Article ID: 2028495 - View products that this article applies to.
This article describes the symptoms and cause of an issue in which Active Directory replication is unsuccessful and generates error 8606: "Insufficient attributes were given to create an object. This object may not exist because it may have been deleted." This article also describes a resolution for this issue.
Symptom 1The DCDIAG reports that the Active Directory Replications test failed with error 8606: "Insufficient attributes were given to create an object."
Starting test: Replications
[Replications Check, <Destination DC>] A recent replication attempt failed:
From <source DC> to <destination DC>
Naming Context: <directory partition DN path>
The replication generated an error (8606):
Insufficient attributes were given to create an object. This object may not exist because it may have been deleted and already garbage collected
The failure occurred at <date> <time>
The last success occurred at <date> <time>
Symptom 2Incoming replication that is triggered by the Replicate Now command in the Active Directory Sites and Services snap-in DSSITE.MSC is unsuccessful and generates the error "Insufficient attributes were given to create an object." When you right-click a connection object from a source DC and then select Replicate now, the replication is unsuccessful and generates the following error: "Access is denied." Additionally, you receive the following error message :
Dialog title text: Replicate Now
Dialog message text: The following error occurred during the attempt to synchronize naming context <%active directory partition name%> from domain controller <source DC> to domain controller <destination DC>:
Insufficient attributes were given to create an object. This object may not exist because it may have been deleted and already garbage collected.
The operation will not continue
Symptom 3Various REPADMIN.EXE commands fail with error 8606. These commands include but are not limited to the following:
Collapse this tableExpand this table
Symptom 4Event 1988 is logged shortly after one of the following events occurs:
Symptom 5NTDS replication event 1988 may be logged in the Directory Service event log of domain controllers that are trying to inbound-replicate Active Directory.
Source: NTDS Replication
Event ID: 1988
User: NT AUTHORITY\ANONYMOUS LOGON
Computer: <hostname of DC that logged event, aka the "destination" DC in the replication attempt>
Description: The local domain controller has attempted to replicate the following object from the following source domain controller. This object is not present on the local domain controller because it may have been deleted and already garbage collected.
Source domain controller:
<fully qualified GUIDED CNAME of source DC>
<DN path of live object on source DC>
<object GUID of object on source DCs copy of Active Directory>
Error 8606 is logged when the following conditions are true:
If the destination domain controller was configured to use loose replication consistency, the object would have been "reanimated" on the destination domain controller's copy of the directory. Specific variations that can cause error are 8606 documented in the "More Information" section. However, the error is caused by one of the following:
When you troubleshoot 8606 errors, think about the following points:
1. Identify the current value for the forest-wide TombStoneLifeTime setting
c:\>repadmin /showattr . "CN=Directory Service,CN=Windows NT,CN=Services,CN=Configuration,DC=forest root domain,DC=TLD> /atts:tombstonelifetime2. For each destination domain controller that is logging the 8606 error, filter the Directory Service event log on NTDS Replication event 1988.
See the "Tombstone lifetime and replication of deletions" section in the following article in the Microsoft Knowledge Base:
(http://support.microsoft.com/default.aspx?scid=kb;EN-US;910205)Information about lingering objects in a Windows Server Active Directory forest.
3. Collect metadata for each unique object that is cited in the NTDS Replication event 1988.
From each 1988 event that is logged on the destination domain controller that cites a new object, populate the following table:
Collapse this tableExpand this table
Columns 1 through 5 of this table can be populated by reading values directly from fields in the NTDS Replication 1988 events that are logged in the Directory Service event logs of the destination domain controllers that are logging either the 1988 event or the 8606 replication status.4. Select the appropriate action based on the object metadata cited in the 1988 event.
The date stamps for LastKnownParent and IsDeleted columns can be determined by running "repadmin /showobjmeta" and referencing the objectguid of the object that is cited in the NTDS replication 1988 event. To do this, use the following syntax:
c:\>repadmin /showobjmeta <fqdn of source DC from 1988 event> "<GUID=GUID of object cited in the 1988 event>"
The date stamp for LastKnownParent identifies the date on which the object was deleted. The date stamp for IsDeleted tells you when the object was last deleted or reanimated. The version number tells you whether the last modification deleted or reanimated the object. An IsDeleteted value of 1 represents an initial delete. Odd-numbered values greater than 1 indicate a reanimation following at least one deletion. For example, an isDeleted value of 2 represents an object that was deleted (version 1) and then later undeleted or reanimated (version 2). Later even-numbered values for IsDeleted represent later reanimations or undeletes of the object.
