extension-dll-exception

Cause

Figured out that the error is happening due to .Net Framework 4 being installed, and FIM attempting to use .Net 4 instead of .Net Framework 2. This is a new issue that occurred in the June 2011 CU for Sharepoint 2010, and unbeknownst to us, the June CU was re-released to fix it. We were still utilizing the original June CU.

Resolution

This issue can be fixed by either installing the latest version of June CU, or by following the steps below.

1. Open

   C:\Program Files\Microsoft Office Servers\14.0\Synchronization Service\Bin\miiserver.exe.config

   for editing.
2. Locate the below section:
   
   

   <startup useLegacyV2RuntimeActivationPolicy="true">
     <supportedRuntime version="v4.0.30319"></supportedRuntime>
     <supportedRuntime version="v2.0.50727"></supportedRuntime>
   </startup>

3. Delete or comment out the reference for the .NET v4 version. Like this:
   
   

   <startup useLegacyV2RuntimeActivationPolicy="true">
     <!-- <supportedRuntime version="v4.0.30319"></supportedRuntime> -–>
     <supportedRuntime version="v2.0.50727"></supportedRuntime>
   </startup>

   or
   
   

   <startup useLegacyV2RuntimeActivationPolicy="true">
       <supportedRuntime version="v2.0.50727"></supportedRuntime> 
   </startup>

4. Restart the two FIM services in the services console.
5. Run the sync again.

SMS_MP_Control_Manager errors:

Navigate to %windir%\\System32\\inetsrv\\config\\schema, take ownership of webdav_schema.xml,  remove the readonly attributes, and edit them to these:

attribute name=”allowAnonymousPropfind” type=”bool” defaultValue=”true”
attribute name=”allowInfinitePropfindDepth” type=”bool” defaultValue=”true”
attribute name=”allowCustomProperties” type=”bool” defaultValue=”false”

If issues, still exist, and your authoring rules are correct, try removing WebDAV, rebooting, re-installing WebDAV.

Clients not talking to server:

• Is the client push installing? If not, check WMI is enabled on the client firewalls
• Are site boundaries defined?
• Is the install flag cleared?
• Is a sitecode defined in your AD schema? If so, are the clients picking it up? Sometimes your clients may appear to have a site code assigned (from the SCCM console), double-check on your client machines to be sure they are actually picking it up. If this has occurred, please see the following note.
• Are you using a GPO with the Configuration Manager ADM template to control the site code? If so, be aware that it works by applying a registry key (also considered a preference, so it’s stickied unless you explicitly define a removal policy for it). This registry key defaults to an x86 portion of the registry. The actual key for an x64 machine is located elsewhere, and needs to be defined in the following key:

‘HKLM\\SOFTWARE\\Wow6432Node\\Microsoft\\SMS\\Mobile Client\\AssignedSiteCode’

I recommend pushing this registry key out by GPP in one of your GPOs, and then initiating a re-install of the SCCM client on the machines affected.

WSUS MP issues:

1. Remove WSUS, delete the existing database.
2. Remove WSUS component from SCCM.
3. Reboot server. Re-add the WSUS role. You may have to manually download WSUS with SP2 from Microsoft Download Center if you are encountering errors re-adding the role. Do NOT configure the WSUS role when re-adding.
4. Re-install WSUS component in SCCM.

Forefront Endpoint Protection 2010 installation on SCCM failing:

When installing FEP2010 on SCCM, you may be hit with an error just before installation completion that ‘Setup was unable to create unknown machines. 0×80070003‘, and/or that sms_def.mof couldn’t be updated. This occurs because the FEP2010 setup is looking for the x86 Program Files directory for one of the last steps, and not using the default Program Files directory on an x64 machine.   To resolve the first error, do the following:

1. Manually create this folder structure:  Program Files (x86)\\Microsoft Configuration Manager\\inboxes\\auth\\ddm.box  (give folder same ACL’s as is on existing installed directory)
2. Run R2 setup again. When install completes, copy the two DDR’s from the C:\\Program Files (x86)\\Microsoft Configuration Manager\\inboxes\\auth\\ddm.box and place them it into the “C:\\Program Files\\Microsoft Configuration Manager\\inboxes\\auth\\ddm.box” (where SCCM is actually installed):
3. Once the DDR’s processed the R2 installation should succeed.

