List the Manager Emails for AD Group Members with PowerShell

Below is a quick script I’ve been working on for a colleague. Its simple purpose is to query an AD group then display each member’s username and the email address of their manager. Hopefully this will help others accomplish this common task with ease.

(Update 04/11/2018)

I modified the script to include the employee’s full name and also to sort the output by that name. Additionally, the way I was handling the variable members and naming bothered me. I knew it could be better. So I did some more research and cleaned that up utilizing the -ExpandProperty option for select.

 

Getting Started with PowerShell Desired State Configuration (DSC)

As I’ve mentioned in other posts, Desired State Configuration (DSC) is a powerful technology with a lot of potential. However, due to how new it is and how rapidly it’s evolving, it can be difficult to get started and figure out how to accomplish your specific goals. My intention here is not to give an exhaustive look at the ins and outs of DSC (I’m not qualified to do that), but rather to give you the tools to get started and be successful with it .

Step 1: Get a Baseline

After years of cobbling together information, then having to go back later and relearn how to do it the right way, I’ve learned the value of getting the framework right from the start. If you have a strong framework in mind for how something is built and how it’s intended to be used, then building useful things on top of it is much easier. I had the same experience when learning DSC. There is a lot of information on the internet about it, but much of it is “old” and out of date. Also, everyone has their own opinions for how it should be used. Eventually I came across two videos on Microsoft Virtual Academy that put everything into perspective. They are taught by “Jeffrey Snover, Microsoft Distinguished Engineer and inventor of PowerShell,  along with Windows PowerShell MVP Jason Helmick”. This is a fantastic starting point for diving into DSC and I can’t recommend them enough.

Getting Started with PowerShell Desired State Configuration (DSC) – Microsoft Virtual Academy

Advanced PowerShell Desired State Configuration (DSC) and Custom Resources – Microsoft Virtual Academy

Step 2: Working With New Resources

Now that you know all about DSC and how it can be used, it’s time to put that knowledge to work. You eagerly download the module you want to use (let’s say xSQLServer, for example) and are ready to have machines install SQL for you. The first inclination is probably to google it to see how it’s used, which will lead you to github or PowerShell Gallery. Those are great for getting information about the package and its change tracking, but not much use for actually implementing the module. So here you are with a brand new toy and no manual.

Examples

The first thing to do with a new module is always check for an examples folder. Your module was probably installed in C:\Program Files\WindowsPowerShell\Modules.

Opening the examples folder within that module will reveal a list of scripts made by the creating team for various scenarios they see the module being used in. Your mileage may vary depending on the module you’re using and who made it, but generally those produced by Microsoft have useful information. This is where I obtained the example file that my last post, “Installing SQL Server Using Desired State Configuration“, was based on. Again, how much explanation is included within each script is completely up to the discretion of the creator. That’s why it’s important to first watch the videos linked above. Then, whether there is proper documentation or not you can make sense of it yourself.

(Upon further inspection, the examples are sometimes also available on the github site.)

Interrogating Resources

Even when you have an example file that closely matches your needs it’s likely that you will still want to customize it. Many times, the module you are working with will have a resource you need but not an example of it listed in the file. Or you simply want to know what all you can do with the module. As usual, it’s PowerShell to the rescue.

Using PowerShell we can easily find which resources are available to us in a module using Get-DscResource.

After finding a resource that interests us, we can dig further down into its specifics. In order to get more than cursory information, it s necessary to expand the properties field.

From this information we can tell which fields are available to us, what their data types are, and which ones are required.

It should also be noted that you can discover these from right within ISE as well. This is the improvised method that I used before discovering the PowerShell cmdlets above. If you type anything within the resource block that it doesn’t recognize, intellisense will automatically suggest the proper fields to use by hovering over the angry red line.

And, if all else fails, in the end resources are just PowerShell scripts. You can go to their folder and open them like any other file (ie C:\Program Files\WindowsPowerShell\Modules\xSQLServer\DSCResources\MSFT_xSQLServerLogin).

Step 3: Be Brave

Armed with all of this knowledge there is but one thing to do, be brave. Start putting some configurations together, make mistakes, then use the lessons learned to make better configurations. This is an exciting technology in which things are rapidly moving and changing. In fact, within days of my last post (and while writing this one) I discovered that xSQLServer had been retired in favor of SqlServerDsc, and I’d had no idea.

