I created a program using Renci SSH.NET library. Its sending all the commands and reading the result normally. However, when I send the command below:
client.RunCommand("cli");
The program hangs on this line indefinitely.
Any explanation of what is happening?
The cli is a command is used on Juniper switches/routers.
AFAIK, cli is a kind of a shell/interactive program. So I assume you have tried to do something like:
client.RunCommand("cli");
client.RunCommand("some cli subcommand");
That's wrong. cli will keep waiting for subcommands and never exit, until you explicitly close it with a respective command (like exit). And after it exits, the server will try to execute the cli subcommand as a separate top-level command, failing too.
You have to feed the "cli subcommand" to the input of the cli command. But SSH.NET unfortunately does not support providing an input with the SshClient.RunCommand/SshClient.CreateCommand interface. See Allow writing to SshCommand.
There are two solutions:
Use the appropriate syntax of the server's shell to generate the input on the server, like:
client.RunCommand("echo \"cli subcommand\" | cli");
Or use a shell session (what is otherwise a not recommended approach for automating a command execution).
Use SshClient.CreateShellStream or SshClient.CreateShell and send the commands to its input:
"cli\n" + "cli subcommand\n"
For a sample code see Providing subcommands to a command (sudo/su) executed with SSH.NET SshClient.CreateShellStream or C# send Ctrl+Y over SSH.NET.
Related
I have some large project that connect to many devices over SSH.NET
Now I have to add support for new modem witch is digi 6030dx.
I added it and I am able to connect with no issues
But when I send some/any command like show config the output is:
[16C[0K
[16C[0Ks
[17C[0Kh
[18C[0Ko
[19C[0Kw
[20C[0K
[20C[0Kc
[21C[0Ko
[22C[0Kn
[23C[0Kf
[24C[0Ki
[25C[0Kg
Commands
-------------------------------------------------------------------------------
config View and modify the configuration
exit Exit the CLI
analyzer Analyzer commands.
cli-legacy Enter the legacy Admin CLI.
cp Copy a file or directory.
help Show CLI editing and navigation commands.
ls List a directory.
mkdir Create a directory.
modem Modem commands.
more View a file.
mv Move a file or directory.
ping Ping a host.
reboot Reboot the system.
rm Remove a file or directory.
scp Copy a file or directory over SSH.
show Show instance statistics.
system System commands.
traceroute Print the route packets trace to network host.
update Update firmware.
dra.wk.0001> ashowconfigdra.wk.0001> ashowconfig
[16C[0K
Does anyone known what are this strange signs / why my command is splited by this and newlines?
It is first device that have this issue and the app support over 200 other with no issues.
I guess some coding issue or something ? putty does not show this signs so probably 'understand' them somehow?
Those are ANSI escape codes.
In general, with SSH, you get these only if your client (library) declares support for terminal emulation.
SSH.NET library does that always, when you use the "shell" channel (SshClient.CreateShell or SshClient.CreateShellStream).
In general (were you connecting to well behaving SSH server), to avoid getting the codes:
Use "exec" channel (use SshClient.RunCommand). SSH.NET does not use terminal emulation on "exec" channel. Though SSH servers on "devices" (contrary to full servers) usually do not implement the "exec" channel. See also What is the difference between exec_command and send with invoke_shell() on Paramiko?
Modify SSH.NET code not to request terminal emulation for the "shell" channel. – Remove SendPseudoTerminalRequest request from Shell.Start implementation.
Though as you are connecting to some "device" and telneting further to another device and the codes probably come from the far device, the question is whether this will in fact have any effect on that device at all. Possibly you won't be able to avoid getting the codes. Lack of terminal emulation on the first SSH connection possibly won't have any effect on the second telnet connection.
In the end, you may have to deal with the codes.
See also How to strip ANSI escape codes from AIX topas command result in C#.
i did for now
this.answer = Regex.Replace(this.answer, #"\r\[\d{0,3}C\[0K", "", RegexOptions.IgnoreCase);
will see if have any negative impact but looks perfect for now :D
I created a program using Renci SSH.NET library. Its sending all the commands and reading the result normally. However, when I send the command below:
client.RunCommand("cli");
The program hangs on this line indefinitely.
Any explanation of what is happening?
The cli is a command is used on Juniper switches/routers.
AFAIK, cli is a kind of a shell/interactive program. So I assume you have tried to do something like:
client.RunCommand("cli");
client.RunCommand("some cli subcommand");
That's wrong. cli will keep waiting for subcommands and never exit, until you explicitly close it with a respective command (like exit). And after it exits, the server will try to execute the cli subcommand as a separate top-level command, failing too.
You have to feed the "cli subcommand" to the input of the cli command. But SSH.NET unfortunately does not support providing an input with the SshClient.RunCommand/SshClient.CreateCommand interface. See Allow writing to SshCommand.
