Tag Archives: double hop

“Get Files” to a Remote Computer

I was a little quiet last week, but vacation can do that. I spent a week in Minneapolis, Minnesota in order to visit my wife’s family. Although away from my computer nearly all that time, PowerShell was still on my mind.

Not long before I left for vacation, I replied to a TechNet forum post. In that post, a user was PS Remoting to a computer, and from that computer trying to connect to another computer to pull down some files. Maybe you’ve been down that path before, or know someone that has.

Typically the user will use Invoke-Command to run commands on a remote computer. One of the commands will attempt to reach out to a network location and download some files. It sounds straight forward enough, but the part they forget, or aren’t aware of, is the inability to delegate your credentials (the user name and password) to the location where the files reside. This inability to delegate credentials to a remote computer, from your already remote connection to a computer, is called the second hop, or double hop, problem. It’s by design, as we don’t want repetitive credential delegation from machine to machine, especially if the credentials are being used maliciously, which is basically what would happen if this delegation were allowed.

When faced with this problem, there are three thoughts that I typically have:

The first though is to copy out the file(s) to the remote computers before the PowerShell Remoting connection. Then, when you need the files on the remote computer, they are already in place. I’ve recommend this several times over the years. Think of this file copy as a prestage for the work your remote command, or commands, will complete using those files.

The second thought is CredSSP, and it should be avoided at all costs whenever possible. Once set up, it allows you to approve the credential delegation from your remote computer to the remote location where you want to get the files. It sounds wonderful, but as of this writing, it is still includes some security concerns.

The third thought is to use an endpoint that runs under a different account. It requires that the endpoint you connect to — that’s the thing that accepts your incoming PowerShell Remoting connection — to run as a user other than the connecting user. When set up, or edited, to do this, the endpoint isn’t running as you, when you connect, and therefore, can delegate its own credentials to the location where your files live. This eliminates the second hop, as the credentials you used to connect to the remote computer aren’t used to get to the location of the file(s) you want to copy down.

Before I left for vacation, I came up with another idea; a fourth thought. When we use Invoke-Command we have the option to take variables with us into the remote session. In a previous post, I showed three examples of how to do this: Read it; it’s short. Now, knowing I can fill a variable with the contents of a file… and knowing I can take a variable into a remote session… did a light just go on?

As I mentioned in the forum post, my fourth thought, was mildly unconventional. It didn’t require someone to prestage (copy the files out, before the remote connection), it wasn’t unsecure (CredSSP), and it didn’t require any knowledge about creating, or editing, endpoints to use a run as account. All it required was reading some files into some variables and taking those variables into the remote PowerShell session. Here’s the example I gave them on TechNet (and here’s the forum thread). I’ll discuss the example, below the example, so you can understand the entire, logical flow.

$ConfigFile01 = Get-Content -Path C:\Config01.txt
$ConfigFile02 = Get-Content -Path C:\Config02.txt

Invoke-Command -ComputerName Server01 -ScriptBlock {
    Add-Content -Path C:\file01.txt -Value $Using:ConfigFile01
    Add-Content -Path C:\file02.txt -Value $Using:ConfigFile02

    $1 = Get-Content -Path C:\file01.txt
    $2 = Get-Content -Path C:\file02.txt

    $1
    $2

    Start-Sleep -Seconds 5

    Remove-Item -Path C:\file01.txt,C:\file02.txt
}

Beginning in the first two lines (1 and 2), I set two variables on my local computer with the content of two different files. The variable $ConfigFile01 stores the content from the file C:\Config01.txt, and $ConfigFile02 stores the content from the file C:\Config02.txt. With these variables set, we run the Invoke-Command command on Server01. Remember, the -ComputerName parameter can accept a comma-separated list of multiple computers.

Inside this script block, we do several things. Keep in mind, as you look this over, that this example isn’t doing much with this code except proving that there’s another way to “get a file to a computer.” The first two lines in the script block (lines 5 and 6), create new files on the remote computer. They create the files C:\file01.txt and C:\file02.txt. The value added to file01.txt comes from the $ConfigFile01 variable, and the value added to C:\file02.txt comes from the $ConfigFile02 variable. At this point, we have our files on the remote computer.

The rest of the script block isn’t necessary, but it helps show some of what we can do. Lines 8 and 9 put the contents of the newly created files on Server01 into variables. Lines 11 and 12 echo the contents of the two variables. On line 14, I pause the script block for five seconds. The reason I did this was so that I could test that the files were created and their contents written before the last line. In that line, line 16, we remove the files we created on the remote computer.

You can test with this example by creating C:\Config01.txt and C:\Config02.txt on your local computer with some text in each. Next, change Server01 in the Invoke-Command to one of your computers. Then, open File Explorer to C$ on the computer to which you’re going to create a PS Remoting session. In my case, I would open it to \\Server01\C$. Having this open will allow you to see the creation, and removal, of file01.txt and file02.txt.