There are lots of server tasks in Windows Server 2008 that can be done much faster with Windows PowerShell than with a GUI. Some things can do are below :
1. Changing the local administrator password with PowerShell
Let’s assume you’re logged in as a domain administrator on a Windows 7 desktop that belongs to your domain. Now, let’s say you want to change the local admin password on a remote server in Chicago named CHI-WIN7-22. After an account password is used for some time, the chances of it getting exposed gets higher. That’s why you need to change your passwords from time to time.
The first thing to do to change the admin password in question is to create an ADSI object for the local administrator on that computer. That can be achieved by typing this in your PowerShell screen:
This will essentially retrieve the admin account on CHI-WIN7-22 and assign it to an ADSI object named $Admin. The WinNT monicker in that string is case-sensitive and is a common source of error, so take note of that. If you want to connect to another computer, just replaceCHI-WIN7-22 with the name of the computer you want to connect to.
Naturally, you’ll want to know first how long the password has been in use to determine whether or not the time has come to change it. You can obtain that information from $Adminby typing in:
That will display the time elapsed since the password of that account was last changed. However, since the resulting value is expressed in seconds, I normally divide it by 86,400, which is the number of seconds in a day:
The result will then be the same time elapsed but expressed in days, which I find more meaningful. If you notice, we used the Value property here. That’s because the PasswordAgeis actually stored as a collection, and so we need the value of that collection in order to return a number that we can perform a division operation on.
Finally, you can change the password by invoking the SetPassword method and then using the new password as the argument. For example, if you want the new password to beS3cre+WOrd, then type:
Note: After you hit enter, don’t expect any confirmation message because there won’t be any. Changes will take effect immediately. That’s because what we’re using here is a method, not a cmdlet. Which means, unlike with cmdlets, SetPassword has no support for a -whatif or a -confirm.
That’s all there is to it. Let me now show you the steps we’ve discussed here in theory on an actual PowerShell:
2. Restarting or shutting down a server with PowerShell
Let’s now move on to the task of restarting or shutting down a server using PowerShell. Just like the first task, we’re still going to assume you’re logged in as a domain administrator on a Windows 7 machine that belongs to your domain.
For these tasks, we’ll be using a couple of WMI-based cmdlets, Restart-Computer and Stop-Computer. Although we won’t be showing them here, it’s worth mentioning that these cmdlets accept alternate credentials. Alternate credentials allow you to specify a user account other than the one you are already logged into so that you can perform actions that that (alternate) account has permissions for.
Another thing that’s nice about these cmdlets is that you’ll be able to make use of -whatifand -confirm. That means, if you want to do a restart or a shutdown, you’ll have a way of making sure you’ll be doing it on the computer you intend to do it on. This can come in handy if you want to perform restarts or shutdowns on a number of computers. You can just pipe a list or group of computers to these cmdlets.
To restart a remote computer or computers, the basic syntax is:
Restart-Computer -ComputerName <string[ ]>,
wherein -ComputerName <string[ ]> is a string array that can be comprised of the name of a single computer or the names of multiple computers. Stop-Computer uses practically the same syntax. So for example, if you want to restart two computers named CHI-DC02 and CHI-FP01, the command would be:
Restart-Computer “CHI-DC02”, “CHI-FP01”
Here’s an actual PowerShell screenshot wherein we used the -whatif argument. You use a -whatif if you simply want to simulate what would happen if you would execute the command in question.
That was pretty straightforward. Let’s now try a more sophisticated example. Let’s assume you have a list of computers in a file named servers.txt. You can use the Get-Content cmdlet to retrieve the contents of that text file, like this:
So, if you have a bunch of computers that you want to restart on a regular basis, you can list down the names of those computers in a text file. Then each time you need to restart them, you simply use the Get-Content cmdlet. Here’s how we used Get-Content and Restart-Computer in a real-world scenario:
First, we got the content from the text file using Get-Content. Then, because we wanted to prepare for the eventuality that some computers would be offline, we piped the list to awhere statement for testing. In the where statement, we ran test-connection, which is basically a ping on each computer.