Error 8606 / NTDS Replication event 1988 event is most frequently caused by long-term replication failures that are preventing domain controllers from inbound-replicating knowledge of all originating deletes in the forest. This results in lingering objects on one or more source domain controllers.
Review the metadata for object that is listed in the table that was created in step 4 of the "Resolution" section.
If the object in the 1988 event is either ((live on the source domain controller) but (deleted on the destination domain controller for longer than tombstone lifetime expiration)), see "Removing Lingering Objects" and "." Objects in this condition must be manually removed by an administrator.
Deleted objects may have been prematurely purged from the deleted objects container if system time jumped forward in time on the destination domain controller. Review the "Check for Time Jumps" section.
If the object that is cited in the 1988 event exists in the deleted objects container of the source domain controller and its delete date is right at the cusp of tombstone lifetime expiration in such a way that the object was reclaimed by garbage collection by one or more destination domain controllers and will be reclaimed by garbage collection at the next garbage-collection interval on source domain controllers (that is, the lingering objects are transient), you have a choice. Either wait for the next garbage collection execution to delete the object, or manually trigger garbage collection on the source domain controller. See "Manually starting garbage collection." The introduction of the first Windows Server 2008 R2 domain controller, or any change in the partial attribute set, can cause this condition.
If repadmin /showobjmeta output for the object that is cited in the 1988 event has a LastKnownParent value of 1, this indicates that the object was deleted, and an IsDeleted value that of 2 or some other even-numbered value, and that date stamp for IsDeleted is at the cusp of the tombstone lifetime number of days different from the date stamp for LastKnownParent, then the object was deleted and then undeleted / auth-restored while it was still live on the source domain controller but already reclaimed by garbage collection by destination domain controllers logging error 8606 / Event 1988. See Reanimations at the cusp of TSL expiration
How to remove lingering objectsThe following two commands in REPADMIN.EXE can remove lingering objects from the following directory partitions:
REPADMIN /REMOVELINGERINGOBJCTS can be used to remove lingering objects from writable and read-only directory partitions on Windows Server 2003 source domain controllers. The syntax is as follows:
c:\>repadmin /removelingeringobjects <Dest_DSA_LIST> <Source DSA GUID> <NC> [/ADVISORY_MODE]
REPADMIN /REHOST can be used to remove lingering-objects domain controllers that host a read-only copy of a domain directory partition from domain controllers that are running Windows 2000 SP4 or a later version. The syntax is as follows:
<Dest_DSA_LIST> is the name of a domain controller that is running Windows Server 2003 or a later version and that contains lingering objects (such as the source domain controller that is cited in the NTDS Replication 1988 event).
<Source DSA GUID> is the name of a domain controller that is running Windows Server 2003 or a later version and that hosts a writable copy of the directory partition that contains lingering objects to which the domain controller in <Dest_DSA_LIST> has network connectivity. The DC to be cleaned up (first DC specified in the command) must be able to connect directly to port 389 on the DC that hosts a writable copy of the directory partition (specified second in the command).
<NC> is the DN path of the directory partition that is suspected of containing lingering objects, such as the partition that is specified in a 1988 event.
c:\>repadmin /rehost DSA <Naming Context> <Good Source DSA Address>
DSA is the name of a domain controller that is running Windows 2000 SP4 or a later version and that hosts a read-only domain directory partition for a nonlocal domain. For example, a GC in root.contoso.com can rehost its read-only copy of child.contoso.com but cannot rehost root.contoso.com.
<Naming Context> is the DN path of a read-only domain directory partition that is residing in a global catalog.
<Good Source DSA Address> is the name of a domain controller that is running Windows 2000 SP4 or a later version and that hosts a writable copy of <Naming Context>. The domain controller must be network-available to the DSA computer.If the lingering object that is reported in the 1988 event is not removed by repadmin, evaluate whether the object on the source domain controller was created in USN gap, or whether the objects originating domain controller does not exist in the source domain controller's up-to-dateness vector as documented in Microsoft Knowledge Base article 948071.