So get at it, make your own creations, keep your eyes open daily for changes, and let me know if you have any questions. I look forward to learning with you.

Installing SQL Server Using Desired State Configuration

(Update: I’ve since discovered that SqlServerDsc has replaced xSQLServer.)

One of my growing passions is using PowerShell Desired State Configuration (DSC) to automate all the things. I started out with simple configurations for testing but wanted to dive into more complex\useful situations for my day-to-day DBA life. To be honest, I was intimidated by the idea of doing a SQL installation. Configuring simple parameters or creating a directory are easy enough to wrap my head around, but something as complex as a DBMS installation gave me pause. I’m here to tell you that my worries were unfounded, and that you should have none as well.

The blessing and curse of DSC is that it’s so new. It is without doubt a very powerful tool, but as of yet there isn’t a lot of documentation around the individual resources. Or worse yet, the pace of improvement moves so quickly that information from two years ago is now out of date. I plan on doing a separate post for how to approach DSC given these realities. With this post, however, I wanted to fill one of those documentation gaps. Specifically, how to install and configure an instance of SQL server. I based my file off of an example one provided by Microsoft in the “Examples” folder of the xSQLServer module named “SQLPush_SingleServer.ps1”. Pro tip: always look for example folders in the modules you want to work with. It should be noted that you can address much more complicated scenarios, such as setting up clusters or Availability Groups, but for simplicity this configuration will be creating a single instance on one node.

If you have experience with DSC or simply don’t want to listen to me drone on about the details, the full configuration is at the bottom. For those interested in the play by play, or just bored and looking for something to do, I’ll address each piece individually.

The script starts out with compulsory documentation on the script and what it does. Kidding aside, get into the habit of doing small sections like this. Your coworkers (and you years from now when you’ve forgotten what you did) will thank you.

Next, we hard-code a couple of items specific to your individual run of the script. List the computer(s) that you want to deploy to as well as a local path for the configuration file that DSC will create.

Following that, we will set how the Local Configuration Manager on the target nodes is to behave. We’re specifying that the configuration is being pushed to it, that it should automatically check every so often for compliance to this configuration and auto-correct anything that’s not aligned, that modules on the node can be overwritten, and that it can reboot if needed.

Following that is the actual configuration details, where all the fun is defined. Mine is named “SQLSA”, but it really doesn’t matter what you name it. This is like defining a function; so as long as you call it by that same name later, little else is relevant. You’ll see at the top of this section there are three “Import-DscResource” lines. This tells the configuration which DSC modules will be needed to perform the actions we’re requesting.

The WindowsFeature item is one of the most handy in DSC. This allows us, as you might guess, to install Windows Features (in this case the .NET Framework).

Next I’ve created a firewall rule to make sure our instance’s port will be open (this is defined later under xSQLServerNetwork). It’s worth noting that there is a resource built into xSQLServer that allows you to configure firewall rules for SQL. However, I did not like the behavior of it and found that xFirewall from the module xNetworking provided a lot more flexibility.

Up next is the actual meat of installing SQL Server. The if($Node.Features) block is something I picked up from the example file. I’d say it’s redundant to check for whether you’re installing SQL when you came here to install SQL, but hey, it works well so I left it.

One way I’ve altered this section from the original is to parameterize everything. If you look further down there is a $ConfigurationData section. Having all of our customizable fields there allows us to easily change them for each deployment (dev, test, prod) without having to search through the code. You and your team will know exactly where to go and what to change for each situation.

I’ve also included some examples of basic SQL Server tasks like creating a database, disabling the sa account, disabling a feature like xp_cmdshell, and configuring the network port (referenced earlier). The naming on these items looks odd but makes sense. By adding in the node name we can ensure that they are unique should we deploy to more than one target node. And adding a friendly name to the configuration item, like “sa”, makes it easy to tell DSC which item depends on which. Speaking of which, note that each of the configurations depends on the base installation. That way DSC will not run those if there is nothing to actually configure.

After the configuration definition we have the $ConfigurationData mentioned earlier. It’s a great idea to get in the habit of using sections like this. It will make your transition between various environments much easier.

The next section details what we’d like the instance name to be as well as what features should be installed. It’s very picky about the feature names, and they don’t line up exactly with a standard command line install. So be careful with what you place here. It won’t install anything incorrectly, just simply cause the configuration not to run and you to lose your mind.

