Systems not processing maintenance windows

[Shrumpsed]

We had a major SCCM issue a few weeks back caused by a hardware failure leading to database replication failures. This was "resolved" by manually syncing and rebuilding a few of the secondary site servers but since then we're having one major issue that we cannot track down or even have anything that points to what the cause could be. Not a single PC is receiving their collection based maintenance windows. They only seem to be getting their "business hours" window which is defined by the times in the software center.

What we have tried:

• Creating new collection and applying the maintenance window times
  
• Verified maintenance windows are actually set on the collection via powershell
  
• Resetting the times for the maintenance window on the already existing collection
  
• Deleting systems from SCCM and having them be re-detected and repopulate in SCCM
  
• Uninstalling/Reinstalling SCCM client on workstation
  
• Running triggers
  
• Checked as many logs on the clients and server as I could think of
  
• Reset machine policy
  
• Reimaging computer
  

None of these have had any changes or given me any indication on what is failing. If I run the report "Maintenance windows available to a specified client" it shows the correct information, but the PCs themselves don't seem to be receiving the information for what the system should be seeing. It doesn't matter which primary/secondary site server, all clients seem to be affected. The only thing I see in the policy for CCM_ServiceWindow have ServiceWindowType = 6, none that are ServiceWindowType = 1 which is what they should be coming from the collection settings.

It just seems this all goes back to database replication issues, I just don't know what it could be. Once the replication link issues were fixed I had to redeploy all DCM baselines as well as the Client policy for enabling SCEP. So I created a clone of our default client policy and deployed that to a newly created collection with a custom maintenance window, but still nothing on the workstations. Since you don't actually deploy the default client settings I'm not sure if those are being applied properly.

I'm missing something, I just don't know what it is. I don't even really know if maintenance windows not applying is the issue or a symptom of a different issue that if that is fixed the maintenance windows will start working. But that's just an idea, I have no way to really prove that.

We're running SCCM 2012 R2 version 5.00.8239.1000 with a CAS, 3 Primary sites, and 17 Secondary sites. We were previously running version 5.00.7958.1000 but updated earlier this year. This environment had been functioning (more or less) properly for over a year prior to the database replication issues.

Any advice would be greatly appreciated.

I'm sure the advice will be to contact Microsoft, and I'd love to but our company doesn't really let us do that as Microsoft support is handled by the server team who doesn't actually have much knowledge about using SCCM itself so if that's what needs to be done I'll figure that when it happens but I'd like to be able to fix it without them.

Thank you,

[Shrumpsed]

If anybody ever has this problem I got off with Microsoft and after a few hours we got it working.

 

Run the following query against the Primary servers SQL database:

select top (1) rowversion as rowversion from CEP_CollectionExtendedProperties order by rowversion

 

You should get a value showing the version of the collection according to the database. Compare this to the following registry key on the Primary server itself:

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\SMS\Components\SMS_COLLECTION_EVALUATOR\Last Row Version

 

In our case the SQL record was 0x0000000000000EF9 and the registry was 0x000000002416D1CA. MS had me change the value to 0x0000000000000EF8, then restart SMS_EXECUTIVE and monitor the colleval.log on the server. It reprocessed all collections and deployments, then within a few minutes running a machine evaluation cycle showed maintenance windows on the workstations as expected.

 

This had to be done on each primary server.

[anyweb]

thanks for posting the solution, did Microsoft explain why this happened and why you had to do the query ?