To resolve the error about sms_def.mof not updating, do the following:

1. Manually create C:\\Program Files (x86)\\Microsoft Configuration Manager\\inboxes\\auth\\clifiles.src\\hinv
2. Copy C:\\Program Files\\Microsoft Configuration Manager\\inboxes\\auth\\clifiles.src\\hinv\\sms_def.mof to the folder you just created
3. Re-run FEP2010 setup and then copy C:\\Program Files (x86)\\Microsoft Configuration Manager\\inboxes\\auth\\clifiles.src\\hinv\\sms_def.mof back to the original Program Files path.

Next I checked the ISA 2006 event logs. I started seeing a lot information events about packets dropped because of invalid data. These packets were coming from the servers configured in the root hints. It seemed odd, as they were literally filling my event logs. I stumbled across this¹ gem. Windows Server 2008 comes with a new protocol called EDNS turned on by default. These EDNS UDP packets are often well over 512 bytes. ISA 2006 apparently has issues handling these packets. Quickest solution: turn off EDNS.

EDNS can be turned off by the following command from an elevated command prompt: “dnscmd /config /enableednsprobes 0″

Immediately after disabling EDNS, clients had normal DNS lookups again. No more failed page lookups resolved by a refresh. There was one catch though: web browsing was drastically slow. In some cases it appeared an “ipconfig /renew” would fix it briefly, but for the better part of the day web browsing was slow. What was particularly odd was that speed tests would report the performance that we expected from our lines. After some Googling, I found this² Microsoft KB which describes slow web browsing performance between ISA and SBS 2008. While I don’t have SBS in my environment, I thought it might be related. I ran the hotfix, and after it did its thing, it restarted the firewall. I’m not sure whether the hotfix or the firewall restart (or both) was the solution, but after that our browsing performance was speedy again!

In the registry, browse to HKLM\\SOFTWARE\\Microsoft\\Microsoft Data Protection Manager\\Notification\\

Delete the SMTPPassword and SMTPUserName keys

That’s it, anonymous SMTP authentication works now!

The article makes a number of interesting observations, particularly about hardware manufacturers and suppliers. It also makes mention of a number of upcoming hardware designs and the features they will bring. I’d recommend reading the piece, even if just for the factual knowledge it contains.

I’m afraid however, that I have to disagree with the author. In my opinion, I don’t believe that 2010 is the year of the Linux tablet. As a matter of fact, I don’t think that any Linux device will ever out-iPad the iPad. Linux aficionados have been proclaiming the ‘year of Linux’ on desktops for as far back as I can remember. Every year, every distribution, Linux fans rejoice and proclaim that the end is nigh for Microsoft and Apple domination, and that Linux is going to take over the world. Why? Because they believe that ‘free’ will triumph over commercial.

The author summarizes my perspective exactly (though completely unintended) in a later paragraph, where he says ‘In my opinion, Linux won’t offer this kind of magic. To bring magic like Apple does, Linux and the companies bringing Linux have to act like Apple. That means secrecy, a closed development model, a one-size-fits-all solution and one and only one manufacturer / organization which oversees hardware, software, marketing and the developer community. Not going to happen for free software and Linux!’ If we break down this statement meaningfully, we see that that in order to bring an almost perfect user experience, one needs to have a closed development model, and a single manufacturer designing both the hardware AND software. He continually pushes Android, an flavor of Linux being driven by a company (Google) with a lot of money behind it, and a development structure (management) that is not available in a community driven environment.