Also in this section, we’re copying over the modules that each node will need in order to perform this configuration. This isn’t necessary when using DSC in pull mode, but that’s a story for a different post.

I know you thought it’d never come, but at last it’s time to actually do something with all of this. We call our “SQLSA” configuration, passing in the $ConfigurationData and specifying to place the resulting .mof file in $OutputPath. After that, configuration is started on each node using Start-DscConfiguration and calling the .mof that was just created. Lastly, the node is tested to make sure it’s not out of compliance.

If all goes well, your output will lack red and eventually will end in a message like this:

And that’s all there is to it! Not so scary after all. I deployed my first DSC SQL Server while making tea and wondered why I’d been doing it any other way…

 

 

My First PowerShell Module – PSOraenv

Most of my experience with Oracle has been on Linux, but recently I began working with it on Windows as well. It came to my attention very quickly that oraenv, my beloved friend in Oracle administration, is not present on Windows installations. When you have more than one database installed on a single server, or perhaps ASM (which has its own Oracle home), manually swapping back and forth to work with the different pieces gets really old really fast. For this reason, and because I’ve been anxious to learn how to make my own PowerShell modules, I created PSOraenv. Its purpose is to closely mirror the capabilities of oraenv and allow a similar level of simple, yet powerful, command-line flexibility on Windows systems. Let me give you some examples.

By default, the necessary environment variables for Oracle aren’t set when you fire up PowerShell or Command Prompt. (It seems to default to the most recently installed product.) Using my cmdlet Get-OraEnv, we can verify that they start empty.

In order to see what options are available to us, we run my cmdlet Get-OraSID to pull back a list of the database SID’s and associated Oracle homes on the local machine.

Now that we know what SID’s are available to us, my cmdlet Set-OraEnv can be used.

Now, if we run Get-OraEnv again, we can see that the environment variables have indeed been set.

And, just to prove that actually matters, sqlplus can verify which database we are currently working with. (My sandbox system defaults to lkftest2.)

And there you have it! We have easily viewed which databases are on the local server and quickly swapped to the one that’s needed. It’s just as easy to swap back to the lkfowler2 SID if needed.

This module can be installed from PowerShell Gallery using Install-Module, like the example below.

(You can also find the PowerShell Gallery page at https://www.powershellgallery.com/packages/PSOraenv/1.0)

If you’re interested to see what’s in the code, that has been placed below. I welcome any comments, feedback, and functionality requests. Hope you enjoy!

 

Windows Server Core Jumpstart

Recently I’ve been looking into the potential that Windows Server Core holds for our environment. Like most eager new Core users, I imagine, I jumped in with grand visions of spinning up a VM quickly and being off to the races administering it from my desktop. The reality wasn’t quite the same, as I ran into a chicken and egg situation wondering how I could set up the machine when I could not yet connect to it. To complicate the issue, I couldn’t find a concise list of information on exactly what is needed to simply make the machine available so that I could begin to work with it.

With that in mind, I’ve compiled the following information in hopes of saving others the same headache. There’s nothing earth shattering here, but hopefully it will allow people to get started with Server Core quickly so that they can move on to more important things, like how the server will actually be used.

Let me know if you have any questions or suggestions. Hope it’s a help to you.

  • Ports to request from your firewall team.
    •  TCP
      • 5985, 5986 (WinRM)
      • 445 (SMB) –This is up to you. I wanted to be able to move files to/from the server.
      • 135
  • Local firewall rules to allow remote administration.
    • Enable Remote Management groups
      (Note: If you enable “Remote Service Management” on the host first, then you can do the others via PowerShell remoting. This can be helpful since copy/paste in things like VMWare console doesn’t always work.)

Default outbound traffic to allow

Enable Ping (optional)

  • Remote management tools
    • Add the remote computer to Server Manager (available on Windows desktop and server versions).
      • Once added, you can easily launch Computer Management and PowerShell for that specific machine by right-clicking it.
    • Connect via PowerShell remoting.

      • Cross-domain PowerShell Remoting (ie Dev or Test domains)
        • If remoting isn’t enabled on your local machine, enable it.
        • Add machines to the TrustedHosts list. (Depending on your setup, you might have to substitute IP addresses for the machine names in -Value.)