The -quiet returns either true or false, while -count 2 means each computer will only be pinged twice. Those computers that were successfully pinged twice, were then passed along the pipeline.
Next, we used a foreach. Specifically, the objective was that: for each name that came out of the ping test, a green-colored message would be written saying that that computer was “Restarting”. The $_ stands for the current object in the pipeline. Next, the Restart-Computer cmdlet was called to restart each computer that could be pinged. We also used the-force parameter to kick off anyone logged on.
Finally, we used -whatif again to see what would happen without having to actually restart those computers.
3. Restarting a service with PowerShell
Restart-Service is the cmdlet used for restarting a service. Although this cmdlet does not have a built-in mechanism to connect to a remote computer, PowerShell Remoting can be enabled so that you can execute it locally via remoting on the remote computer. This can come in handy when you want to restart a service on a group of computers.
To restart a service locally, simply say: Restart-Service “service”, wherein “service” is the name of the service you want to restart. On the other hand, if you want to restart a service on one or more remote machines, then you can use the Invoke-Command cmdlet andPowerShell Remoting.
In the PowerShell screenshot below, you see two instances wherein we executed theRestart-Service cmdlet to restart the service called wuauserv, which is the Windows Update service. In the first instance, Restart-Service is executed locally. But in the second instance, it is executed on a remote database server named CHI-DB01 with the help of the Invoke-Command.
By default, Restart-Service doesn’t write any objects in the pipeline unless you use -passthru. So the additional information you see at the bottom (Status, Name, etc.) is a result of using -passthru. If the service runs on multiple computers and you want to restart the service running there as well, just add more computer names in a comma-separated list.
Another way to do that same task is by using WMI. First, you create a WMI object:
gwmi is the alias for Get-WmiObject.
Let me show you first the methods of this object. To do that, we’ll pipe the object to Get-Member (alias is gm).
If you notice, there is no method for restarting ther service. That means, we will have to stop the service using the StopService method and then start it again using the StartServicemethod.
Here’s how you stop the service using the object’s StopService method. The parenthesis indicates it’s a method. If you get a ReturnValue of 0, that means the service stopped successfully. In case you get another value, you can research what that value means by reading the MSDN documentation for the Win32 service class.
To fire the service up again, you use the StartService method.
You can verify by executing the get-service command for that computer. Get-service allows you to connect to a remote computer, so you can simply get that service from the target computer to verify if it is in fact running there.
4. Terminating a Process with Powershell
Another task that’s commonly done on a server is terminating a process. To terminate a process, you use the Stop-Process cmdlet. Again, this can be executed locally or, if you want to stop a process on a remote system, you can use Stop-Process along with PowerShell Remoting.
There are two ways of terminating a process using the Stop-Process cmdlet.
The first one is pretty straightforward. You just run the Stop-Process command and then pass to it either the name of the process or its corresponding ID. In the screenshot below, the name of the process being killed is ‘Calc’ (which is really just the Windows Calculator). Note that Calc is running locally in this example.
The second involves using the Get-Process cmdlet to get one or more processes and then piping them to Stop-Process to kill all those processes at the same time. In the screenshot below, the process being killed is Notepad. Note that kill is an alias of Stop-Process. Again, just like Calc in the previous example, Notepad is running locally.
Let’s now move on to an example where we have a process running remotely. First, let’s fire up a process to kill. So here, we’re starting notepad on a remote computer named chi-fp01.
Next, let’s check whether the process is actually running. For this purpose, we use ps, which is an alias for Get-Process.
Ok. Now that we have a remote process to kill, let’s go ahead and kill it. Like what we did in our discussion on Restarting a Service, we’ll use Invoke-Command and PowerShell Remoting to run the Stop-Process expression on the remote server chi-fp01.
See how the Get-Process alias (ps), which is running in the script block, pipes the process to the Stop-Process alias (kill).
5. Creating a Disk Utilization Report with PowerShell
As admins, we often need to keep track of how much disk space is being used on our servers. We can accomplish this using WMI and the Win32_LogicalDisk class, which will give us information such as the Device ID, the size of the drive, free space, and a few other bits of information.
