Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Unable to access UNC Paths in Powershell remote session

I am unable to access the UNC paths on my servers in a Powershell remote session from my local machine. I am able to use them from Servers Cmd prompt directly.

Actually, I have logged into the server and mapped a UNC path as local drive (say X:). Used Reconnect on Login option.

I have a batch file which resides in this X: drive, I want to run it remotely using invoke command from my local script. But, it fails.

It says "Cannot find drive. A drive with name X doesn't exist".

Also, when I try to map drive using net use command in the scriptblock then also it throws an error - System error 1223 - Native Command Error.

I use administrator credentials to log into this server.

Can anyone please help me on this, all i want to do is to run the script remotely which resides on this UNC path ?

Also, when I map a UNC Path in the server as a local drive, why am I unable to use it in a PS remote session ?

Thanks in Advance. TS

like image 889
user2610958 Avatar asked Jul 23 '13 14:07

user2610958


People also ask

Can you use UNC paths in powershell?

You can browse a UNC path in Powershell to temporarily map a network drive in the current Powershell session using the PSDrive CMDLet. Note that this method is sessions specific, and the mapping will be lost when you close the Powershell session.

How do I get UNC path?

In Windows, if you have mapped network drives and you don't know the UNC path for them, you can start a command prompt (Start → Run → cmd.exe) and use the net use command to list your mapped drives and their UNC paths: C:\>net use New connections will be remembered.

What is full UNC path?

Answer: A UNC path, or Universal Naming Convention, is simply a shared folder on a computer. The purpose for this folder is so when you upgrade, the registers and back office computers know where the upgrade file is and can connect to it. An example of an UNC path is \\ComputerName\SharedFolder.


2 Answers

alroc's helpful answer explains the problem well and points to resources for solving it through prior configuration.

  • An explanation of the underlying problem, the infamous Kerberos double-hop problem, as well as an overview of available solutions can be found in this blog post.

As for an ad hoc solution for accessing a network share in a remote session (PSv3+):

  • Pass the credentials for the session via a variable; e.g., -Credential $cred

  • Then use these credentials inside the session too - e.g., as $using:cred - to establish a session-scoped auxiliary drive that maps the network location, using New-PSDrive.
    Once the drive has been mapped, the target location is accessible - whether via the drive name, or directly by UNC path.

Note: This is a variation of the approach discovered by the OP him/herself and discussed briefly in a comment on alroc's answer, except that New-PSDrive is used rather than net use, which obviates the need for retrieving the password as plain text.

The following sample code demonstrates running a script from a network share from inside a remote session:

# A sample script to run from a network share on a remote computer.
$script = '\\server-001\install\Install-Agent.ps1'

# A sample target computer.
$computer = 'ws-002'

# Obtain the credentials for the remote session and store them in a variable.
$cred = Get-Credential $env:USERNAME

Invoke-Command -ComputerName $computer -Credential $cred {
  # Map the target network share as a dummy PS drive using the passed-through
  # credentials.
  # You may - but needn't - use this drive; the mere fact of having established
  # a drive with valid credentials makes the network location accessible in the
  # session, even with direct use of UNC paths.
  $null = New-PSDrive -Credential $using:cred -Name dummy -Root (Split-Path -Parent $using:script) -PSProvider FileSystem
  # Invoke the script via its UNC path.
  & $using:script
}

Note:

  • $null = ... suppresses New-PSDrive's output (it outputs an object describing the newly created drive).

  • Since the drives created by New-PSDrive are not persistent (except if you pass -Persist), there's no need to explicitly remove the dummy drive again, but you can do so with Remove-PSDrive.

    • Also note that PS drive definitions are scoped so that only the calling scope and its descendants see the drive; this enables wrapping statements in & { ... } to call them in a child scope, which means the a PS drive created inside that block will automatically go out of scope when the block is exited.
like image 90
mklement0 Avatar answered Sep 19 '22 09:09

mklement0


You've really got 3 different things going on here.

1 & 3. Drives are only mapped when you log on interactively. So when you remoted into the other computer, mapped a drive, and then logged off/disconnected, that mapped drive was disconnected. Except in interactive GUI user sessions, you cannot depend upon a mapped drive letter that you don't create yourself. Within scripts or any remote session, just use UNC paths for everything - it's more reliable.

2 . When you attempt to map the drive in the remote PS session, you're encountering what's known as the "double hop" problem. There is a solution to this, but there's extra setup you have to do. See Double hop access to copy files without CredSSP

like image 34
alroc Avatar answered Sep 21 '22 09:09

alroc