          Verify with:
        • Use PSSession to connect
      • IIS management (run on remote machine)
        • Set HKLM\SOFTWARE\Microsoft\WebManagement\Server\EnableRemoteManagement to 1.
          (This can be achieved using the local regedit tool and connecting it to the remote machine.)
        • Restart the WMSVC service.
        • Connect from local IIS Manager for Remote Administration with the local administrator credentials of the remote machine.
      • You can either use sconfig or the following remote PowerShell commands to allow Remote Desktop. (This is especially helpful for quickly getting to sconfig and other commands that do not operate properly with remote PowerShell.)
  • Common configuration tasks
    • The utility “sconfig” can be used for most setup items.
    • For a more speedy and scriptable setup, below are some common configurations via PowerShell.
      • Change date\time
      • Change computer name
      • Add to the domain

         

 

The Final Push – CRM 2016 is Live

After months of ups and downs, plans built and destroyed, and countless hours invested, CRM 2016 is live. (Actually, it’s been live since Easter weekend. But keeping a blog up to date is obviously not one of my talents.) For the last installment in this series I’d like to go over how we went about the upgrade and highlight some successes. Before I get into that, though, there is one essential thing that needs to be understood about this entire project. From the moment we really got started, it was placed in the hands of Jesus Christ. From planning to implementation, and everything in between, it was submitted to Him. I don’t say that to be cheesy or legalistic, but rather to give credit where credit is due.

As I’ve mentioned before, we were attempting to go from our production Dynamics CRM 2011 environment all the way to the latest Dynamics CRM 2016. That environment has a database that exceeds 1.5TB and numerous integration points with systems like our dialer. During planning we reached out to a vendor, who shall remain nameless, for their advice. The overview of their plan looked something like this:

Following this approach, the migration would flow as such:

  • Shut off the PROD 2011 servers.
  • Back up the PROD 2011 organization database.
  • Restore the PROD 2011 database to the temporary CRM 2013 environment.
  • Import the PROD 2011 database to CRM 2013.
  • Back up the upgraded CRM 2013 database.
  • Restore the upgraded CRM 2013 database to the temporary CRM 2015 environment.
  • Import the upgraded CRM 2013 database to CRM 2015.
  • Back up the upgraded CRM 2015 database.
  • Restore the upgraded CRM 2015 database to the temporary CRM 2016 environment.
  • Import the upgraded CRM 15 database to CRM 2016.
  • Back up the upgraded CRM 2016 database.
  • Restore the upgraded CRM 2016 database to PROD 2016.
  • Import the upgraded CRM 2016 database to PROD 2016.

We built out the “hop” environments as this plan suggested and began to test the theory. Aside from multiple technical challenges, some of which I’ve addressed in other posts, even when it worked the timeline was simply too long. End to end it took 1.5 weeks or more, and that’s not counting any additional steps such as importing our customizations or the work for our data warehouse. Taking down a tier zero application for one to two weeks was NOT an option for us.

The more I thought over the proposed plan, our past CRM experiences, and the options we had available to us, a new plan came into view. I wasn’t sure it would even work, but decided to do a “mock migration”, if you will, and test it out. The new approach looked like this:

The smaller database symbols represent the config database for each temporary environment. The temporary 2016 servers were eliminated altogether since our final (PROD) hop was 2016 anyway. Basically, I just made the 2011 organization database a central upgrade spot on a new SQL instance and pointed all the app servers there.

Following this new plan, the work flow became:

  • Shut off the PROD 2011 servers.
  • Remove the PROD 2011 organization database from our PROD Availability Group.
  • Detach the PROD 2011 organization database and reattach it under the new PROD 2016 instance.
  • Import the PROD 2011 database to CRM 2013.
  • Remove the organization from CRM 2013.
  • Import the upgraded CRM 2013 database to CRM 2015.
  • Remove the organization from CRM 2015.
  • Import the upgraded CRM 2015 database to CRM 2016 (using PROD App servers).
  • Remove the other copy of the CRM 2011 database and sync the upgraded 2016 database to the other Availability Group node.