Note Lingering objects can also be removed by using repldiag.exe.
(http://activedirectoryutils.codeplex.com/releases/view/13664)This tool automates the repadmin /removelingeringobjects process.
Monitoring Active Directory replication health dailyIf error 8606 / Event 1988 was caused by the domain controller's failing to replicate Active Directory changes in the last tombstone lifetime number of days, make sure that Active Directory replication health is being monitored on a day-to-day basis going forward. Replication health may be monitored by using a dedicated monitoring application or by viewing the output from the one inexpensive but effective option to run "repadmin /showrepl * /csv" command in a spreadsheet application such as Microsoft Excel. (See "Method 2: Monitor replication by using a command-line" in Microsoft Knowledge Base article 910205).
Domain controllers that have not inbound-replicated in 50 percent of tombstone lifetime number of days should be put in a watch list that receive priority admin attention to make replication operational. Domain controllers that cannot be made to successfully replicate should be force-demoted if they have not replicated within 90 percent of TSL.
Replication failures that appear on a destination domain controller may be caused by the destination domain controller, by the source domain controller, or by the underlying network and DNS infrastructure.
Check for time jumpsTo determine whether a time jump occurred, check date stamps in Event and diagnostic logs (Event Viewer, repadmin /showreps, netlogon logs, dcdiag reports) on destination domain controllers that are logging error 8606 / NTDS Replication 1988 events for the following:
Microsoft Support teams have seen system time on production domain controllers incorrectly jump hours, days, weeks, years, and even tens of years in the past and future. If system time was found to be inaccurate, you should correct it and then try to determine why time jumped and what can be done to prevent inaccurate time going forward vs. just correcting the bad time. Possible questions to ask include the following:
Microsoft Knowledge Base article 884776
(http://support.microsoft.com/default.aspx?scid=kb;EN-US;884776)documents steps to help protect domain controllers from “listening” to invalid time samples. More information about MaxPosPhaseCorrection and MaxNegPhaseCorrection is available in the W32Time Blog
(http://blogs.msdn.com/w32time/default.aspx)post Configuring the Time Service: Max[Pos/Neg]PhaseCorrection
(http://blogs.msdn.com/w32time/archive/2008/02/28/configuring-the-time-service-max-pos-neg-phasecorrection.aspx). Microsoft Knowledge Base article 961027
(http://support.microsoft.com/default.aspx?scid=kb;EN-US;961027)describes some helpful precision updates when you are configuring time-based settings in policy.
Check for lingering objects by using "repadmin /removelingeringobjects /advisorymode," and then remove them as required.
Relax "Allow replication with Divergent and corrupt partner" as required.
How to manually start garbage collectionSeveral options exist to manually trigger garbage collection on a specific domain controller:
Run "repadmin /setattr "" "" doGarbageCollection add 1"
Run LDIFDE /s <server> /i /f dogarbage.ldif where dobarbage.ldif contains the text:
dn:Note The final "-" character is a required element of the .ldif file.
Reanimations at the cusp of TSL expirationFor this condition to exist, repadmin /showobject "<GUID=object guid for object in 1988 event>" should report that the object is
"not found" on the destination domain controller but is live on the source domain controller and is either a deleted or a nondeleted object.
A review of key fields from repadmin /showobjmeta on the source domain controller should report that the following are true:
LastKnownParent has a value of 1, and its date stamp is at the cusp of TSL days in the past. Its date stamp indicates the delete date of the object.
IsDeleted has a version number of 2 (or another even-numbered value), where version 1 defined the original deletion and version 2 is the restore / reanimation. The date stamp for "isDeleted=2" should be ~ TSL number of days later than the last change date for LastKnownParent.
Evaluate whether the object in question should remain a live object or a deleted object. If LastKnownParent has a value of 1, something or someone deleted the object. If this was not an accidental deletion, chances are good that the object should be deleted from any source domain controllers that have a live copy of the object.
