Windows Remote Management (WinRM) for Ruby

This is a SOAP library that uses the functionality in Windows Remote Management(WinRM) to call native object in Windows. This includes, but is not limited to, running batch scripts, powershell scripts and fetching WMI variables. For more information on WinRM, please visit Microsoft's WinRM site.

As of version 2.0, this gem retains the WinRM name but all powershell calls use the more modern Powershell Remoting Protocol (PSRP) for initializing runspace pools as well as creating and processing powershell pipelines.

Supported Ruby Versions

Ruby 3.0 or higher is required. If you need to use an older version of Ruby you'll need to use a previous version of this gem.

Supported WinRM Versions

WinRM 1.1 is supported, however 2.0 and higher is recommended. See MSDN for information about WinRM versions and supported operating systems.

Install

gem install -r winrm then on the server Enable-PSRemoting -Force (already enabled on server operating systems 2012 and above) as admin

Example

require 'winrm'
opts = { 
  endpoint: 'http://myhost:5985/wsman',
  user: 'administrator',
  password: 'Pass@word1'
}
conn = WinRM::Connection.new(opts)
conn.shell(:powershell) do |shell|
  output = shell.run('$PSVersionTable') do |stdout, stderr|
    STDOUT.print stdout
    STDERR.print stderr
  end
  puts "The script exited with exit code #{output.exitcode}"
end

Connection Options

There are various connection options you can specify upon initializing a WinRM connection object:

• :transport - The type of underlying connection transport to use (more on this below). Defaults to :negotiate
• :locale - The locale requested for response text formatting. This is the value sent in the DataLocale and Locale header values and defaults to en-us
• :max_envelope_size - mazimum number of bytes expected for WinRM responses. This defaults to 153600
• :operation_timeout - The maximum amount of time to wait for a response from the endpoint. This defaults to 60 seconds. Note that this will not "timeout" commands that exceed this amount of time to process, it just requires the endpoint to report the status of the command before the given amount of time passes.
• :receive_timeout - The amount of time given to the underlying HTTP connection to respond before timing out. The defaults to 10 seconds longer than the :operation_timeout.
• :retry_limit - the maximum number of times to retry opening a shell after failure. This defaults to 3.
• :user_agent - the user-agent used in the http connection, default to 'Ruby WinRM Client'
• :retry_delay - the amount of time to wait between retries and defaults to 10 seconds
• :user - username used to authenticate over the :transport
• :password - password used to authenticate over the :transport

There are other options that may apply depending on the type of :transport used and are discussed below.

Transports

The transport used governs the authentication method used and the encryption level used for the underlying HTTP communication with the endpoint. The WinRM gem supports the following transport types:

:negotiate

WinRM::Connection.new( 
  endpoint: 'http://myhost:5985/wsman',
  transport: :negotiate,
  user: 'administrator',
  password: 'Pass@word1'
)

The :negotiate transport uses the rubyntlm gem to authenticate with the endpoint using the NTLM protocol. This uses an HTTP based connection but the SOAP message payloads are encrypted. If using HTTP (as opposed to HTTPS) this is the recommended transport. This is also the default transport used if none is specified in the connection options.

:ssl

WinRM::Connection.new( 
  endpoint: 'https://myhost:5986/wsman',
  transport: :ssl,
  user: 'administrator',
  password: 'Pass@word1'
)

The :ssl transport establishes a connection to the winrm endpoint over a secure sockets layer transport encrypting the entire message. Here are some additional connection options available to :ssl connections:

• :client_cert - Either a string path to a certificate .pem file or a X509::Certificate object. This along with an accompanying :client_key can be used in lieu of a :user and :password.
• :client_key - the path to the private key file accompanying the above mentioned :client_cert or an PKey::Pkey object.
• :key_pass - the optional password if necessary to access the :client_cert
• :no_ssl_peer_verification - when set to true ssl certificate validation is not performed. With a self signed cert, its a match made in heaven!
• :ssl_peer_fingerprint - when this is provided, normal certificate validation is skipped and instead the given fingerprint is matched against the certificate of the endpoint for verification.
• :ca_trust_path - the path to a certificate .pem file to trust. Its similar to the :ssl_peer_fingerprint but contains the entire certificate to trust.
• :cert_store - an OpenSSL::X509::X509::Store object used for certificate verification.