This had a profound affect on our timeline. Each of our backup and restore operations had cost us around six hours each, for a total cost of around 48 hours. By pointing all of the app nodes at one central SQL server, that was completely eliminated. It also allowed us to utilize the beefier PROD hardware. Instead of 1.5 weeks our tentative database upgrade timeline became 31-36 hours, very doable for a weekend maintenance window. And that was on slower storage that had been provisioned only for the mock migration tests. In the end our actual time for full migration was less than 24 hours. That includes not only the database upgrade portion but also importing customizations, changing our data warehouse structure, reconfiguring connected systems, and troubleshooting issues.

There was truly a fantastic team that worked on this project and I’m honored to have shared in the adventure with them. There is a great deal more work not detailed in these blog posts because I didn’t have direct insight into those portions. I’d like to call out a few of them at a high level though.

  • Our development team spent months updating old code, much of which predated their employment here. They also developed innovative new ways of approaching old problems and bringing everything into compliance with CRM 2016. With the amount of customization we have, this is more than commendable.
  • My co-upgrader, Matt Norris, put in many long hours and handled with stride any challenge thrown at him. He was new to CRM when we started but I now refer to him as “battle-hardened”.
  • My fellow DBA, Bret Unbehagen, single handedly worked out the difference in table structures between CRM 2011 and 2016, determined how that would affect streaming data to our Oracle data warehouse, and created a process to mitigate those issues for the go-live weekend. I can’t even begin to describe to you how impressive this is because I don’t even fully understand it myself.
  • Our networking and storage teams were indispensable in preparing for and carrying out the upgrade. I asked for terabytes upon terabytes of space for various tests and mock migrations, as well as innumerable firewall requests, and they delivered every time. The compute guys also gave me additional resources on the app server virtual machines, to which I credit much of our surprisingly small timeline.

It’s a blessing and a privilege to have been a part of such a successful effort. There is nothing more satisfying than doing good work to God’s glory.

If you have any questions about how we approached the upgrade, mock migrations, etc please feel free to ask. I’ll be glad to answer anything that doesn’t put my job at risk 😉

Technical Details

App Server Versions at Upgrade Time

  • 2013: 6.1.0001.0132
  • 2015: 7.0.0001.0129
  • 2016: 8.1.0000.0359

SQL Server Version

  • Microsoft SQL Server 2014 SP2

Dynamics CRM Install\Import “The SQL Server {sqlserver_name} is unavailable”

My apologies for the significant gap between my last post on our CRM 2016 Upgrade Adventure and this one. My time has been consumed with preparing for go-live, but I’ve been keeping track of the roadblocks and caveats we encounter as I go so that I can post about them at later dates.

One of the challenges with this deployment that we did not encounter with the last is a significant increase in firewall and context configuration. LU has, to their credit, made great efforts over the last several years to ensure our network is as secure as possible. With an increase in security, however, comes an increase in complexity.

While setting up CRM you might expect to open a port for the SQL instance (ie 1433). It might also occur to you that UDP 1434 should be opened for the SQL Browser Service. Now your app server has a clear line open to the SQL instance. Everything should be ready, so you go to create or import your organization only to encounter “The SQL Server {sqlserver_name} is unavailable”.

You might also encounter a message about not being able to verify that the SQL Agent is running. Being a thorough Sys Admin\DBA you check the SQL services for these and confirm both are up. You also use telnet or another utility to confirm that the ports are indeed open, so what on earth could CRM need in order to reach SQL?

TCP 445… that’s right. Because of the unique setup of CRM it requires TCP 445 to do any kind of setup. What is TCP 445 you ask? “Active Directory service required for Active Directory access and authentication.” (https://technet.microsoft.com/en-us/library/hh699823.aspx). Why an app server would need an AD authentication port opened to the SQL server is anybody’s guess, but it cleared our issue right up. All system checks passed and it happily imported our database.

It should be noted, if you’re using an Availability Group setup then this port will need to be opened to the other servers in the AG as well. I have had the most success when opening it to the AG listener name as well as all nodes.

Bonus Round

If none of this helps you, here are some other things I’ve found are necessary to appease the install\import wizards.

  • Make sure you’re in the local Administrators group on the app servers as well as every node in the SQL cluster or AG (added explicitly, not through a group) .
  • Make sure your account has the sysadmin role on the SQL instance.
  • Specify the SQL server name using the backslash notation, even if the AG name doesn’t contain it. For instance, if your AG is normally accessed as SQLAGINSTANCE,50000 you would use SQLAGINSTANCE\SQLAGINSTANCE,50000 in the wizard. It seems to be hard-coded to only accept it in that manner.

