Confused Amused » sql

OCS2009-DBUpgrade.msi with a 32-bit SQL Server Back-End

Tom Pacyk — Wed, 11 Nov 2009 17:02:20 +0000

I wanted to point out a quick note about KB 969834 aka the OCS2009-DBUpgrade.msi file – The KB article suggests running the package from your Back-End database server, but if you’re running SQL 2005 x86 you’ll be greeted with the following error:

This installation package is not supported by this processor type.

Basically, the MSI needs to be run from an x64 machine so your only option now is to run the update directly from your Front-End server. If you try to launch from there you might receive this error:

You must install Microsoft SQL Server 2005 Client Tools before you install Microsoft Office Communications Server 2007 R2 (KB969834).

You could try install the SQL Tools and Service Pack updates from installation, but OCS is looking for very specific versions of the SQL tools. The quickest and easiest way is to just use a couple of downloads from the Feature Pack for Microsoft SQL Server 2005 – February 2007.

You’ll want to download and install the following on your R2 Front-End before running the update:

Microsoft SQL Server 2005 Backward Compatibility Components (x64 package)
Microsoft SQL Server 2005 Management Objects Collection (x64 package)

After running those installers you should be able to run the DB upgrade successfully. Don’t forget – you need to run that MSI from a command line with the poolname (Non-FQDN version) parameter. And if you’re using Server 2008 be sure open the command prompt as Administrator so it runs with elevated rights. Example:

OCS2009-DBUpgrade.msi POOLNAME=MyFirstPool

Your OCS Front-End and DPM 2010 Part 3: Recovery

Tom Pacyk — Fri, 09 Oct 2009 22:55:46 +0000

Now would normally be the time where everyone is running around like their head has been cut off because your Front-End server is totally hosed, but because you followed the backup procedures in Part 1 (you did run the backup, right?) restoring service to your OCS server is fairly simple.

Restore the Database

Open up the DPM console.
Click the Recovery tab at the top.
We need to restore the SQL database and files separately, but let’s start with the database. Expand the tree to \\All Protected SQL Instances\\RTC\rtc
Highlight a suitable recovery date in the calendar and select the RTC database below.
Right-click and select Recover…
Press Next.
We’ve successfully screwed up the server to where we might as well recover to the original SQL server. Select that option and press Next.
Select Leave database operational and press Next.
No options needed. Just press Next.
Yup, those are the files we need. Press Recover.
Press Close while the recovery operation occurs.
If you click the Monitoring tab you can view the jobs in process.

Restore the Files

Now we need to restore files separately. Expand the tree to \\All Protected Protected Volumes\
Highlight a suitable recovery date in the calendar and select the Program Files folder below.
Right-click Program Files and select Recover…
Press Next.
Select Recover to the original location and press Next.
Select to Overwrite the existing versions (if any), and then select to Apply the security settings of the recovery point version. Press Next.
Now press Recover.
Press Close while the recovery operation occurs.
If you click the Monitoring tab you can view the jobs in process.

Fix SQL Database Chaining

One thing DPM won’t restore is an option within SQL. If you miss this step your Front-End services will fail to start.

Open SQL Management Studio (Express).
Press the New Query button.
Enter the following text: sp_dboption 'rtc','db chaining',TRUE

Press Execute.
```
 
```

Bounce the Server

If you check your OCS Front-End you’ll find all the files you deleted previously have now returned. You could probably get away with restarting services as this point, but since the machine was completed hosed I’m just going to restart the server and cross my fingers.

Check Functionality

After the restart all of my OCS services started successfully and my errors have gone away. You can see now my Communicator list still has my contacts and access levels defined. Likewise, Device Updates and client auto updates should function normally now.

Your OCS Front-End and DPM 2010 Part 2: Destruction

Tom Pacyk — Fri, 09 Oct 2009 22:55:35 +0000

Now that we’ve verified the DPM backups are running successfully on a regular schedule we can get to really destroying the environment. First up: the RTC database. So shut down your OCS Front-End and SQL services. Then go and delete the RTC.mdf and RTC.ldf files. I know that doesn’t sound like a good idea, but really, delete them.

