Fix it now
Event ID 7016 is one client-side extension reporting that it finished in an error state, not Group Policy failing as a whole. The entry names the extension and how long it ran, so read that name and repair what that extension was pointed at.
gpupdate /target:computer /force
gpresult /h %userprofile%\Desktop\gpreport.html /f
reg query "HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon\GPExtensions" /s
- Open Event Viewer and expand Applications and Services Logs, Microsoft, Windows, GroupPolicy, Operational.
- Filter for Event ID 7016 and open the newest entry. The description names the extension and the milliseconds it spent before failing.
- On the Details tab, copy the ActivityID. Every event from that one processing cycle carries it, so filtering on it replays the run in order.
- Compare against Event ID 5016, the same completion record written when an extension succeeds. If everything else logged 5016 and one logged 7016, that is your extension.
- Repair what that extension was pointed at: the share it reads, the account it references, or the package it installs. Not SYSVOL, and not replication.
The ActivityID changes on every refresh, so re-read it after each gpupdate rather than reusing the one you copied.
If that fixed it you can stop here. If not, the next section explains why one extension can fail while the rest of the cycle succeeds.
Why it happens
The Group Policy service does the shared work once per cycle: find a domain controller, build the filtered list of policy objects, read the policy files. It does not itself know how to map a drive, install software or grant a user right. Each of those belongs to a client-side extension, a DLL registered under HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon\GPExtensions, one subkey per extension, named by GUID.
So the operational log reads in pairs. Event ID 4016 records an extension starting. Event ID 5016 records it completing, with the elapsed time in milliseconds. Event ID 7016 is the error counterpart: the extension reached the end of its work in a failed state. Read the pair and you know which extension failed and how long it spent failing.
The companion identifiers answer different questions from other logs. Event ID 1085 goes to the System log when settings owned by a named extension could not be applied. Event ID 1202 comes from SceCli in the Application log. Event ID 1091 is the one most often misread: the settings did apply, and only the Resultant Set of Policy record failed to be written, so the policy is in force and the reporting is what is wrong.
The security extension cannot resolve an account named in policy
You have this one if 7016 names the security settings extension and an Event ID 1202 appears in the Application log in the same minute, usually with status 0x534 or 0x2.
- Turn on the security extension’s own logging, described below, then refresh computer policy.
- Read out the names it could not resolve:
find /i "cannot find" %SystemRoot%\security\logs\winlogon.log - Re-add the account by browsing to it rather than typing it, usually in User Rights Assignment or Restricted Groups.
An unreachable trusted domain looks identical to a deleted account, so check the trust first.
The extension is waiting on a network path that never answers
You have this one if the elapsed time dwarfs every other extension, and the one named reaches out: folder redirection, software installation, scripts or preferences.
- Take the UNC path out of the setting and test it from the affected client, not from your own workstation.
- Computer-side settings run as the machine account, so permissions must grant read access to the computer objects, not only to users.
- If the path is a namespace, test the target behind it. One pointing at a dead target fails slowly.
A third-party extension is registered but broken
You have this one if the extension named in the entry is not a Microsoft one, or it shows only as a GUID with no friendly name.
- Run the reg query from the fix section and find the subkey matching the GUID in the event.
- Its default value gives the friendly name and DllName gives the file. Confirm that file is there.
- Reinstall or fully remove the product that owns it rather than deleting the registration by hand.
Foreground processing runs before the network is ready
You have this one if 7016 appears at every boot or logon but never on a manual refresh, and it is worse on wireless and VPN-connected machines.
- Enable Computer Configuration, Administrative Templates, System, Logon, Always wait for the network at computer startup and logon, which turns off fast logon optimisation.
- Where the switch port authenticates after the operating system starts, review the 802.1X configuration.
- For laptops that reach the domain only over VPN, use a client that connects before logon.
Full reference
Reading one processing cycle end to end
The operational log is Microsoft-Windows-GroupPolicy/Operational, and every event from a single processing instance shares an ActivityID. Take it from the Details tab, under the System node, and filter the log on it. That turns a log full of interleaved cycles into one readable run: the service starting, each extension’s 4016, and each extension’s 5016 or 7016 with its elapsed time. Refreshing policy assigns a new ActivityID, so the correlation identifier from your last run will not match the next one.
Two readings look like failures and are not. An extension that fails on a laptop off the network and succeeds on the next cycle is behaving correctly. And the audit extension can complete with status 2147483658, which is E_PENDING, meaning it started an asynchronous thread successfully; that is expected rather than a fault.