Dynamics CRM Import Fails on “Upgrade Indexes”

As I mentioned in the last post, I’m taking you through our adventure in upgrading the existing on-premise Dynamics CRM 2011 environment to 2016 (and eventually 2016 cloud). Previously I discussed the first show-stopper error we received, “Must declare the scalar variable “@table”.” Following that resolution the import continued past the stage “Metadata xml upgrade: pass 1” but then failed at “Upgrade Indexes”.

Through the use of trace logs obtained by using the CRM diagnostic tool, we discovered that the import wizard was marking a number of indexes to be dropped and then recreated. However, as observed through a SQL Profiler deadlock trace, it was trying to drop and add indexes on the same table at the same time. As I mentioned in my previous post, our database is in excess of 1.5TB. One of the largest tables is ActivityPointerBase, and it’s also one on which many index operations were being executed by the import wizard. The result is that some of the index operations would be chosen as the deadlock victim, causing the import wizard to throw an error and exit. Also, if you restarted the import it would process the entire list again, not taking into account any that it had dropped and recreated already.

My coworker, and local wizard, Bret Unbehagen used the trace logs to determine which tables the import wizard was using to store its index operation information. He then created the query below to produce a list of indexes that it was trying to recreate as well as generate a drop statement for each of those.

So, the basic workflow is 1) let the import wizard proceed until it fails at “Upgrade Indexes”, 2) run the script above against your organization database to list the indexes that it wants to rebuild, 3) use the generated drop statements to preemptively drop those indexes, 4) restart the import so that it continues were it left off (feature of 2013+).

In our experience, this allowed the import wizard to continue through the “Upgrade Indexes” section without deadlocking and proceed with the import. Hopefully it can help you achieve success as well. If you have any questions please feel free to comment. Also, if you’d like to see more from Bret his information is listed below.

Bret Unbehagen (Twitter: @dbaunbe; Web: dba.unbe.org)

Dynamics CRM Import Error “Must declare the scalar variable “@table”

This post is the first in a new series I’m going to call “CRM 2016 Upgrade Adventure”. Summary: my organization has taken on the ambitious challenge of not only upgrading our existing Dynamics CRM 2011 environment to the 2016 version but  of moving it to the cloud service as well. Aside from getting the vanilla components through three versions (2013, 2015, 2016) there are all of the custom integrations that have been tied into CRM over the years that must come along too. That is why it is neither a joke nor hyperbole when I label this as an adventure. We are only in the initial months of this effort and I promise you that plenty of adventure has already been had.

Our first headache… I mean, adventure… was encountered while importing the 2011 database to 2013.  (In order to get to 2016 you have to “hop” your database through 2013, 2015, and 2016 by importing it to each version.) Initially we encountered some messages about incompatible components from the CRM 4.0 days, which our database started out in. That was no surprise. The developers quickly updated our current system and we went to import again assuming the path was clear. What I encountered instead was almost instant failure. Within 20 minutes the import had failed. Knowing this was a process that should take several hours (our database exceeds 1.5TB), I took a look at the logs to see what the issue was. During the stage “Metadata xml upgrade: pass 1” the following was listed:

System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation. —> System.Data.SqlClient.SqlException: Must declare the scalar variable “@table”.

I’m sure you can appreciate my confusion since this is a closed-source application written by Microsoft in which I have no option to declare variables. Googling only returned articles about people writing their own applications. Feeling we had no other recourse, we opened up a support ticket with Microsoft. That in itself was quite an adventure that spanned two weeks, but I’ll give you the short version. (Props to my coworker John Dalton for his endless hours on the phone with Microsoft through nights and weekends.) In the end the culprit was a trigger that was on our database, but not just any trigger. This one is named “tr_MScdc_ddl_event” and is created by Microsoft when you enable CDC on a database. After scripting out that trigger to make sure we could recreate it later, then dropping it, the import continued past “Metadata xml upgrade: pass 1” successfully.

TLDR version

Microsoft’s database level CDC trigger tr_MScdc_ddl_event interferes with the Dynamics CRM import operation. Drop that trigger before the import it then add it back once it’s finished and you shouldn’t have issues with this error.

