Blog Archives

Microsoft support and the BSOD

I was on a call to Microsoft support today for an issue with Communication server 2007.  During the initial steps of the support call when the technician was trying to setup the Easy Assist session, his computer started a memory dump with a BSOD.  He had to jump to another machine while rebooting the now inoperable computer.  He apologized politely, but it was no big thing for me, I actually got a kick out of it.  It also helped that the Guy (Hi Patrick), was a BSG fan and we chatted a little about the upcoming Season 4 of Battlestar Galactica

Advertisements

FIOS customers – better DNS solution

If you have Verizon FIOS, and haven’t already customized your router to use better DNS Servers, do yourself a favor and do so asap.  By default Verizon uses “DNS assistance” on their *.*.*.12 DNS servers.  Set your router to use custom DNS servers and change the ending .12 to a .14.  Alternatively, using any of the other official Verizon DNS Servers in the range of 4.2.2.1 through 4.2.2.5 is also a good bet. 

If you don’t customize your DNS, the Verizon DNS assistance configuration can potentially cause issues for Tech Savy users who do VPN, host monitoring, etc.  You are also helping Verizon make more money by being served ads on their DNS assistance page. 

I came across this help link on the Verizon website that will guide you through changing your router config to optomize your DNS settings…

Black logon screen for Server 2003

This is the first time I’ve run across this issue, but what was happening is I’d see a black screen both through RDP and from the actual server console.  You could see a little bit of the windows logon box, but it was cut off where the username and password fields would be.  If you can fumble your way through the logon screen you’d be able to logon but everything was still black, the only color you’d see would be in the desktop icons.  This makes it virtually impossible to do anything on the server.  This situation does not cause a work stoppage, as all the background services ran fine, you just can’t logon and do anything. 

So I started searching around and ultimately ran across a MS KB article 906510.  I’ll paste in the information below, but the KB said it was an issue with the color scheme, and the registry values were all set to “0”.  At least that is whats listed as the cause of the black screen symptom, but it does not offer an explanation of what would cause the numbers to get set to “0” in the first place. 

To resolve this I had to do the following:  (NOTE: the drive mapping step was only needed for my solution as the server was not on the domain and not using internal DNS, so I could not authenticate on the domain).

1. Map a drive from another network machine to this server using local admin credentials.
2. Open regedit and connect remote registry to the server in question. (this will not work unless you map the drive first – see above note)
3. Use an export from a working server of the registry key noted in the MS KB and import that into the remote server.
4. Attempt logon through RDP or at the console, the color issue should be resolved. 

I wish I knew what caused this issue, but I can’t find much as to the actual cause.  I heard talk that maybe it was a disk space issue, but my server had 10GB of free space, so that wasn’t it.  Hope this information helps someone in the future.

Help and Support service missing in Server 2003

I ran into a situation in which the Help and Support service in server 2003 dissappeared.  I had no idea why, and for a long time left it alone.  I then did some digging and found a KB article from Microsoft describing how to re-install the service.  Follow these instructions (taken from MS KB article 937055):

To resolve this problem, reinstall the Help and Support service. To do this, follow these steps:

1. Click Start, click Run, type cmd in the Open box, and then click OK.
2. At the command prompt, locate the following folder:

%windir%PCHealthHelpCtrBinaries
3. At the command prompt, type the following commands, and then press ENTER after each command:

HelpSvc.exe /regserver /svchost netsvcs /RAInstall
HSCUpd.exe -i hscmui.cab
HSCUpd.exe -i hscsp_s3.cab

Note The following commands do not apply to some language versions of Windows Server 2003:

HSCUpd.exe -i hscmui.cab
HSCUpd.exe -i hscsp_s3.cab

The reason is that the Hscmui.cab compressed file is not included in some language versions of Windows Server 2003, such as the Japanese version of Windows Server 2003.

Terminal server – strange behavior

Something really strange just happened, and I can’t explain it.  Yesterday I was trying to install Terminal Services on a Widnows 2003 server.  I went through the normal install routine, and set my options and everything installed ok.  But when I went to view the configuration, it was set to Remote administration mode, and did not prompt me for this before install.  I was also trying to install Citrix and couldn’t get past the first part of the installer because it detected TS wasn’t in application mode.  I tried uninstalling and re-installing and nothing helped. 

Here is the weird part, this morning when I tried doing the same thing, uninstalling and re-installing, this time it worked fine.  I don’t even see a mode option in the configuration now and the Citrix setup is allowing me to proceed.  I didn’t do anything differently this time, same options and same license server, but this time it worked.  I have no idea why, but I’m not complaining, I just hate coming across things like this that don’t make sense and I can’t explain what happened.