Open Explorer, jump in to the following volumes and delete the content there:

\\Application Host\Application Data
\\Web Components\AutoUpdate
\\Web Components\Data MCU Web\Web
\\Web Components\Data MCU Web\Non-Web
\\Web Components\DeviceUpdateFiles

Now go and start your SQL services and try starting the OCS services up again. You’ll find a few errors and warnings in your OCS application log because it can’t read the RTC database. Communicator and Live Meeting clients won’t be able to connect to the server as this point either. Oops!

Congratulations, you’ve successfully messed up your Front-End server to the point where it is non-functional. The device update files have been lost, the MOC Auto-Update files have been lost and all your meeting content is gone. In the next section I’ll demonstrate how to get the server back to an operational state with DPM.

Your OCS Front-End and DPM 2010 Part 1: Backup

Tom Pacyk — Fri, 09 Oct 2009 22:55:25 +0000

The goal of this series is to demonstrate how to recover your OCS Front-End’s RTC database in the event of a disaster where your database or disk hosting the RTC database has become corrupted. Or maybe you’ve recovered a server’s installation and configuration, but now need to recover the user information. I’m going to do this in 3 different parts: backing up, destroying, recovering. To get started we need to have a semi-realistic OCS environment running so in this example I have a Standard Edition Front-End running where I’ve done the following:

Added a few users to my contact list and changed the access levels around.
Uploaded and approved the latest UCUpdates.cab package for phone devices.
Added a MOC hotfix to the auto update feature.
Created a couple of conferences with content in Live Meeting.

These items may seem a little random, but they’ve been done to illustrate what’s restorable from the RTC database and the file shares on a Front-End. I also have another machine called OR1DEVDPM01 running the beta of DPM 2010, which is what we’ll be using for the backup and restore.

Now that we have a machine running we first need to get this thing backed up before we trash it. You’ll want to create an exception on the OCS machine for the firewall to allow any traffic from the DPM machine. This will allow installation of the DPM agent, and allow backups and restores to occur.

Install the DPM Agent

Open the DPM console.
Click the Management tab at the top.
Click the Agents tab below the main navigation line.
Click Install in the action pane.
Select Install agents and press Next.
Select the server you’re pushing the agent to (OR1DEVOCS01) and press the Add button. Then press Next.
Enter the credentials of an account with administrative rights on the server and press Next.
Since this is a lab and I’m using Server 2008 I’m not too concerned the server restarting, but in production I’d advise opting for the manual restart.
Press Install and then you can click close while the agent deploys.
After a minute or two the agent status should change to OK. Now we can start backing up the server.

Add the OCS Protection Group

In the DPM console again click on the Protection tab.
Click Create protection group in the action pane.
Select Servers and press Next.
Expand the OCS Front-End, OR1DEVOCS01 and you’ll see a few different nodes such as shares, SQL, volumes and system state. The OCS Backup and Restore guide provides some guidance on what actually needs to be backed up from the server. These options pertain to a Standard Edition Front-End so be sure to check the document for any other role. Here are the items we need to select:
- All SQL Servers\<\RTC\rtc<>
- All volumes\\\Application Host\Application Data
- All volumes\\\Web Components\AutoUpdate
- All volumes\\\Web Components\Data MCU Web\Web
- All volumes\\\Web Components\Data MCU Web\Non-Web
- All volumes\\\Web Components\DeviceUpdateFiles
You can press OK and ignore the warning that pops up about adding the system state backup. Press Next to continue after selecting all of the above options.
Name the Protection Group something descriptive. I’m going out on a limb here, but I used OCS Front-Ends as the name. I don’t have any tape libraries hooked up, so I’ll just be opting for short-term protection to disk. Press Next.
I imagine you’ll generally want more than 5 days of backups, but this works for the purpose here. 15 minute synchronizations are OK, but keep in mind OCS uses a simple recovery model in SQL meaning you take full backups and you restore full backups. None of this full, plus incremental and rolling logs forward fun. Just flat out restore of the entire DB and logs at once. The problem here is a simple database recovery model cannot leverage the synchronization feature of DPM like incremental backups can, so we’re limited to being able to restore only from a full backup, or an “Application recovery point” in DPM terms. You’ll see the default is to back up every day at 8 PM which may or may not be acceptable for you.