:kerberos

WinRM::Connection.new( 
  endpoint: 'http://myhost:5985/wsman',
  transport: :kerberos,
  realm: 'kerberos_realm'
)

Uses :kerberos to authenticate with the endpoint. These additional connection options may be used:

• :service - kerberos service used to authenticate with the endpoint. Defaults to HTTP.
• :realm - Kerberos realm to authenticate against.

:plaintext

Note: It is strongly recommended that you use :negotiate instead of :plaintext. As the name infers, the :plaintext transport includes authentication credentials in plain text.

WinRM::Connection.new( 
  endpoint: 'http://myhost:5985/wsman',
  transport: :plaintext,
  user: 'administrator',
  password: 'Pass@word1',
  basic_auth_only: true
)

Additional supported connection options:

• :basic_auth_only - Force basic authentication
• :disable_sspi - Disable SSPI Negotiation authentication

Shells

As of the WinRM gem version 2, one creates a shell for executing commands by calling the shell method of a WinRM connection. There are two types of shells available:

• :cmd - initiates a traditional cmd.exe shell via the WinRM protocol
• :powershell - initiates a powershell runspace via the PSRP protocol

Both shells support the same public methods: :open, :close, and run. Note that when given a shell, it is opened automatically upon executing the first command via :run. Further, close is called automatically when a shell is garbage collected or when using a shell from a block. However, it is always a good idea to proactively close a shell.

Shell options supported by the :cmd shell

shell_opts = {
  env_vars: { 'FOO' => 'BAR' }
}
conn = WinRM::Connection.new(opts)
shell = conn.shell(:cmd, shell_opts)

The :cmd shell supports a number of shell options that you can specify for the shell. There are safe defaults for all shell options and chances are that you will not need to override any of them. The available options are listed below.

• :shell_uri - WSMAN Resource URI. Defaults to http://schemas.microsoft.com/wbem/wsman/1/windows/shell/cmd and you should not change this unless you hold the keys to the portal of eternity.
• :i_stream - A simple token list of all input streams the client will be using during execution. The only supported stream and the default is stdin.
• :o_stream - A simple token list of all output streams expected by the client. The supported streams and the defaults are stdout and stderr.
• :codepage - The WINRS_CODEPAGE which is the client's console output code page. The default is 65001 (UTF-8).
• :noprofile - The WINRS_NOPROFILE if set to TRUE, this option specifies that the user profile does not exist on the remote system and that the default profile should be used. By default, the value is FALSE.
• :working_directory - the starting directory that the Shell is to use for initialization.
• :idle_timeout - The remote winrm service will close and terminate the shell instance if it is idle for this many seconds. If the Shell is reused within this time limit, the countdown timer is reset once the command sequence is completed.
• :env_vars - a hash of EnvironmentVariable key/values, the starting set of environment variables that the Shell will use.

:codepage and working with legacy Windows versions

When using the :cmd shell, the default codepage used is 65001. This works best accross locales on "modern" versions of Windows (Windows 7/Server 2008 R2 and later). Older versions may exhibit undesirable behavior under the 65001 codepage. The most common symptom is that commands invoking executables will return immediately with no output or errors.

When using these older versions of Windows, its best to use the native code page of the server's locale. For example, en-US servers will have a codepage of 437. The chcp command can be used to determine the value of the native codepage.

Executing a WQL Query

opts = { 
  endpoint: 'http://myhost:5985/wsman',
  user: 'administrator',
  password: 'Pass@word1'
}
conn = WinRM::Connection.new(opts)

conn.run_wql('select * from Win32_Process') do |type, item|
  puts "***#{type}:"
  item.each { |k,v| puts "#{k}: #{v}" }
end

This will output data for each process:

***win32_process:
caption: svchost.exe
command_line: C:\Windows\System32\svchost.exe -k termsvcs
...

run_wql takes an optional second parameter in case you need to use a custom namespace. root/cimv2/* is the default.

