Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

Event 2042 / error 8614 "It has been too long since this machine replicated": rejoin or remove the domain controller

Technical article
Troubleshooting

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.

Event 2042 and error 8614 are a safety stop, not a fault in themselves. A domain controller has refused to accept changes from a partner because the two have not replicated for longer than the forest's tombstone lifetime. Letting them replicate straight away could bring deleted accounts and groups back to life.

The fix is a decision. Either remove the out-of-date domain controller and rebuild it, or check for and remove lingering objects, briefly lift the block, let replication catch up and then put the block back. Each has a cost. Removing the server is faster, but any changes made on it that never reached another domain controller are lost; Microsoft's 8614 page says forced demotion "can result in the loss of originating updates (that is, data loss)". Cleaning and resuming keeps those changes, but only if lingering objects are removed first. Lifting the block before that cleanup risks bringing deleted objects back.

What you see

  • The Directory Service log on the destination domain controller records event 2042: "It has been too long since this machine last replicated with the named source machine. The time between replications with this source has exceeded the tombstone lifetime. Replication has been stopped with this source." The event names the source, the time of the last successful replication and the tombstone lifetime in days (Microsoft Learn, event 2042).
  • repadmin /showrepl and repadmin /replsum show 8614 (0x21a6): "Active Directory cannot replicate with this server because the time since the last replication with this server has exceeded the tombstone lifetime."
  • dcdiag fails the Replications test with the same status, and Replicate now in Active Directory Sites and Services returns the same text.

Why it happens

When an object is deleted, Active Directory keeps a small "tombstone" record of the deletion and replicates it to every domain controller. After the tombstone lifetime the record is removed. A domain controller that missed the whole lifetime never learns about the deletion, so it still holds the object. That leftover copy is a lingering object (Microsoft Learn, lingering objects).

Microsoft lists the root causes of 8614 and 2042 as (Microsoft Learn, error 8614):

  1. A real replication failure that lasted longer than the tombstone lifetime. The partner was offline, cut off by a network or firewall change, or failing with another error (often 1722) that nobody noticed.
  2. A clock jump. If the destination's system time moved forward or back by the tombstone lifetime or more, it looks to the replication engine as if the partner has been silent that long.

Typical real-world triggers named by Microsoft include a domain controller removed from the network and stored, a pre-staged domain controller that took too long to reach its site, a long-running WAN outage, an administrator shortening the tombstone lifetime, and a clock moved forward or back.

How long is the tombstone lifetime? For forests created on currently supported Windows Server versions the default is 180 days. A forest first built on an older version may still use 60 days, because upgrading domain controllers does not change the stored value (Microsoft Learn, lingering objects). It can also have been set much shorter by hand. The 2042 event text states the value your forest uses.

How to prove the cause

These checks are read-only.

1. Read the tombstone lifetime the forest actually uses.

repadmin /showattr <DC name> "CN=Directory Service,CN=Windows NT,CN=Services,CN=Configuration,DC=<forest root domain>,DC=<top level domain>"

If the tombstoneLifetime attribute is absent from the output, an internal default applies.

2. List every partner's last success, oldest first.

repadmin /showrepl * /csv > showrepl.csv

Open the file in Excel and sort the "Last Success" column ascending. Microsoft recommends this as a simple daily check too. Anything near or past the tombstone lifetime is the problem pair. Note that once 8614 is logged, older errors for that partner from within the tombstone lifetime are masked, so look at past logs and other partners to find what originally broke replication.

3. Look for a clock jump. In repadmin /showreps output, dcdiag logs and event logs on the destination, look for dates before the operating system existed or was installed, dates in the future, or gaps with no events at all. Microsoft notes support teams have seen production domain controller clocks jump hours, days, years and "even tens of years".

4. Check for lingering objects without removing anything. On a domain controller you believe holds the latest changes:

repadmin /removelingeringobjects <destination DC FQDN> <source DC DSA object GUID> <partition DN> /advisory_mode

Get the DSA object GUID from repadmin /showrepl <authoritative DC>. Advisory mode only reports. Microsoft's preferred tool for this is Lingering Object Liquidator v2 (LoLv2), a separate download, with Repadmin as the fallback (Microsoft Learn, event 2042).