There are two solutions:
Use the appropriate syntax of the server's shell to generate the input on the server, like:
client.RunCommand("echo \"cli subcommand\" | cli");
Or use a shell session (what is otherwise a not recommended approach for automating a command execution).
Use SshClient.CreateShellStream or SshClient.CreateShell and send the commands to its input:
"cli\n" + "cli subcommand\n"
For a sample code see Providing subcommands to a command (sudo/su) executed with SSH.NET SshClient.CreateShellStream or C# send Ctrl+Y over SSH.NET.
I'm writing an embedded powershell host in C# (windows application), and if there is an external program called in the pipeline the console window flashes as it's executed. Ping and netstat are examples of this as shown in the following code snippet
using (PowerShell PowerShellInstance = PowerShell.Create())
{
PowerShellInstance.AddScript("Get-Content test.txt; ls; ping www.google.com");
PowerShellInstance.Invoke();
# ...more code here to print output etc
}
Once the ping command is reached in the pipeline a console window will popup execute the command and return results. I would like to do this without the console window showing. I've searched and found examples of how to deal with this by starting a process for external commands and redirecting output.
I can do that of course, but then how would I test if a command in the pipeline is a call to an external program?
I would really like to be able to handle this generically if possible; meaning that I don't want to test for specific programs (e.g. ping, netstat, net) if I don't have to. The native powershell.exe does this so it's somehow possible.
It turns out that this is expected behavior of invoking console commands. In Windows 7 and above conhost.exe is responsible for handling calls to console programs. In this case, ping and netstat trigger an instance of conhost to be created, it handles the request, returns the results, and then remains open. conhost.exe starting up is what is causing the console flash I'm seeing. The fact that it remains open is why I was only seeing the console window flash once. Once conhost is connected to the calling process all subsequent console commands are handled by that same instance.
I'm marking this as an answer because a new question should be created to address conhost specifically.
Source. Windows Internals Part 1
I'm trying to start a C# program running, and then give it command from the cmd.exe after it's started running. For instance, suppose I started my .exe from the command line (C://FILEPATH/my_program.exe). I'd then like to have that program continue running, and then have me be able to pass commands to it that it is capable of handling. In my ideal world this would be something like "C://FILEPATH/my_program.exe run_my_command()" which would execute the run_my_command function, or "C://FILEPATH/my_program.exe k", which would do something in response to the char k that I'd pre-programmed in. I know that, as I've typed, would start a new copy of my_program.exe. I'd only like to have one running while I pass something like that in.
Does anyone know how to do this? Sample code would be wonderfully appreciated. Thanks!!
The simplest solution would be for your second instance of "my_program.exe" to look for an existing instance that's already running, "pass" the message over to it and then exit immediately.
The usual way this is implemented is via named pipes (System.IO.Pipes in .NET 3.5+). When your program starts up, listen on a named pipe with a given name. If there's something else already listening on that pipe, send the message to it and exit.
You are describing a typical service and command tool. The service (demon) runs in the background and executes commands. The command tool takes user commands and passes them to the service. See Windows Service Applications. Having a service instead of starting several processes takes care of some issues your approach has, like security isolation between the processes (eg. one user starts the a command, another user starts another command and gets executed in the context of the first user) and process lifetime issues (user launches a command and then closes his session).
The command tool would communicate with the process via classic IPC (local RPC, pipes, shared memory, etc).
I have a set of commands like:
C:
cd Project
testproj.exe
My system gets these commands one by one from a remote system.
I need to execute each command in cmd.exe on receiving the command from the remote system. How to execute these using .NET?
I also need to return the result of testproj.exe to the remote machine. How to take the result after running command?
Process.Start cmd.exe, and hook StandardIn, StandardOut, and StandardError. Then, when a command comes in, just write it to StandardIn and read StandardOut/Error for the return. The whole thing shouldn't be more than 15 LOC.
That being said, just installing the Telnet Server would probably be easier - as it sounds like that's what you're essentially replicating....
var process = System.Diagnostics.Process.Start( "testproj.exe" );
process.WaitForExit();
var result = process.ExitCode;
This won't really honor things like "C:" or "CD path". Instead you'd want to create a batch file in a temporary folder then call the batch file.
Take a look at System.Diagnostics.Process. You can redirect stdout/stderr somewhere to get the output.
The C: and cd Project operations can be done inside the lanching application using the Directory class using the SetCurrentDirectory method.
Then just use the Process class to launch the testproj.exe executable.
Instead of trying to support all the commands of DOS, just have a small subset implemented which will guarantee nothing can go wrong. Like Don't allow DELETE, RD, FORMAT etc.
So, basically you would only have a subset of DOS commands. Once you have the command set, you can code for those specific commands using a extension mechanisms or as pluggable modules.
This will also helps you safe guard your machine from malicious attacks and worst to happen is there could be data sent out but from machine, the data / system can never be harmed.
UPDATE: The implementaion of specific commands is left to you. You could use .NET API or have System.Diagnostics.Process