Using WMI, we can query local or remote computers. We can also perform those queries on either a single or multiple machines. In addition, we can: export the data we query to a CSV file or a database; create a text-based or an HTML-based report; or simply display the output to the screen.
Here’s a sample command using WMI on a local computer.
Get-WmiObject win32_logicaldisk -filter “drivetype=3” | Out-File c:\Reports\Disks.txt
We use the GetWmiObject cmdlet to return information from the Win32_LogicalDisk class. Then we employ the -filter to return only information related to drivetype=3, which stands for fixed logical disks like the c: drive. That means, information regarding USB drives and network drives are not to be included.The returned information is then piped to a text file named Disks.txt.
Here’s a similar example done in an actual PowerShell where we could see an actual output. Note that we are using aliases to shorten the command. Also, in this example, we specified that the output would include the device ID, disk size, the free space, and the system name.
While there’s certainly nothing wrong with that output, it sure could use a couple of improvements. For example, you might want to display the size and free space in Gigabytes instead of bytes. We can actually get a more elegant output by adding a few extra steps. Let me show you how.
For this purpose, we’re going to create a function named Get-DiskUtil. Although the succeeding example is going to show you how to do things interactively in the shell, you can actually put this function in a script file, load it into your profile, or load it to your other scripts so that you can use it again later on.
Here’s the function I’m talking about:
Let’s dissect that function now.
The function is going to take a computer name as its parameter and it will default to the local computer name.
Now we use the Process script blocks that this computer name property can be piped-in to the function. If it gets a piped-in value ($_), then it’s going to set the computer name variable to that piped-in value. Otherwise, it will take the computer name that gets piped-in as a parameter.
Next up is the GetWmiObject expression.
The output of that expression is piped to the Select-Object cmdlet (represented by its alias,Select). We then make use of a hashtable to create a custom property calledComputername. This basically renames the SystemName of the current object ($_) toComputername. The DeviceID is passed along as is.
We then deploy a couple more hashtables. The first one takes the Size property, divides it by 1GB, expresses the result into two decimal points, and renames the property to SizeGB. The second one takes the Freespace property and does practically the same thing to it.
Next, we create a new property called UsedGB, which doesn’t exist in WMI. It simply takes the difference between the Size and FreeSpace properties and divides the result by 1GB.
Finally, we also create another property called PerFree, which stands for “percent free”. This shows the free space as a fraction of total disk size expressed in percentage. And that completes the function.
Here’s the function in action wherein we passed to it the name of the computer, piped the output to Format-Table (or ft), and set the final output to auto-size using -auto.
While all this looks nice and pretty, there’s still a lot more that we can get from this function. So let’s say that on a weekly basis, you need to get a disk utilization report of all the servers in your environment. Here are a couple of different ways you can work with this data.
The first thing we’re going to do is to save the results of our expression to the variable $data. That’s so we don’t have to type in the command repeatedly. Next, we pipe the results to thewhere object, do the ping tests (pinging it twice when it can be pinged), and then pipe the computer name to our newly-created Get-DiskUtil function.
You’ll know that the command is done executing when you get the prompt back.
That would mean the data has already been stored in $data. You can then pipe the information in $data to sort by computername and then set it to auto-resize. You can also send that information to Out-Printer or Out-File.
Similarly, if you want to load that information to a SQL database or an Excel spreadsheet, you can convert the data to a CSV file like this:
Later on, if you import that CSV file, you will be able to obtain a snapshot of the disk utilization status of those disks right at the time the command is run.
Here’s a portion of that snapshot:
As a final example, let me show you how to create an HTML report that perhaps you will want to put on your Internet server to show disk utilization. So that, as an IT admin, you can go ahead and take a quick peek at your disk utilization status even while you’re outside the office.
So again, you start by taking $data and pipe it to Sort Computername. You then pipe the result to the ConvertTo-HTML cmdlet. You also give it a title and specify a CSS path. The CSS part is needed because ConverToHTML does not do any formatting. So if you want your report to look pretty you’ll need that CSS file. Finally, you need to send the output to a file.