Goodlink 5 deployment issue

I have been working on deploying multiple Goodlink 5 messaging servers recently, and came across a problem with the user migration feature in Goodlink 5.  Normally if you have more than one Good messaging server, you have the option to “change good messaging server” when you right-click on a user account.  So in my deployment when I didn’t see this option enabled, I had to call Goodlink support.  What I found out made sense, but is definately irritating. 

I called Goodlink support before deploying Goodlink version 5 to ask about deployment options and get some advice on my project.  I asked about the user migration feature and using multiple GoodAdmin mailboxes, and was told that it should work fine.  Now here is where the issue comes in.  In my deployment scenario, I wanted to deploy multiple individual Goodlink 5 servers in various locations/offices around the world.  The limitation here is network bandwidth and latency, due to the fact that the offices getting these Goodlink 5 servers are spread across the globe on various speed networks, and some of the offices in Asia for example, have very high network latency and bandwidth is limited.  To help with this situation, I asked if it would be advisable to use multiple GoodAdmin mailboxes, one for each new Goodlink 5 server, that way, each location that got a Goodlink 5 server and had a local Exchange server, would get its own GoodAdmin mailbox homed on the local Exchange server.  This would increase performance, reduce the time it takes to re-connect all the mailboxes in the event of a reboot, and provided a better solution for my deployment project. 

When you use a single consolidated GoodAdmin mailbox, and you have more than one Goodlink server which is connected to the network on a low bandwidth link or one that has high latency, it is a good idea to have separate GoodAdmin mailboxes, to keep network traffic local to the office in question.  Its faster, more reliable, and if the central Exchange server that houses the consolidated GoodAdmin mailbox were to go down, all your Good Messaging servers would be rendered useless until the Exchange server is back online and the GoodAdmin mailbox is available again. 

So my scenario of using more than a single consolidated GoodAdmin mailbox was a good idea, still works well and does help with performance and all the other reasons I mentioned above.  The drawback is that in this configuration, from within the Good Messaging console, you only see the server you are connected to under the “Good messaging servers” folder.  This also means that the “change good messaging server” option is grayed out and unavailable.  The fact is that the GoodAdmin mailbox is what keeps track of how many servers you have, but they all have to be configured to use the same GoodAdmin mailbox in order for the server to “know” about the other Good Messaging servers you have in your network.  This basically means that the user migration option is not going to work for us. 

This is not a problem, its just an inconvenience.  And it would have been nice to have known this initially when I talked with Good support eariler.  Lately no matter what company I deal with, I usually end up getting conflicting information about their products and features based on who you talk to.  The first time I call and get info on something, I hear one thing, then when I am working on the deployment and have a question or problem and call back, I hear something completely different.  This is irritating and can cause some major problems for large projects. 

Phone number normalization in OCS 2007

I ran into this issue with OCS 2007 and thought it might come in handy for someone in the future.  Lets say you deploy OCS 2007 in your environment, and in AD you use the phone number field for the office ext, and you want to keep that format.  Lets also assume you are noticing that if you use this configuration you end up with warnings in communicator saying it cannot synchronize the addressbook.  When you check the event log on your OCS server you see errors and warnings about the addressbook server.  You won’t find a service for this component (at least I couldn’t find it), it appears to be built-in to the OCS server software.  You also find a list of invalid AD phone numbers in your C:Program FilesMicrosoft Office Communications Server 2007Web ComponentsAddress Book FilesFiles directory. 

So how can you keep your internal extensions in their current format and still communicator without having to see the annoying addressbook warning.  Its not simple, but the fix is not too complex to impliment.  Here is what to do.  This assumes that you use a format such as “x 5555” for your extensions in AD for the users in the phone number field. 

First, in the OCS 2007 console, right click on the forest, and go to properties, then select voice properties.  Next, click the “Add” button to setup a new default location profile.  Give it a name and a description.  Next, under “Normalization Rules” click the “Add” button.  Give this a name also and a description.  Under translation, enter “^(x|X|x |X )“ without the quotes in the “Phone pattern regular expression” box.  In the next box, enter “$`”  Thats a dollar sign, immediately followed by the key to the left of the number one key on the top number row of your keyboard.  To test this, enter an extension in the same format as you use in AD in the “sample dialed number:“ box.  If it works you should see the resulting number in the format you want in the bottom text box. 

Save you changes by clicking “ok” until you are back at the console screen.  Then you just need to issue some commands on the server to update the address list, which are listed below: (NOTE: CD To the directory of your abserver.exe location)

Click Start, click Run, type cmd, and then click OK. 

Type net start abserver, and then press ENTER.

Type abserver.exe –regenUR, and then press ENTER.

Type abserver.exe –syncNow, and then press ENTER.

Exit and re-start communciator on your PC/laptop and it should now load with no errors or warnings and you will now see your extensions listed.

