Fix it now
Microsoft’s published cause is short: WSUS has downloaded content it cannot use because of a missing decryption key. The key is missing on the server, not on the client, so every client offered that feature update fails identically. The fix is on the WSUS server and Microsoft publishes a specific sequence for it.
$s = Get-WsusServer
$bad = $s.SearchUpdates("version 1607")
$bad | ForEach-Object { $_.Decline() }
$bad | ForEach-Object { $s.DeleteUpdate($_.Id.UpdateId) }
- Confirm the server is in the state Microsoft describes before changing anything. Against SUSDB, count rows in
tbFilewhereIsEncryptedis 1 andDecryptionKeyis NULL, or where the file name is an .esd for the affected build andIsEncryptedis 0. Anything above zero is the condition. - Confirm the WSUS prerequisite is installed. Microsoft names KB3159706 for WSUS on Windows Server 2012 and 2012 R2, and the whole problem comes from synchronising Upgrades before it was in place.
- Run the decline and delete above, adjusting the search string for the build and language you actually synchronised.
- Remove the orphaned rows from SUSDB so the content files are no longer referenced, then re-synchronise.
- On one client, delete the contents of
%windir%\SoftwareDistribution\DataStore, restart the Windows Update service and rescan before you touch anything else.
Declining is what allows the content to be removed; unapproving alone leaves the file in place and the client keeps being offered it.
If a pilot client now downloads the feature update, the server is repaired. If clients still fail, the next section separates the server-side codes from the client-side ones.
Why it happens
WSUS distributes feature updates as ESD packages, and those packages arrive encrypted. The server is supposed to hold the key that lets it hand out usable content. 0xC1800118 is Microsoft saying the server has downloaded content it cannot use because that key is missing. Nothing is wrong with the client, the network, or the machine’s licence, and no amount of resetting the update agent changes the answer.
How the server got into that state matters, because it tells you where to look. Microsoft’s remediation describes WSUS on Windows Server 2012 and 2012 R2 where the Upgrades classification was enabled and synchronised before KB3159706 was installed. The server treated the upgrades as ordinary quality updates and stored them without the decryption keys. Once that has happened, installing the prerequisite afterwards does not repair the rows already in the database – they have to be removed.
The companion codes are the noise the clients make while this is going on, and they are not all the same kind of thing. 0x8024200D is the update handler saying the update needs to be downloaded again. 0x8024401F is the client reporting HTTP 500 from the server. 0x80244008 is the client failing to parse a SOAP fault it was sent. 0x80244007 is the broadest of them, and Microsoft publishes several distinct causes for it, none of which are about decryption keys.
The server cached upgrades before it could decrypt them
You have this one if Every client offered the same feature update returns 0xC1800118, while cumulative updates from the same server install normally.
- Run the SUSDB check above and confirm the count is above zero.
- Install KB3159706 on the WSUS server if it is missing, and complete its documented post-installation steps.
- Decline and delete the affected feature updates with the WSUS PowerShell API rather than through the console one at a time.
- Clear the orphaned file rows, re-synchronise, and only then re-approve.
Let the server finish downloading the re-synchronised content before you let clients scan. A client that scans mid-download fails on an incomplete package and you will conclude the fix did not work.
Clients are still holding the bad download
You have this one if The server has been repaired but individual machines keep returning the same code.
- Stop the Windows Update service, rename
C:\Windows\SoftwareDistribution, and start the service again. - Rescan so the client rebuilds its datastore and fetches the corrected content from scratch.
- If a machine still fails, confirm it is talking to the repaired server by reading
WUServerunderHKLM\SOFTWARE\Policies\Microsoft\Windows\WindowsUpdate.
0x80244007: the client exceeded what SyncUpdates accepts
You have this one if WindowsUpdate.log records a SOAP fault naming parameters.InstalledNonLeafUpdateIDs or parameters.OtherCachedUpdateIDs with *FAILED* [80244007] SyncUpdates_WithRecovery.
- On the WSUS server, edit
%programfiles%\Update Services\WebServices\ClientWebService\web.configand changemaxInstalledPrerequisitesfrom 400 to 800. - Decline all superseded updates on the server, which is the other published remedy for the same fault.
- For the
OtherCachedUpdateIDsvariant, increasemaxCachedUpdatein the same file. - Recycle the WSUS application pool and rescan a client.
0x80244007: the client has no authorisation entry
You have this one if The log shows InternalServerError against GetAuthorizationCookie on the SimpleAuthWebService rather than against SyncUpdates.
- On the client, stop the Windows Update service.