The author highlights another perspective I share (again, unintentionally I’m sure), when he states that ‘Android is much more popular, and the magic comes from Google, Google marketing Android at big device manufacturers and the platform being ‘pretty open’. Also, in contrary to my limited view, Google understood the issue about marketing and education. That’s why, in advent of the ballot screen, they ran huge adds outside on busstops, along highways and buildings at least in NL and UK and they even bought their own front page add on free newspapers read by millions of Dutch readers.’ It’s funny how advertising is so important even for something free. He realizes that the idealistic nature of Linux (free) is not merely enough to make people want it. Money (and lots of it) must be spent in order for something to be adopted. Nothing in life is free, and unfortunately, the majority of the Linux community seem to miss that fact.

The author concludes with a short(sighted) paragraph, stating that ‘Android on the tablet will be what Windows was on the desktop. But it will be easier for other Linux to co-exist with Android than with Windows, so for those who like to tinker the feature is pretty bright.’ But I thought that Linux by its very nature was supposed to be open, so that it can coexist easily with everything? Therefore, it shouldn’t be easier to code for Linux rather than Windows. Unfortunately, developers tend to code more for the platforms that make them money (but hey, everything in life should be free right? ), and the majority of developers will tend to code for the Windows desktop OS. Also, I think it is very short-sighted to conclude that Microsoft will not be targeting the tablet market as well.

This post may come across a little bit harsh toward Linux. However, the fact remains that without proper guidance/direction in the development of a Linux tablet OS, and proper marketing, Linux will never succeed. For this ‘free’ operating system to succeed, money, and lots of it, must be spent. The only problem is that it is not the general Linux community who will pay for it. Rather, it will be (is) big corporations (see Google) who fork out tons of money. It just seems that the Linux community does not get this fact, or, if they do, do not care about freeloading the money to support their dream of a ‘free’ OS.

Let me start off by laying out what I believe are the three tiers of cloud computing. I believe that cloud computing consists of hosted services (IBM), hosted applications (VMWare, Citrix), and hosted storage/data. This contrasts with internal services (SharePoint, Exchange), internal applications (MS Office, Adobe Acrobat), and internal data (File server). The internal computing may be run on virtualized servers, or on their dedicated boxes. In my opinion, businesses are concerned firstly with security regarding hosted documents/storage layer, secondly web services, and thirdly, most open to hosted web applications.

We are seeing a shift away from applications run on local machines to run on a server and/or virtualized (Citrix, VMWare). These applications tie into the internal services and storage. We are also seeing a shift away from local and/or virtualized applications to web applications. While not all applications can be run as hosted applications (Multimedia applications, lab machines, etc), most of the basic productivity apps are able to be run online or on a VPN server. Web applications (whether hosted or run internally) tie into the company’s services. These services in turn create mountains of data to be stored. Where does the data get stored? Should companies use hosted storage or should they use internal storage? If they want internal storage, they are going to have to create a framework infrastructure to dump this data internally, whether it be from web services or from hosted/web applications. There is no competing with the onslaught of hosted applications. Some are transitioning to web applications, but most (if not all) will be transitioning from local to being hosted.

Here’s a brief synopsis of what companies are offering what (in my model of a three tier system):

• Vmware vCloud – virtualized hosted apps, tie into internal infrastructure, thinapps
• Microsoft Azure- hosted services, storage, web applications
• Cisco VN- virtualized apps, network
• Citrix – hosted apps
• IBM – hosted services

Let’s move to the next tier, the web services layer. This layer provides services such as email or forums (Microsoft Exchange Online, Drupal, Joomla, IBM services). I would argue that it also includes web serving services as well. The problem with this layer is that this is where more and more malicious apps are being targeted. Microsoft recently did a study that showed that malicious apps are targeted more and more at the browser or specific apps. They target the authentication component of the web services. If compromised, an attacker can use the information gathered to gain access to the data contained (emails, documents in the document management system). However, with enough security measures, this risk can be mitigated for the most part. 

The other negative to having hosted web services is that your emails and forums are being hosted on someone else’s server. How much this affects your organization is something you have to decide. 

The benefit of hosted services is that the cost to run them is miniscule. Organizations no longer need to employ internal staff to maintain these, and pay a lot less on implementation costs. It also provides arguably better failover services. The benefits are very attractive for organizations. I would argue that competing with hosted web services is a waste of time for the most part.

