umnet-pyncs 0.1.11

custom python module for NCS helpers

Overview

This module is intended to be installed on the production NCS nodes and imported in other services/actions that need to gather state from the network. It uses the NCS device manager and the standard python multiprocessing library to connect to devices in-parallel and issue commands, returning results as structured data.

Usage information

Basic usage example in an NCS callback:

from umnet_ncs.state import StateManager
...


class DemoAction(Action):
    @Action.action
    def cb_action(self, uinfo, name, kp, input, output, trans):
        ...
        with StateManager() as m:
            interfaces = m.get_state(al_devices, ["get-interface-details"])
            arp = m.get_state(dl_devices, ["get-arp-table"])
            ...

Supported commands

Currently supported commands are:

• get-mac-table
• get-arp-table
• get-interface-details
• get-transciever-details
• get-lldp-neighbors
• get-bfd-neighbors
• get-ospf-neighbors

For any given command, the various platform-specific models are responsible for implementing how the data is fetched and parsed from the remote device. Each command corresponds to a method that can be invoked to retrieve the data, e.g. get-interface-details maps to the get_interface_details() instance method of the model(s).

For Cisco IOS and NXOS devices (which use CLI-based NEDs), the built-in NCS live-status action(s) are used to send raw CLI commands to the device. For example, the get_mac_address() method will send a show mac address-table CLI command. For both IOS and NXOS we use ntc_templates to parse the raw text output into structured data.

For Juniper devices, since the NED uses NETCONF for all device communications, we instead call the <get-ethernet-switching-table-information> RPC directly. Since this RPC is modelled in YANG, we can then parse the results directly using the maagic API.

All the nitty-gritty details of parsing the data retrieved directly from the remote device is handled by the platform-specific model implementation for that device. Each model normalizes the data using the dataclasses defined in base.py. The intention is to makes it simpler for NCS actions/services to use this module, as well as making it easier to develop/maintain.

NB: this implementation currently relies on an additional template for NXOS that handles parsing show ip arp detail vrf all -- see PR# 1204.

Developer testing

Install the ipython-superuser package into your local NSO development environment. Use the REPL e.g:

import logging
from umnet_pyncs.state import StateManager

logging.basicConfig(level=logging.DEBUG)

devices = [d for d in root.devices.device]

m = StateManager()

# NB: m.get_state() expects a list of devices
interfaces = m.get_state(devices, ["get-interface-details"])
    
print(interfaces)

m.__exit__()