- Delete the
PingID,AccountDomainSid,SusClientIdandSusClientIDValidationvalues fromHKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\WindowsUpdate. Errors saying the value was not found are expected. - Start the service and run
wuauclt.exe /resetauthorization /detectnow. - Wait ten to fifteen minutes before judging the result; the client re-registers on its own schedule.
That registry branch is client state, not policy. It is the correct place for these four values and the only place Microsoft documents deleting them.
Full reference
Separating a server fault from a client fault
| What you see | Where the fault is |
|---|---|
| Every client fails on one feature update, quality updates fine | The server’s cached content for that update |
| One client fails, others on the same server succeed | That client’s datastore or its authorisation entry |
| SOAP faults naming SyncUpdates parameters | The WSUS client web service limits |
| HTTP 500 reported as 0x8024401F | Something internal to the server, not the content |
| Clients fail immediately after a re-sync | The server has not finished downloading the content yet |
The database check, in full
Microsoft’s verification step runs against SUSDB and counts rows in tbFile where IsEncrypted is 1 and DecryptionKey is NULL, or where the file name matches the affected build’s .esd and IsEncrypted is 0. A total above zero is the condition; zero means you are looking at something else and should stop before deleting updates.
Doing the removal in the right order
- Confirm the state with the SUSDB count.
- Install the prerequisite update on the server if it is not already present, and reboot it.
- Decline the affected updates. Deletion is not available until they are declined.
- Delete the updates through the WSUS API rather than the console; the API takes the update GUID and removes it cleanly.
- Remove the orphaned file rows so the content files stop being referenced.
- Re-synchronise, wait for the content download to finish, then re-approve.
- Clear one client’s DataStore, rescan, and only then release the approval to the rest of the estate.
Client-side codes you will see alongside it
| Code | What it is telling you |
|---|---|
0x8024200D |
The update handler wants the update downloaded again |
0x8024401F |
The server answered with HTTP 500 |
0x80244008 |
The client could not parse the SOAP fault it was returned |
0x80244007 |
A SOAP fault, with several distinct published causes |
If the feature update still will not go through WSUS
WSUS handles quality updates without any of this complication, and feature updates are the part that goes wrong. Where a small number of machines need the new build now, an in-place upgrade from mounted installation media reaches the same result without the content path that is failing, and does not change the licence on a machine that is already activated. That is a way round the immediate problem, not a repair: the server stays broken until the cached content is removed.
Every code this article covers
| Code | What it points at | Source |
|---|---|---|
0xC1800118 |
WSUS has downloaded content that it cannot use because of a missing decryption key | Microsoft Learn |
0x8024200D |
WU_E_UH_NEEDANOTHERDOWNLOAD: the update handler did not install the update because it needs downloading again | Microsoft Learn |
0x80244007 |
WU_E_PT_SOAPCLIENT_SOAPFAULT: a SOAP fault from the server, with several published causes including the SyncUpdates prerequisite limit | Microsoft Learn |
0x8024401F |
WU_E_PT_HTTP_STATUS_SERVER_ERROR: the same as HTTP 500, an error internal to the server | Microsoft Learn |
0x80244008 |
WU_E_PT_SOAPCLIENT_PARSEFAULT: the SOAP client failed to parse a SOAP fault | Microsoft Learn |
Confirm the fix worked
- The SUSDB count that identified the condition now returns zero.
- The affected feature update no longer appears in the console as approved with cached content.
- After the re-sync, the content folder has grown by roughly the size of the package before any client is allowed to scan.
- A pilot client downloads and starts the upgrade rather than failing at the same point.
- A second client that had already failed once succeeds after its SoftwareDistribution folder is rebuilt.
Questions people ask about this
Does fixing this cost anything?
No. It is server maintenance: a prerequisite update, a database check, a decline and delete, and a re-sync. No licence, key or edition is involved anywhere in it.
Why decline rather than just unapprove?
Because deletion and content cleanup act on declined updates. Unapproving stops the offer but leaves the unusable file in place, so the condition is still there when someone re-approves months later.
Is WSUS still a sensible way to deliver feature updates?
It works, but this failure mode is specific to the Upgrades classification and to servers that synchronised it without the prerequisite. If you deliver feature updates rarely, mounted media or your deployment tooling avoids the content path entirely; if you deliver them at scale, keep the server patched and the classification prerequisites in place.
Do I need to touch the clients at all?
Yes, the ones that already failed. They cached the unusable package. Renaming SoftwareDistribution and rescanning is enough; machines that never got as far as downloading it need nothing.
What if the SUSDB count comes back as zero?
Then this is not your problem and deleting updates would be destructive for nothing. Look at the companion codes instead – 0x80244007 in particular has its own published causes that have nothing to do with decryption keys.
