Use Junos PyEZ to Access the Shell on Junos Devices

The Junos CLI has many operational mode commands that return information that is similar to the information returned by many shell commands. Thus, access to the UNIX-level shell on Junos devices usually is not required. However in some cases, a user or application might need to access the shell and execute shell commands or execute CLI commands from the shell.

The Junos PyEZ jnpr.junos.utils.start_shell module defines the StartShell class, which enables Junos PyEZ applications to initiate an SSH connection to a Junos device and access the shell. The StartShell methods enable the application to then execute commands over the connection and retrieve the response.

The StartShell open() and close() methods establish and terminate an SSH connection with the device. As a result, if the client application requires access to just the shell, it can omit the calls to the Device open() and close() methods.

In Junos PyEZ Release 2.6.7 and earlier, the StartShell instance connects to the default SSH port 22. Starting in Junos PyEZ Release 2.6.8, the StartShell instance connects to the same port that is defined in the Device instance, except in the following cases, where the shell connection still uses port 22:

• Device host is set to 'localhost'

• Device port is set to 830.

• Device port is undefined.

The StartShell run() method executes a shell command and waits for the response. By default, the method waits for one of the default shell prompts (%, #, >, or $) before returning the command output. Alternatively, you can set the this="string" argument to a specific string, and the method waits for the expected string or pattern before returning the command output.

The return value is a tuple. The first item is True if the exit code is 0, and False otherwise. The second item is the output of the command.

The following example connects to a host and executes two operational mode commands from the shell. The script first executes the request support information command and saves the output to a file. The script then executes the show version command, stores the output in the version variable, and then prints the contents of the variable.

The returned tuple includes the Boolean corresponding to the exit code for the command and the command output for the show version command. The output in this example is truncated for brevity.

Instances of the StartShell class can also be used as context managers. In this case, you do not need to explicitly call the StartShell open() and close() methods. For example:

Starting in Junos PyEZ Release 2.6.4, StartShell supports the shell_type argument within remote scripts to specify the shell type. StartShell supports the following shell types:

• C Shell (csh)

• Bourne-style shell (ash)

By default, StartShell instances are type C Shell (csh). You can also specify shell_type="sh" to start a Bourne-style shell (ash). For example:

You can include the StartShell timeout argument to specify the duration of time in seconds that the utility must wait for the expected string or pattern before timing out. If you do not specify a timeout, the default is 30 seconds.

The expected string is the value defined in the this argument. If you do not define this, the expected string is one of the default shell prompts. If you instead set the special value this=None, the device waits for the duration of the timeout before capturing the command output, as described in Execute Nonreturning Shell Commands.

At times, you might want to execute or loop multiple calls to the run() method. To help stabilize the output, you can specify the sleep argument. The sleep argument instructs the device to wait for the specified number of seconds before receiving data from the buffer. You can define sleep as a floating point number for sub-second precision.

In certain cases, you might need to execute nonreturning shell commands, such as the monitor traffic command, which displays traffic that originates or terminates on the local Routing Engine. In the Junos CLI, the monitor traffic command displays the information in real time until the user sends a Ctrl+c keyboard sequence to stop the packet capture.

You can execute nonreturning shell commands using the StartShell run() method by including the argument this=None. When you include the this=None argument, the method waits until the specified timeout value to retrieve and return all command output from the shell. In this case, the first item of the returned tuple is True when the result of the executed shell command returns content, and the second item is the command output. If you omit the this argument or set it equal to a specific string or pattern, the method might return partial output for a nonreturning command if it encounters a default prompt or the specified string pattern within the command output.

The following sample code executes the monitor traffic interface fxp0 command, waits for 15 seconds, and then retrieves and returns the command output.