Matching what the log shows to where the fault is
| What the log shows | Where the fault is |
|---|---|
| 7016 for one extension, 5016 for the rest | That extension’s payload: a path, an account or a package |
| 5016, but an elapsed time far above the others | It is reaching something that answers slowly or not at all |
| 7016 naming an extension you do not recognise | A third-party extension whose DLL is broken or gone |
| 7016 at boot but never on a manual refresh | The network is not ready when foreground processing runs |
| 1091 for an extension that otherwise logged 5016 | Policy applied; only the reporting record failed |
Turning on the security extension’s own log
The security configuration engine writes a plain text log, but not by default. It appears once you set a debug level on the security extension’s own registry key and refresh computer policy. The GUID below is the security extension; the log is written to %SystemRoot%\Security\Logs\Winlogon.log.
reg add "HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon\GPExtensions\{827D319E-6EAC-11D2-A4EA-00C04F79F83A}" /v ExtensionDebugLevel /t REG_DWORD /d 2 /f
gpupdate /target:computer /force
find /i "cannot find" %SystemRoot%\security\logs\winlogon.log
Set the value back to 0 when you have finished. Leave debug logging on and the file grows on every refresh.
Status codes SceCli reports with 1202
| Status | Meaning | Usual explanation |
|---|---|---|
| 0x534 | No mapping between account names and security identifiers was done | A name in policy that no longer resolves |
| 0x2 | The system cannot find the file specified | A path or object named in policy that is not there |
| 0x5 | Access denied | The engine could not apply a setting it was asked to apply |
| 0x6fc | The trust relationship failed | The account lives across a trust that is currently broken |
| 0x4b8 | An extended error has occurred | Read the log rather than the status; it is a catch-all |
What the extension registration actually holds
Each subkey under GPExtensions is named for the extension’s GUID and describes how the service should call it. DllName gives the file, and ProcessGroupPolicyEx names the entry point. The rest are behaviour switches the extension’s author set: NoBackgroundPolicy, NoSlowLink, NoMachinePolicy, NoUserPolicy, EnableAsynchronousProcessing and NoGPOListChanges among them. Read them to understand why an extension runs when it does. They are the extension author’s declarations, not administrative tuning, so treat the key as evidence rather than as something to edit.
When none of the above fits
- Check whether the same extension logs 7016 on a second client in the same organisational unit. That separates a machine problem from a policy problem.
- Read the 1085 entry’s event data rather than its message. ErrorCode and ErrorDescription are usually more specific than the General tab.
- Compare a working and a failing client’s registered extensions. A GUID on one and not the other names the product to repair.
Every code this article covers
| Code | What it points at | Source |
|---|---|---|
Event ID 7016 |
GroupPolicy Operational log: a client-side extension completed processing in an error state, naming the extension and the time it ran | Microsoft Learn |
Event ID 1085 |
System log: settings owned by the named extension could not be applied, with the extension name, GUID and status in the event data | Microsoft Learn |
Event ID 5016 |
GroupPolicy Operational log: the success counterpart of 7016, carrying the milliseconds that extension took; 4016 is the matching start event | Microsoft Learn |
Event ID 1091 |
System log: the settings did apply, but the Resultant Set of Policy record could not be written, so reporting tools may be inaccurate | Microsoft Learn |
Event ID 1202 |
Application log, SceCli: security policy was propagated with a warning, carrying its own status value | Microsoft Learn |
Confirm the fix worked
- Refresh policy and confirm the operational log now shows 4016 followed by 5016 for the extension that was failing.
- Restart the client and confirm the same pair appears during foreground processing, not only on a manual refresh.
- Check the System log and confirm no new 1085 entries are being written for that extension.
- Generate a report with gpresult /h and confirm the settings that extension owns appear as applied.
- If you enabled security extension logging, set ExtensionDebugLevel back to 0.
Questions people ask about this
Does fixing this cost anything?
No. Every step here is diagnosis and configuration on machines you already own and licence. A client-side extension failure has nothing to buy attached to it.
Is 7016 always serious?
No. An extension that fails on a machine that was off the network and succeeds on the next cycle is behaving as designed. What matters is the same extension failing every cycle on clients that are reliably connected.
Why does gpresult say the policy applied when the event log says it failed?
Two different questions, and sometimes a third answer. gpresult reports which policy objects were in scope, which is not the same as whether every extension inside them succeeded. And if you are also seeing Event ID 1091, the reporting data itself failed to be recorded, so gpresult is working from an incomplete picture.
Can I stop one slow extension holding up logon?
Turning off the wait for the network makes logon return sooner, but it does that by applying policy asynchronously, so the extension still fails and now fails later. Treat it as a way to reduce the pain while you fix the payload, not as the fix.
The event names a GUID and no friendly name. What now?
Look the GUID up as a subkey under GPExtensions. Its default value is the friendly name and its DllName value is the file, which together tell you which product registered it.