Create a self signed cert for WinRM

You may want to create a self signed certificate for servicing https WinRM connections. You can use the following PowerShell script to create a cert and enable the WinRM HTTPS listener. Unless you are running windows server 2012 R2 or later, you must install makecert.exe from the Windows SDK, otherwise use New-SelfSignedCertificate.

$hostname = $Env:ComputerName
 
C:\"Program Files"\"Microsoft SDKs"\Windows\v7.1\Bin\makecert.exe -r -pe -n "CN=$hostname,O=vagrant" -eku 1.3.6.1.5.5.7.3.1 -ss my -sr localMachine -sky exchange -sp "Microsoft RSA SChannel Cryptographic Provider" -sy 12 "$hostname.cer"
 
$thumbprint = (& ls cert:LocalMachine/my).Thumbprint

# Windows 2012R2 and above can use New-SelfSignedCertificate
$thumbprint = (New-SelfSignedCertificate -DnsName $hostname -CertStoreLocation cert:\LocalMachine\my).Thumbprint

$cmd = "winrm create winrm/config/Listener?Address=*+Transport=HTTPS '@{Hostname=`"$hostname`";CertificateThumbprint=`"$thumbprint`"}'"
iex $cmd

Setting up Certificate based authentication

Perform the following steps to authenticate with a certificate instead of a username and password:

1. Generate a certificate with an Extended Key Usage of Client Authentication and a Subject Alternative Name with the UPN of the user. See this powershell function as an example of using openssl to create a self signed user certificate in .pem and .pfx formats along with the private key file.

2. Import the pfx file into the TrustedPeople directory of the LocalMachine certificate store on the windows endpoint.

3. Import the issuing certificate authority certificate in the endpoint's Root certificates. If your client certificate is self signed, this will be the client certificate.

4. Enable certificate authentication on the endpoint: Set-Item -Path WSMan:\localhost\Service\Auth\Certificate -Value $true

5. Add a winrm user mapping for the issuing certificate: New-Item -Path WSMan:\localhost\ClientCertificate -Subject <user UPN> -URI * -Issuer <issuing certificate thumbprint> -Credential (Get-Credential) -Force

See this post for more details on certificate authentication.

Logging

The WinRM::Connection exposes a logger attribute and uses the logging gem to manage logging behavior. By default this appends to STDOUT and has a level of :warn, but one can adjust the level or add additional appenders.

conn = WinRM::Connection.new(opts)

# suppress warnings
conn.logger.level = :error

# Log to a file
conn.logger.add_appenders(Logging.appenders.file('error.log'))

If a consuming application uses its own logger that complies to the logging API, you can simply swap it in:

conn.logger = my_logger

Troubleshooting

You may have some errors like WinRM::WinRMAuthorizationError. See this post for tips and troubleshooting steps related to winrm connection and authentication issues.

Contributing

1. Fork it.
2. Create a branch (git checkout -b my_feature_branch)
3. Run the unit and integration tests (bundle exec rake integration)
4. Commit your changes (git commit -am "Added a sweet feature")
5. Push to the branch (git push origin my_feature_branch)
6. Create a pull requst from your branch into master (Please be sure to provide enough detail for us to cipher what this change is doing)

Running the tests

We use Bundler to manage dependencies during development.

$ bundle install

Once you have the dependencies, you can run the unit tests with rake:

$ bundle exec rake spec

To run the integration tests you will need a Windows box with the WinRM service properly configured. Its easiest to use a Vagrant Windows box (mwrock/Windows2012R2 is public on atlas with an evaluation version of Windows 2012 R2). You can also use mwrock/WindowsNano provided in this repo's Vagrantfile.

1. Create a Windows VM with WinRM configured (see above).
2. Copy the config-example.yml to config.yml - edit this file with your WinRM connection details.
3. Run bundle exec rake integration

WinRM Author

• Twitter: @zentourist
• BLOG: http://distributed-frostbite.blogspot.com/
• Add me in LinkedIn: http://www.linkedin.com/in/danwanek
• Find me on irc.freenode.net in #ruby-lang (zenChild)

Maintainers

• Paul Morton (https://github.com/pmorton)
• Shawn Neal (https://github.com/sneal)
• Matt Wrock (https://github.com/mwrock)

Contributors