I'd just finished upgrading my Config Mgr lab to SP1, and found that as soon as I'd restarted my machine, SMS_EXECUTIVE started to fail about 30 seconds or so after starting.  I didn't get a lot of information to know this was happening, apart from the usual "SMSexec has crashed, do you want to send an error report". 

Putting on my troubleshooting hat, I had a look at the CrashDumps directory under the default logs directory.  This is where a dump of various logs and bits of the Config Mgr state get dropped should a component fail.  The interesting log in this case was crash.log which is a summary of the failure, and the various thread states of each of the components.  I had an error which looked like:

EXCEPTION INFORMATION

Time = 06/05/2008 15:21:29.358
Service name = SMS_EXECUTIVE
Thread name = SMS_HIERARCHY_MANAGER
Executable = D:\Program Files\Microsoft Configuration Manager\bin\i386\smsexec.exe
Process ID = 3548 (0xddc)
Thread ID = 4256 (0x10a0)
Instruction address = 78141e3a
Exception = c0000005 (EXCEPTION_ACCESS_VIOLATION)
Description = "The thread tried to read from the virtual address 00000080 for which it does not have the appropriate access."
Raised inside CService mutex = No
CService mutex description = ""

The key message in this case is the description, which points to a virtual memory address (00000080).  To find this, I searched the crash.log for that error, and found this:

STACK TRACE FOR SMS_HIERARCHY_MANAGER THREAD 4256 (0x10a0) AT 06/05/2008 15:21:29.358

EAX: 00000080  CS: 001b  EIP: 00000000  EFLAGS: 00010202
EBX: 2ec8a0fe  SS: 0023  ESP: 013df35c
ECX: 7ffffffe  DS: 0023  EBP: ffffffff
EDX: 00000073  ES: 0023
ESI: 00000000  FS: 003b
EDI: 00000080  GS: 0000

This seemed to indicate that the issue was happening in Hierarchy Manager.  Opening hman.log  the last lines were:

Update the Sites table: Site=XXX Parent=~  $$<SMS_HIERARCHY_MANAGER><Thu Jun 05 15:21:29.137 2008 New Zealand Standard Time><thread=4256 (0x10A0)>
Nothing has changed for the boundary  $$<SMS_HIERARCHY_MANAGER><Thu Jun 05 15:21:29.268 2008 New Zealand Standard Time><thread=4256 (0x10A0)>
 No profile in DB, will try to create first version of AMT profile  $$<SMS_HIERARCHY_MANAGER><Thu Jun 05 15:21:29.298 2008 New Zealand Standard Time><thread=4256 (0x10A0)>
Trying Update Amt profile, new version will be created  $$<SMS_HIERARCHY_MANAGER><Thu Jun 05 15:21:29.338 2008 New Zealand Standard Time><thread=4256 (0x10A0)>

This seemed a little strange, as there should have been more information in there after that.  It didn't indicate why the Hierarchy Manager would have failed, just that it had suddenly stopped.  On a hunch I went looking in the hman.box directory (<Config Mgr install path>\Inboxes\hman.box) and found a bunch of CT2 files in there.  In a steady state you wouldn't expect to see any files in there - they should get processed and moved on.  I decided the best thing I could do would be to move the files out (but keeping them in case they were important) and try restarting the SMS_EXECUTIVE service.  As soon as I did this, all sorts of activity started to take place.  All the components that should have been reinstalled as part of the SP1 upgrade were reinstalled, and the system eventually started functioning as normal.

 Lessons learned:

• All the information you need to troubleshoot is there in the logs
• It's not necessarily that obvious what log you need to look at
• Follow your hunches

 

When you're implementing Windows Server 2008 RODC's there are a couple of compatability issues that you may get struck with on Windows XP or Windows Server 2003 clients.  This KB article details the various issues that you may strike, and provides downloads for the compatability pack.

 There are ten issues which you might come across, a brief summary of each is below:

• If a client can access only read-only domain controllers, Windows Management Instrumentation (WMI) filters that are configured for Group Policy are not applied.
• Internet Protocol security (IPsec) policies cannot be applied, and Win32 error code 8219 (ERROR_POLICY_OBJECT_NOT_FOUND) is returned when only Windows Server 2008 read-only domain controllers are available.
• Windows Server 2003 member computers and Windows XP member computers do not synchronize Win32 time with Windows Server 2008 read-only domain controllers.
• Computers in a perimeter network cannot join the domain.
• In a site that has only read-only domain controllers available, users try to change their passwords on computers that are running Windows 2000, Windows XP, or Windows Server 2003. When the users do this, the password change operation fails.
• Windows Server 2008 read-only domain controllers cannot retrieve or create the public key certificate by using the LsaRetrievePrivateData function or the LsaStorePrivateData function.
• When you try to publish a printer, the published printer may not work correctly.
• In a site that has only read-only domain controllers available, you use the Find Printer dialog box on a client computer that is running Windows 2000, Windows XP, or Windows Server 2003. When you do this, the Find Printer dialog box stops responding.
• Active Directory Service Interfaces (ADSI) API functions in Windows Server 2003 and in Windows XP always send requests to a remote writable domain controller instead of to a local read-only domain controller.
• Domain controllers that are running Windows Server 2003 perform automatic site coverage for sites that have read-only domain controllers.
  

The patch for System Center Virtual Machine Manager (VMM) 2008 that allows it to manage Hyper-V RC1 has been released to the http://connect.microsoft.com site.  To install this patch, the VMM machine and the Hyper-V hosts must all be running the RC1 release of Hyper-V, and after you've installed it you won't be able to manage any RC0 Hyper-V machines.

Just a bunch of things that caught my eye this morning:

Hyper-V Failover Clustering Options: http://blogs.technet.com/josebda/archive/2008/06/17/windows-server-2008-hyper-v-failover-clustering-options.aspx

Microsoft App-V (SoftGrid) 4.5 RC Available: http://blogs.technet.com/virtualworld/archive/2008/06/17/microsoft-application-virtualization-4-5-release-candidate-is-now-available.aspx

Update released for Config Manager to fix issue with June Security Updates: http://support.microsoft.com/?kbid=954474

To Dep or not to DEP: http://blogs.technet.com/askperf/archive/2008/06/17/to-dep-or-not-to-dep.aspx (in fact the whole Ask The Performance blog is worth reading every time they post)

Performance tuning guidelines for Windows Server 2008 now include Hyper-V & Power: http://www.microsoft.com/whdc/system/sysperf/Perf_tun_srv.mspx

All topics performance - a great blog discussing various performance bits and bobs: http://blogs.msdn.com/tvoellm/default.aspx, if you want just the Hyper-V performance stuff, use the tags: http://blogs.msdn.com/tvoellm/archive/tags/Hyper-V/default.aspx

I was just talking to one of my colleagues about the difference between Revert & Apply when using the Hyper-V management console.  They are almost the same thing, just that Revert is a special case of Apply.  Apply is used on any snapshot that you want to start using - you right click the snapshot and choose Apply.  This then takes the virtual machine to the state it was in at the time the snapshot was taken.  Revert is a special case, and takes you back to the last snapshot that you took or applied.  This is the one that is shown with a green triangle in the Snapshots pane.

 For more information about Snapshots in Hyper-V I recommend Ben Armstrong's blog post: http://blogs.msdn.com/virtual_pc_guy/archive/2008/01/16/managing-snapshots-with-hyper-v.aspx