If the object should exist on all replicas, the options are as follows:
Causes of lingering objects
Cause 1: The source domain controller sends updates to objects that have already been reclaimed by garbage collection on the destination domain controller because the source domain controller either was offline or failed replication for TSL elapsed number of days.The CONTOSO.COM domain contains two domain controllers in the same domain. Tombstone lifetime = 60 days. Strict replication is enabled on both domain controllers. DC2 experiences a motherboard failure. Meanwhile, DC1 makes originating deletes for stale security groups daily over the next 90 days. After it is offline for 90 days, DC2 receives a replacement motherboard, powers up, and then originates a DACL or SACL change on all user accounts before it inbound-replicates knowledge of originating deletes from DC1. DC1 logs 8606 errors for updates security groups that are purged on DC1 for the first 30 days that DC2 was offline.
Cause 2: The source DC sends updates to objects at the cusp of TSL expiration that have already been reclaimed by garbage collection by a strict mode destination DC.The CONTOSO.COM domain contains two domain controllers in the same domain. Tombstone lifetime = 60 days. Strict replication is enabled on both domain controllers. DC1 and DC2 replicate every 24 hours. DC1 originates-deletes daily. DC1 is in-place upgraded to W2K8 R2. This stamps new attributes on all objects in the configuration and writable domain partitions. This includes objects that are currently in the deleted objects container. Some of them were deleted 60 days ago and are now at the cusp of tombstone expiration. DC2 reclaims some objects by garbage collection that were deleted TSL days ago before the replication schedule opens with DC2. Error 8606 is logged until DC1 reclaims the blocking objects by garbage collection.
Any updates to the partial attribute set can cause temporary lingering objects that will clear themselves up after source domain controllers garbage-collect deleted objects at the cusp of TSL expiration (for example, the addition of the first W2K8 R2 domain controller to an existing forest).
Cause 3: A time jump on a destination domain controller prematurely speeds up the garbage collection of deleted objects on a destination domain controller.The CONTOSO.COM domain contains two domain controllers in the same domain. Tombstone lifetime = 60 days. Strict replication is enabled on both domain controllers. DC1 and DC2 replicate every 24 hours. DC1 originates deletes daily. The reference time source that is used by DC1 (but not DC2) rolls forward to calendar year 2039. This causes DC2 to also use a system time in CY2039. This causes DC1 to prematurely delete objects that are deleted today from its deleted objects container. Meanwhile, DC2 originates changes to attributes on users, computers, and groups that are live on DC2 but deleted and now prematurely-garbage collected on DC1. DC1 will log error 8606 when it next inbound-replicates changes for the premature deleted objects.
Cause 4: An object is reanimated at the cusp of TSL expirationThe CONTOSO.COM domain contains two domain controllers in the same domain. Tombstone lifetime = 60 days. Strict replication is enabled on both domain controllers. DC1 and DC2 replicate every 24 hours. DC1 originates deletes daily. An OU that contains users, computers, and groups is accidentally deleted. A system state backup that is made at the cusp of TSL in the past is auth-restored on DC2. The backup contains objects that are live on DC2 but already deleted and reclaimed by garbage collection on DC1.
Cause 5: A USN bubble is triggered the logging of the 8606Say that you create an object in a USN bubble in such a way that it does not outbound-replicate because the destination domain controller "thinks" it has the object because of the bubble. Now, after the bubble closes, and after changes start to replicate again, a change is created for that object on the source domain controller and is displayed as a lingering object to the destination domain controller. The destination controller logs event 8606.
Requirements for end-to-end replicate knowledge of originating deletesActive Directory domain controllers support multi-master replication where any domain controller (that holds a writable partition) can originate a create, change, or delete of an object or attribute (value). Knowledge of object / attribute deletes are persisted by the originating domain controller and any domain controller that has incoming replicated knowledge of an originating delete for TSL number of days. (See Microsoft Knowledge Base articles 216996
Active Directory requires end-to-end replication from all partition holders to transitively replicate all originating deletes for all directory partitions to all partition holders. Failure to inbound-replicate a directory partition in a rolling TSL numbers of days results in lingering objects. A lingering object is an object that was intentionally deleted by at least one domain controller but that incorrectly exists on destination domain controllers that did not inbound-replicate the transient knowledge of all unique deletions.
(http://go.microsoft.com/fwlink/?LinkId=151500)for other considerations.
Article ID: 2028495 - Last Review: October 17, 2013 - Revision: 21.0