The third layer is the data storage layer. This is where all your documents are stored. This would include Microsoft SharePoint Online, Live Mesh, or other web storge services. This is critical. Are you willing to have your documents hosted elsewhere? Do you have thousands of employee SSN’s or financial data stored? This is a weighty matter to consider. What about if somehow an employee at the company hosting your data manages to break in. How big is this risk? Are you willing to take it? 

Here is where I see things going:
Government & Financial institutions will likely graduate to hosted applications or web apps, while moving to virtualized services and internal data. VMWare/Cisco will be the big players. Security is paramount, so services will all be virtualized internally.

Medium to Large organizations will move to hosted services, hosted AND virtualized applications, while maintaining data internally. Big players include Microsoft (Azure) and VMWare. Cisco will also play a fairly large role. The employees will be working from home/remotely more frequently, as well as have intermittent (flights, dialup, etc) internet outages, so offline virtualization is more important.

Small to Medium organizations will graduate to hosted or virtualized applications, hosted services, and hosted data. Big player: Microsoft. The cost outweighs the benefit, so being heavily internet based will be more attractive than maintaining ability to work offline.

Microsoft obviously is not going to take the iPad challenge lying down, and will be touting Windows 7 as a more complete experience (Flash anyone?) than the iPad OS. Unfortunately however, I don’t see Windows 7, good as it may be, to be a worthwhile competitor to the iPad, due to the reason outlined above. So what should Microsoft do?

I firmly believe that if Microsoft wants to take on the iPad, it needs to build a tablet from the ground up. They’ve got some great software and hardware engineers¹, and should leverage that talent to build a complete Microsoft tablet experience. Integrate it with Zune and Xbox. They’ve got the Zune app store and Xbox download on demand. Utilize the existing frameworks and build a custom, performance tuned device that can compete with the iPad. Microsoft has proven it has both the hardware and software talent… now they need to be combined.

There are no 128 bit capable processors available. Microsoft is building for a market that hasn’t even started yet. They’ve barely transitioned to 64 bit! Windows 7 is the first Windows OS that has a truly viable 64 bit market. Manufacturers have had a few years to develop 64 bit drivers, and consequently, it is far more likely that you can install Windows 7 x64 now than you could install Windows XP x64. Add to that that almost every new computer ships with a 64 bit processor, and you have yourself one viable market. Software developers also now have mature SDK’s, and have had a few years to transition to 64 bit computing.

Microsoft has finally shipped Windows 7, with a much shorter timeframe than Windows Vista did. I suspect that they are going to try to maintain such an active pace, shipping software every 18-24 months or so, rather than the 5 years it took for Vista. There is no way that either the hardware market or the software market will be remotely ready for 128 bit computing in 2 years. Yes, they may make it ‘functional’, like XP x64 was ‘functional’, but it will not truly be 128 bit. I suspect that Windows 9 will be as 128 bit mature as Windows Vista was 64 bit mature. Don’t expect true 128 bit computing till Windows 10.

The other thing I found interesting was that Google Chrome does not even run on Linux yet (at least, no public builds). There is an alpha version for Mac, and a full version on Windows, but no public Linux builds. They have a long way to go on this project.

Another interesting facet that the concept of the web browser as the user interface brings to the game, is that of localized storage. If the only accessible UI is that of the web browser, then what happens when the user does not have access to the internet? Fortunately, HTML 5 brings localized database storage to the web browser. Unfortunately, Internet Explorer 6 & 7 do not support this. Internet Explorer 8 supports it somewhat, and most other web browsers (Opera, Firefox, Chrome, Safari) support it to varying greater extents. This is a boon for web developers. It will help shift the web experience to more standards compliant browsers.

The other option for localized storage is Google Gears, which Google will no doubt push developers to use. I think however, that the majority of developers will choose to go the open standards route, rather than Google’s technology.

That’s my perspective on the ‘web OS’ news this week. Feel free to leave yours in the comments.