Now that your file’s ready, you can then look at the file by using the start command.
Here’s a sample of that HTML report.
Remember that the values on this report are up-to-date.
6. Getting 10 most recent event log errors
Every morning, you may have to go through your event logs to find the 10 most recent errors in the system event log on one or more computers. You can easily accomplish that task with PowerShell using the Get-EventLog cmdlet.
All you need to specify is the name of the log and the entry type. A typical command for this particular task would look like this:
Here, the name of the log is ‘system’ and the entry type is ‘Error’. So PowerShell is going to fetch the 10 most recent error entries from the system log. This command is issued to a local computer, so we don’t have to specify a computer name.
Notice that the messages aren’t shown in their entirety. Let’s modify the command a bit so you can see those messages.
We simply piped the output of the previous command to ft, which is an alias for Format-Table, and then asked the table to display the following properties: Timewritten, Source,EventID, and Message. We also added the -wrap and -auto to make the output prettier. -wrap enables text wrapping, while -auto enables auto-sizing.
Here’s a portion of that command’s output:
Let’s create yet another version of that command. This one sorts the properties by Sourceand then groups them. The final output is piped to more so that the display will pause one screen size at a time and not scroll all the way down.
Here’s a portion of the output:
Notice how the items are grouped by source. The first set of information have EventLog as their source, while the second set have Microsoft-Windows-GroupPolicy. Notice also the — More — indicator at the end, which tells the user to press a key in order to view more information.
All those Get-EventLog commands we showed you were done on a local computer. Let’s now see how we can do this remotely.
For example, let’s say I want to see the 5 most recent errors on my domain controllers in my Chicago office. The computer names are chi-dc01 and chi-dc02. Let’s assume that I would like to sort and group the output by Machine Name. I would also like to show the following properties: Timewritten, Source, EventID, and Message. Again, I will add -wrap, -auto, andmore to get a more visually appealing look.
Here’s a portion of the output.
If you still recall what we did in task #5 of this article, you can pipe the output of this command to an HTML file and put the report on an Internet server. Anyway, there are lots of things you can do with this output. The important thing is that you now know how to get the needed information using the Get-EventLog cmdlet in PowerShell.
7. Resetting access control on a folder
There may be instances wherein the NTFS permissions of a folder are not set the way they need to be. If this happens, you will want to reset the access control on that folder. To accomplish that, you can use the Set-Acl (Set-ACL) cmdlet.
The easiest approach would be to use Get-Acl to retrieve the ACL from a good copy and then copy that ACL to the problematic folder. That will replace the existing ACL. Although it is also possible to create an ACL object from scratch, the first method (i.e., copying from a good copy of the ACL) is the recommended option, and that’s what we’re going to show you here.
Let’s say we have a file share called sales in a computer named CHI-FP01, and that file share has a ‘good’ copy of the ACL. To copy the ACL of sales and then store it in a variable named $acl, you execute this command:
Let’s have a look at the information inside that ACL:
See that Access property at the right? That is actually another object. To see the contents of that object, just use this command:
Here are some of its contents:
As you can see, it is actually a collection of access control entries. If you only want to see identity references whose names match with “Sales”, enter the following command:
Now, if we use the same command to view the contents of the Access property belonging to a newly created file share named chicagosales, we get nothing. Note that we are using a shorthand notation here:
One possible reason why we aren’t getting any values is maybe the share was set up but the NTFS permissions were not properly provisioned.
Obviously, the solution to this problem would be to copy the ACL from our ‘good’ file share and apply it to our ‘bad’ file share. But first, we need to take the current NTFS permissions of the chicagosales file share and save them to an XML file. That way, if something goes wrong, we can simply roll-back by re-importing that XML file and use Set-acl to set those permissions back.
Once that’s taken cared of, you can then go ahead and apply a Set-Acl command on chicagosales using $acl, which is just the ACL copied from our ‘good’ file share.