If you press the Modify button you can add in additional time slots to run an express full backup. Unfortunately (hoping this is a beta bug), you can’t select all the time slots and press Add. So just press Add quite a few times until each timeslot is added and you’ll have a recovery point every 30 minutes for your database. The trade off to running with this kind of frequency is the disk space used. Pick a schedule that’s appropriate for your deployment. Press OK to accept the schedule and then press Next to save the short-term goals.
On the next page you’ll see the disk allocation. Press Next to continue.
Select when to create the replica of your data (now) and press Next.
Choose when to run consistency checks and press Next.
On the last page you can review your selections and then press Create Group. The initial replica jobs will be created and then you can press Close.
If you click the Monitoring tab you can view the jobs in process.

At this point we should have the backups running from the Front-End server. The next part of this will be destroying the data and blowing up the server. After that I’ll show how to recover everything we destroyed.

Checking Communicator Endpoint Versions On Your OCS Pools

Tom Pacyk — Fri, 11 Sep 2009 20:00:35 +0000

One of the questions that comes up with OCS deployments that have been around for a while is the question of what clients are connecting to the pool. This can be controlled with client version filters and the auto update feature of R2, but more often than not there are some straggling clients out there. The challenge for those without some sort of configuration management tool is identifying what users have those old clients.

Microsoft has been nice enough to provide a handy tool within the OCS 2007 R2 Management Console that checks what kind of endpoints are connected to your Front-End server. If you open the OCS MMC, click the pool object and then click the database tab you’ll see a number of reports you can run. One of the more useful ones is the client version summary. Just press Go next to it and you’ll see it return a list of endpoints.

You can see from the results we still have quite a mix, and even someone still using a Communicator 2005 client! This is useful in providing an overall picture of what’s been used, but the question I immediately hear next is “Who’s using that version?” Unfortunately, there’s no easy way to tell in the console. You can run a per-user-report which will tell you the endpoints a particular user is signed in with, but that’s going to be a tedious effort to chug through a long list of names trying to find the offenders who haven’t updated their clients. You can see below what searching for a single user returns.

In order to answer the question of who’s using what we need to run a SQL query against the RTCDyn database. I’ll say this is definitely not a recommended/supported task, so be sure you know what you’re doing here before you start messing around in SQL Management studio. You have the potential to really hose your OCS installation if you start changing database content. The query we’ll run is just a SELECT statement so we shouldn’t cause any problems. Still, you’ve been warned.

Open SQL Management Studio. If you have a standard edition pool you can download and install SQL Management Studio Express for free. Press the New Query button and paste in the following query. Then just press the Execute button. You’ll get a list back of SIP URIs along with the endpoint they are currently using.

SELECT CAST([SipHeaderFrom] as varchar) as "SIP URI"
      ,CAST([ClientApp] AS varchar) as "Endpoint"
FROM [rtcdyn].[dbo].[Endpoint]

That will give us a nice long of everything in use and what SIP URI is signed in with that client.

Say we want to filter because we’re looking for people with a specific version. In this case, we want to find everyone still using the R1 MOC client so we can add a WHERE clause that searches for strings that match the agent header.

SELECT CAST([SipHeaderFrom] as varchar) as "SIP URI"
      ,CAST([ClientApp] AS varchar) as "Endpoint"
FROM [rtcdyn].[dbo].[Endpoint]
WHERE CAST([ClientApp] as varchar) like '%2.0%'

You could replace that 2.0 with anything else returned in the agent headers such as 3.5, OC, LCC, etc. This only queries the clients that are connected at a specific point in time so you may want to run this from time to time to catch clients that may not have been connected the first time you Hope this helps you identify your clients.