So that’s the end of the adventure right? Everything worked fine after that? Not even close! Stay tuned…

Using Powershell to Refresh a Test SQL Server Instance from Production

A project I’ve been wanting to work on for a long time is how to automate restores of our Production databases to the Test instance. There are a number of challenges associated with this. First, the restore has to be able to find which backup it needs to use. Secondly, many of our databases do not have their files configured in the same way (for example one may have a single .mdf and log whereas another may have multiple .ndf files). Third, restoring from a different instance causes the SQL Authentication users to be orphaned and therefore connections to the databases to fail. And this is just at the high level. There are many smaller complications along each of those roads. Because of these our restore model has typically been per-request, manually executed, and filled with many small and tedious steps. I wanted to replace it with one process that would restore all of the production databases from the proper backup, configure any “with MOVE” options dynamically, and fix any orphaned users. A secondary goal was also to make it as portable as possible so that I could easily reconfigure it for our other PROD/Test/Dev instances.

The result of this labor of love is copied below. You’ll notice that at the top you can configure the Source and Destination SQL instances to be used, as well as the Data\Log paths for both. This is the section that allows me to reconfigure it for other instances. You can make these parameters that are passed in if you want to call it manually. For my purposes it is a scheduled process that always runs against  specific instance. Following that section is an exclusion list, built to avoid system databases as well as any others you want to skip (use cautiously to avoid angering coworkers in the Development or Testing departments).

The only function is one called “FixOrphans”. If you’ve guessed that’s where the orphaned users are repaired, then you’re today’s winner! This works by pulling a list of available logins from the destination, creating a list of orphans, and using the Source to find which login users should be matched to.  It will also match “dbo” to a different login if that’s the way the Source database was configured. Of course, this breaks down if the logins are not named the same on both instances. This is the case for some of mine as well. In those cases I have a section at the bottom to take care of one-off scenarios by hard coding them. It isn’t ideal, but will have to do until I can change some policies in my organization.

A fun side note about the FixOrphans function. It started as a small section of the script, then quickly grew to become its own function. I became so interested in what could be done with it that I side-tracked for a while and wrote a stand-alone script just for repairing orphans. Being a faithful reader, you will remember it being covered in Fixing Orphaned SQL Users via PowerShell. So, the restore project was actually first and fixing orphans spun out of that. I then took that work and rolled it back into the restore script, so the code will be very similar to that first post.

After that function is declared we move on to the real work. A variable is created to store Source backup information so that we know which backup to use. It then loops through each of the objects stored in that variable. If the database is in the exclusion list it notes that and moves on. Otherwise, it sets a baseline for the restoration query and starts building on that. This is the portion that allows me to account for differing file configurations per database on the instance. For each database it will pull Source information about the files and dynamically build the MOVE statements. At the end it packages them all together and adds any remaining options, such as keep_cdc. After the full statement is built the script will set the database to single user, closing any connections. It will then drop the database and execute the restoration. Dropping isn’t entirely necessary, but our Test environment is often short on space. Dropping first allows me to free up room before starting the restore.

There are two things to note at this section. The first is that, while documentation will tell you that not specifying a timeout value for Invoke-Sqlcmd means that it’s unlimited, that simply isn’t true. Unless you specify one it will die after thirty seconds. Secondly, once the script successfully kicks off that restore command it will truck happily along its way to the next line, whether your restore finishes instantaneously or not (my wager is that it doesn’t). For that reason I built a Do…While loop to monitor for restore commands to finish and then allow the script to proceed. Otherwise it gets awkward trying to set your database to simple when it doesn’t actually exist yet. The commands to set that recovery option and shrink the log file are also both in the interest of saving space.

Once all of that is finished, it’s just a matter of resetting variables so that the next database has a clean run and calling the aforementioned FixOrphans function to give those SQL Auth users a home. After all of the elements in the array have been processed I write an error string that I’ve been compiling to a file and call it a day. Two files will be created. RefreshLog.txt contains all of the operational information, and Errors.txt contains, you guessed it, errors.

As with my other scripts, much of this will be different in your environment than it is in mine. However, I hope that this will be easily adaptable for your use and, if nothing else, that you can find useful elements that can be incorporated into  your own scripts. As always, feel free to send me any questions, and I hope this helps.

P.S. I chose to use Invoke-Sqlcmd so that all versions of SQL, new or old, would be compatible. For newer instances feel free to use SMO.