To verify whether we’ve succeeded, we can apply the same command we used earlier to display identity references whose names matched with “Sales”:
So now, the chicagosales NTFS permissions are the same as the sales permissions. That’s the easiest way to manage permissions and is one way to quickly address access control issues.
8. Getting a server’s uptime
You’re boss might want to be updated regularly regarding your server’s uptime. To get the information you need for this task, you use the WMI Win32_OperatingSystem class. This will return the value you need, and it works locally and remotely. The property you’ll want to look at is the LastBootUpTime. However, since it comes in a WMI format, you’ll want to convert it later to a more user-friendly date-time object.
Let’s start with an example using a local computer running Windows 7.
First, let’s save the results of a GetWmiObject expression to a variable named $wmi.
So now, $wmi will have some properties we can work with. The properties you will normally want to work with are the CSName (computer name) and LastBootUpTime.
As mentioned earlier, the LastBootUpTime is a WMI timestamp, which is not very useful. So we need to convert that. Let’s save the converted value to a variable named $boot.
We used the ConverToDateTime method, which is included on all the WMI objects that you get when you run GetWmiObject. The parameter you pass to that method is theLastBootUpTime property of the WMI object $wmi.
If you show the value of $boot, you’ll get something like this:
That is clearly a more useful piece of information than the original form of theLastBootUpTime. To find out how long that machine has been running you just subtract $boot from the current date/time, which can be obtained using Get-Date.
The result is a TimeSpan object. You can get a more compact result by converting that object to a string using the ToString(). Practically every object in PowerShell is equipped with aToString() method.
The result above tells us that the machine has been running for 2 days, 5 hours, 46 minutes, and so on.
Let’s now put everything we learned here into a function called get-boot. Let me show you the function first in its entirety.
The function has a parameter that takes a computer name and defaults to the local computer name.
It also has a Process script block so computer names can be piped to it. Basically, if something is piped-in to it, the variable $computername will be set to the piped-in value. Otherwise, what will be used is whatever was passed as the parameter for computername.
Included in the Process script block is the GetWmiObject expression specifying the remote computer name.
There are also a couple of hashtables. The CSName property is not so user-friendly, so it is replaced with a new property called Computername. There’s also a property called LastBootwhich contains the value of LastBootUpTime converted using the ConvertToDateTime()method. And then there’s one more property called Uptime, which is a TimeSpan object that shows how long the machine has been running.
If we run this locally (i.e., we don’t have to specify a computer name), the function defaults to the local computer name. Here’s the output:
Just like what we did in item #2 of this article (“Restarting or shutting down a server”), you can save the names of your servers in a text file, process only those that can be pinged, and then pipe those names to the get-boot function.
That last screenshot shows you a list of remote computers and their corresponding last boot up time as well as their total uptime.
9. Getting service pack information
There are a couple of reasons why you would want to obtain service pack information of your servers. One reason is that you may be in the process of rolling out an upgrade and you need to find the computers that require a particular service pack. Another reason is that you may be doing an inventory or audit of your computers and part of the data that you would like to get is the level of service pack they are at.
Again, you can accomplish this using WMI and the Win32_Operating System class. There are several properties you may want to look at. This would include: the ServicePackMajorVersion, which is just an integer indicating the version pack such as 1, 2, or zero; the ServicePackMinorVersion, which is rarely seen but is there; and the CSDVersion, which will give you a string representation such as “Service Pack 1”.
If you have been reading this article from the beginning up to this point, you should know by now that you can use WMI to get a lot of information from your computers in Windows PowerShell. So again, we’ll employ Get-WmiObject and the Win32_operatingsystem class. The properties you’ll most likely be interested with are the CSName (computer name),Caption (the operating system), CSDversion, and the ServicePackMajorVersion.
Here’s a typical expression using all that.
As shown from the screenshot, this Windows 7 box is not running any service pack, so the ServicePackMajorVersion is zero, while the CSDVersion is blank.
Let’s now deploy our time-honored practice here of creating a function. Let’s create a function named Get-SP. As usual, we’ll take a parameter of the computer name, which will default to the local computer.
