own independent USN series. GC-1 and GC-2 each has two entries in the up-to-dateness vector. This condition indicates that both GCs were restored from a backup. This output provides enough data to derive the following information:
-
GC-1 was restored from backup some time between 15:07:21 and 17:03:15. The changes that were recorded at 15:07:21 were the last changes to be recorded before the restoration.
-
When GC-1 started recording changes after the restoration, it created a new entry in its up-to-dateness vector. The most recent changes that originated on GC-1 occurred at 17:03:15.
-
GC-2 was restored from backup some time between 15:10:48 and 16:56:31. The changes that were recorded at 15:10:48 were the last changes that replicated to GC-1 before GC-2 was restored.
-
When GC-2 started replicating changes after the restoration, GC-1 assigned it a new entry in its up-to-dateness vector. The most recent changes that originated on GC-2 and then replicated to GC-1 occurred at 16:56:31.
-
The old entry for GC-2 is labeled as "retired" to distinguish it from the entry that records replication that occurred after GC-2 was restored.
[!NOTE]
The first entry of the first DC in the forest might not be marked as "retired," even if it's no longer active.
To ge a slightly different view of the up-to-dateness vector, run the same repadmin command together with the /nocache option. This command resembles the following example:
repadmin /showutdvec GC-1 dc=contosotest,dc=com /nocache
This time, the vector entries in the output use GUIDs instead of site and server names.
3e04086d-da99-465f-b82d-e74f31a1caf1 @ USN 24987 @ Time 2026-02-10 16:56:31
8e294159-140d-41f8-b041-23896e86cb00 @ USN 28677 @ Time 2026-02-10 15:07:21
9789f50b-6b9f-4632-ba93-dccf617325a2 @ USN 28958 @ Time 2026-02-10 17:03:59
ae04509e-574a-4487-914c-7c08df548db9 @ USN 24580 @ Time 2026-02-10 15:10:48
You can compare these GUIDs to the DSA and invocation ID GUIDs from the output of repadmin /showrepl. You can also match USNs and timestamps between this set of entries and the previous example. In this example, replication continued after the first run of repadmin /showutdvec. Therefore, the second run produces some slight changes in USNs and timestamps. The two examples together provide enough data to derive the following information:
-
Each entry has a unique GUID. In this example, you can compare the GUID version of the output to the site or server name version of the output to associate GUIDs and GCs and identify the two entries for each GC.
-
To further identify the GUIDs, compare the GC and GUID combinations to the output of
repadmin /showrepl. From this data, you can derive the following information:- The oldest entry for GC-1 contains GC-1's DSA GUID. This entry indicates that GC-1 is the first DC in the forest. The second entry for GC-1 contains its current invocation ID.
- None of the entries for GC-2 use its DSA GUID. GC-2 is the second DC in the forest. When you add a DC to an existing forest, Windows immediately assigns an invocation ID to it. The oldest entry for GC-2 uses this first invocation ID.
In this example, GC-2 also accepts inbound replication from GC-1. When you run repadmin /showutdvec against GC-2, GC-2 generates its view of its own changes and the changes that were replicated from GC-1.
Caching GUIDs.
..
Default-First-Site-Name\GC-2 @ USN 25016 @ Time 2026-02-10 17:03:02
Default-First-Site-Name\GC-1 @ USN 28677 @ Time 2026-02-10 15:07:21
Default-First-Site-Name\GC-1 @ USN 28957 @ Time 2026-02-10 17:02:30
Default-First-Site-Name\GC-2 (retired) @ USN 24580 @ Time 2026-02-10 15:10:48
You can compare the historical data for GC-1 and GC-2 to show that the up-to-dateness vectors are consistent:
| Entry | Reported by GC-1 | Reported by GC-2 |
|---|---|---|
| GC-1, DSA GUID | @ USN 28677 @ Time 2026-02-10 15:07:21 |
@ USN 28677 @ Time 2026-02-10 15:07:21 |
| GC-2, retired | @ USN 24580 @ Time 2026-02-10 15:10:48 |
@ USN 24580 @ Time 2026-02-10 15:10:48 |
| Entry | Reported by GC-1 | Reported by GC-2 |
|---|---|---|
| GC-1 | @ USN 28958 @ Time 2026-02-10 17:03:59 |
@ USN 28957 @ Time 2026-02-10 17:02:30 |
| GC-2 | @ USN 24987 @ Time 2026-02-10 16:56:31 |
@ USN 25016 @ Time 2026-02-10 17:03:02 |
This table shows that each GC's view of the other is slightly out of date. GC-1 hasn't yet replicated change 28958 to GC-2, and GC-2 hasn't yet replicated changes 24988-25016 to GC-1. In this situation, even though the numbers don't match, the highest USNs correspond with the latest timestamps. This condition is expected, and the next replication cycle should synchronize changes on both GCs.
Finding indicators of abandoned objects in the replication partner data and the up-to-dateness vectors
If the naming context that you're investigating has abandoned objects, you might see the following discrepancies (in addition to the error messages that the "Symptoms" section describes):
- The invocation ID (or DSA GUID) of the source GC never appears in the up-to-dateness vector. This result indicates that the source GC never finished the initial full replication cycle of the naming context to the destination GC.
- The invocation ID (or DSA GUID) of the source GC exists in the up-to-dateness vector, but its timestamp of the most recent entry isn't current. This result indicates that the source GC stopped replicating or was removed from the replication topology.
To help determine what caused the issue, you can retrieve up-to-dateness information from other DCs or GCs that use the affected GC as a replication source. For example, you can identify which servers didn't get information from the affected source GC and which did to help isolate a network infrastructure issue.
References
- Active Directory replication Event ID 1388 or 1988 - A lingering object is detected
- Active Directory replication Event ID 2042 (It has been too long since this machine replicated)
- The AD Recycle Bin: Understanding, Implementing, Best Practices, and Troubleshooting
- Deploy AD DS in an Azure virtual network
- Lingering objects in an AD DS forest
- [MS-ADTS]: Tombstone Lifetime and Deleted-Object Lifetime
Remove-ADObject- Safely virtualizing Active Directory Domain Services (AD DS)
- Troubleshoot AD Replication error 8240 There is no such object on the server
- Troubleshoot replication error 8614
- Virtualized Domain Controller Deployment and Configuration
- A Windows Server domain controller logs Directory Services event 2095 when it encounters a USN rollback
Source: Microsoft Learn / IT Pro troubleshooting by Microsoft and contributors. Documentation licensed CC BY 4.0. Code samples are covered by the MIT License. Converted from Markdown, with links normalized and embedded media omitted.