The fix

Choose the route

Microsoft describes two action plans (Microsoft Learn, error 8614):

  • Remove the out-of-date domain controller. Force-demote it, then perform metadata cleanup from a healthy domain controller in the same domain, and promote a fresh server if you still need one. Microsoft notes this may be easier and faster. It is usually the right choice when the stale domain controller holds nothing unique and other domain controllers in the domain are healthy. Metadata cleanup is a required step after any forced removal; with current tools, deleting the domain controller's computer object in Active Directory Users and Computers performs it automatically, and ntdsutil provides a command-line route (Microsoft Learn, metadata cleanup).
  • Clean and resume. Use this when the domain controller must stay, for example when it is the only one in a site or there are too few healthy domain controllers to rebuild from.

Clean and resume, in order

  1. Fix the clock first if a time jump caused it, and find out why the time moved. Microsoft's list of things to check includes whether the forest root PDC emulator uses a reliable external source, whether domain controllers take time from the domain hierarchy, and whether virtual machines take time according to the hypervisor vendor's guidance.

  2. Remove lingering objects from each partition the destination holds, using LoLv2 or repadmin /removelingeringobjects without /advisory_mode, from the domain controller you chose as authoritative. Repeat for every domain controller that has them.

  3. Resolve the original replication failure (network, DNS, authentication, database or topology). Otherwise the quarantine will simply return.

  4. Lift the block temporarily on the destination:

    repadmin /regkey <destination DC FQDN> +allowDivergent
    

    This sets "Allow Replication With Divergent and Corrupt Partner" under HKLM\SYSTEM\CurrentControlSet\Services\NTDS\Parameters.

  5. Let it replicate once, confirm success, then put the block back:

    repadmin /regkey <destination DC FQDN> -allowDivergent
    

Microsoft is explicit that replication should be restarted "only after you have removed all lingering objects", and that the key should be removed once the systems have replicated to reinstate the protection.

How to confirm the fix

  • repadmin /replsum shows no failures for the pair, and repadmin /showrepl shows a fresh last success for each partition.
  • repadmin /removelingeringobjects … /advisory_mode finds nothing further.
  • allowDivergent has been reversed on every domain controller where you set it.
  • Strict replication consistency is enabled on the domain controllers. It is on by default in forests created on Windows Server 2003 or later; Microsoft documents repadmin /regkey * +strict to enable it everywhere.

What to watch afterwards

  • Daily replication check. Microsoft recommends monitoring end-to-end replication every day and putting domain controllers on a watch list when failures reach 50 per cent of the tombstone lifetime, with a strong push to fix them then, and considering forced removal at 90 per cent. With a 180-day lifetime, that is 90 days and 162 days.
  • Time source. A domain controller restored from an old backup or snapshot, or a virtual machine whose clock is set by the host, can trigger the same stop. Keep domain controller time configuration documented and checked after any move or restore.
  • Stored or offline domain controllers. Do not keep a domain controller switched off "as a spare" for months. It will come back with this error.

Risks and limits

  • Do not set allowDivergent or the registry value to 1 just to make the error go away. If lingering objects remain, deleted users and groups can reappear, and a group that no longer exists can stay in users' access tokens.
  • dcpromo /forceremoval, mentioned in Microsoft's 8614 article, is the legacy command. On current Windows Server versions forced demotion is the "Force the removal of this domain controller" option in Server Manager, or Uninstall-ADDSDomainController -ForceRemoval in PowerShell. Microsoft warns that any changes on that domain controller that had not replicated, "such as passwords or new user accounts, are lost forever", that it leaves orphaned metadata, and that you "must manually perform metadata cleanup immediately". If the server held operations master roles, seize them on another domain controller straight after (Microsoft Learn, demote domain controllers).
  • Removing lingering objects needs Domain Admins rights or equivalent and the right choice of authoritative domain controller. If you are unsure which domain controller holds the latest data, repeat advisory mode on several before removing anything.
  • An 8614 on one domain controller does not mean the fault is on that server. The cause may be the source, the destination or the network between them.

Sources

Next step

If a domain controller has hit the tombstone limit and you need to decide whether to rebuild it or clean it, or you want replication monitored so this never reaches 50 per cent of the lifetime again, NetLeverage has looked after business servers since 2002. Talk to us through managed support and bring your repadmin /showrepl * /csv output and the text of the 2042 event.