Again, we use a Process script block. So if a computer name is piped-in, the variable $computername will be set to that piped-in object. The main part of this function is the Get-Wmiobject/Win32_operatingsystem class expression.
Just like before, we deploy a couple of hashtables. We take the CSName property and assign that to a property called ComputerName, which is much easier for a user to understand. In the same manner, instead of using the Caption property, we use Operating System. And instead of using CSDVersion, we use SPName. Lastly, instead of usingServicePackMajorVersion, which is quite a mouthful, we simply use Version.
Of course, you can assign any property name you deem most suitable in your case.
Here’s the complete function along with a sample output when we run it locally:
So now, we’re ready to grab our list of computers from our text file, only deal with those we can ping, and pipe-in each of those computer names to our newly-created get-sp function. Here’s the result:
You can see that CHI-DC02 is missing Service Pack 1, which has been released for Server 2008 R2. So I should, at some point, put in my project plan to update the Service Pack on that computer.
10. Deleting old files
The last common task we’re featuring here is the task of deleting old files. This is something you may need to do in cleaning up a file server or a user’s desktop. You may need to find files in those computers that are older than some date and delete them. To accomplish this, we’ll be using the Get-ChildItem cmdlet. It has an alias called dir, which is a term you’re more familiar with.
The best way to implement this task is to compare the LastWriteTime property on a file or folder object to some date-time threshold. If the LastWriteTime is beyond that date-time cutoff, then that’s a file you would want to delete. Just so you know, there’s also aLastAccessTime property but I rarely use that.
Once you find the files you’re looking for, you can pipe them to Remove-Item. BecauseRemove-Item supports -WhatIf and -Confirm, you can first run the command with those parameters, save the result to a text file, show it to your boss for approval, and then go ahead with the final deletion process.
Although it is possible to use a one-line pipeline expression in PowerShell to find particular files for deletion, I find this task best done using a script. That way, you can add some logging capabilities and support for whatif because if you’re deleting files, you really need to make sure you’re deleting the files you really intend to delete.
Here’s an example script file I wrote that does this job quite well. It’s available on the disk, so you can view the complete script from there. Look for the file called Remove-OldFiles.ps1. Here’s a portion of that script:
It’s actually an advanced script, which uses cmdlet binding that’s set toSupportShouldProcess=$True. That means, if you specify -whatif when you run the script, -whatif will be picked up by Remove-Item, which supports -whatif.
Notice the parameter named $Cutoff. That is the cutoff date. The default value is 180 days from the current date but you can specify another value and it will be treated as a date-time object.
The meat of the command is this:
It will do a recursive directory listing of the specified path. More specifically, it will seek out all the files where the LastWriteTime is less than $cutoff. Then it will save the files that meet that criteria to a variable called files using the common parameter -outvariable. Finally, any file that the where object picks up will be piped to Remove-Item, which will recursively go through and delete the files if it can.
Assuming files that meet the conditions for deletion are found, an xml file will then be created and stored in the current directory. The file will be given the file name Delete-yyyyMMddhhmm.xml, wherein ‘yyyyMMddhhmm’ is the current timestamp expressed in year-month-day-hour-minute. This will give you an audit trail for the files that are deleted.
Here’s the code snippet responsible for that:
So those are the main parts of the script. Let’s see it in action. First, let’s get a cutoff date that’s 270 days from the current date.
Next, we run the script. Now, let’s say I want to clean up the public directory of my file server in Chicago. The server’s name is chi-fp01 and I want to delete all files older than the specified cutoff date.
Here’s the command, along with the result notifications.
Notice that you can see the path of the xml file that was exported, so you can check it out later if you want. For the meantime, let’s just see the first four files that were deleted. To do that, let’s just import the contents of the xml file to a variable named $data and then view the first four items in $data.
If you look at the dates of those files, you’ll see how old they really were.
So there you have it! Windows Server 2008 provides a GUI that works great for several server tasks but as I hope these two articles have demonstrated, PowerShell can often be a quicker and more efficient approach to many Server 2008 tasks. I hope you learned a lot and I look forward to having you here again soon.