RPC over HTTP(S) for Exchange 2003 – single server

I’m pleased to announce that I’m now able to access my exchange mailbox from anywhere in the world using RPC over HTTP(s).  This was a lot easier to setup than I thought it would be, and following the steps provided by Daniel Petri helped a lot!  I also utilized the RPCFrontend tool that he mentions in the link.  This made things very easy and I got it working the first time I attempted the configuration. 

What this means is that I can now access my mailbox on my Exchange server, from anywhere with the only requirement being an internet connection.  I can just open Outlook, and go straight to my mailbox as if I were on my home network.  I highly recommend this for anyone with Exchange servers, it makes remote connectivity so much easier and can be fully secured with SSL and other security options. 

What no one ever tells you about cluster hardware redundancy

I was recently reminded about this, so I thought I’d post about it and share the knowledge.  I use mostly HP hardware, so in this setup, lets say we have two HP DL585 servers for use in a Windows cluster.  I also have dual fibre channel interfaces in each server.  Externally lets say we have two separate fibre channel switches.  For storage, its an MSA1500 external fibre channel SAN with over 1TB of RAID storage, partitioned for use in a cluster.  The MSA1500 has dual controllers and each controller has its own fibre chennel interface. 

To connect it all together, each server connects one of its two fibre channel interfaces to one physical SAN Switch.  The MSA1500 SAN controllers each get connected to a separate physical SAN switch.  The goal here is to optomize hardware redundancy and make this configuration as fault tolerant as possible.  But here is where the issue comes in. 

No one tells you that you need special software running on each node of the cluster to handle the hardware redundancy.  If you don’t have this software in place, what you will find is that Windows will see two SAN storage arrays.  One of them will be inaccessible since you can only have one physical connection active.  I had to call HP support and go through several explanations of what I was trying to accomplish with all the hardware redundance before I was told about the software I needed.  Its called MultiIO or “Multi Path IO”.  If you do a search on the HP support website, you will find HP MPIO Basic (there are two versions, basic is fine for most people).  Download this software and install on each node and amazingly you will find that things work much better.  From the documentation I saw, there is no indication of needing this software, but in this kind of highly resiliant configuration it is necessary to use MPIO to control the hardware redundancy. 

Ran into issue with LCS 2005 and OCS 2007

LCS 2005 (Enterprise Edition) is installed on a single server with SQL 2005 server for the backend database. ?There is no warning or indication that this would cause any problem and up until now it?hasn’t.? The software does allow for this type of deployment and works fine in this configuration.? OCS 2007 server was recently installed and configured.? During initial testing it was observed that users on the 2005 LCS server could not see presence data and communicate with users on the OCS 2007 server (and vice-versa).??After searching online for solutions it was found that there are two critical hotfixes for LCS 2005 that were said to correct this issue.?

Installing the hotfixes:
Following the instructions on the first hotfix documentation, I installed hotfix 911996.? The installer completed successfully and there was no indication of a problem.? Next I installed the second hotfix 921543, again the installer ran successfully and there was no indication of a problem.? I then checked the services for LCS and found that the LCS Server services would start, but then stop on its own with an error:

Event ID: 12299
Source: Live Communications Server
Details: The service is shutting down due to an internal error.
Error Code: 0xC3EE78F8.

The hotfix documentation mentioned this error as a possibility if you install the patch on the front-end server before the backend patch is installed.? However, since this was a single server deployment with both the back-end database and front-end server components on the same physical server, the hotfix only installed the update for the front-end components.? Again, there is no documentation saying this is not a supported configuration and the installer will gladly allow you do install both components together without a problem in the LCS 2005 setup.? (NOTE: OCS 2007 server prevents this configuration in the installer and will not install both components on the same box when using Enterprise Edition).? There is also no warning in the hotfix documentation not to install the update if you have both components on the same physical box.?

For a resolution, I called Microsoft support services ticket number SRX080212602272 and spoke to an afterhours support specialist.? After some initial troubleshooting and research the technician was able to review a similar case they had a few months ago.? He attempted the resolution listed in their issue history which did ultimately fix the issue.? To resolve, a string value must be deleted from the LCS 2005 registry branch (HKLMSoftwareMicrosoftReal-Time Communications{Your GUID}.? This key “role=EE” ?had to be deleted from the registry in order for the hotfix to install on the back-end database.? Once this was done, the LCS 2005 server service started successfully and we tested connectivity with a test account to verify that the server was working normally.? Basically this tricks the installer into thinking that the server is the backend database only, and it will allow the hotfix to install the back-end components.?

This issue could have been avoided completely if Microsoft had updated their documentation for the hotfix, especially since they had issues with this months ago and should have updated the documentation.? I am recommending to the technical support manager that the documentation be updated since this is a known issue and others have experienced the same problem.? Hopefully this will get done and prevent others from having the same issue.?

Tags: , , , , , ,