nornir_csv

Use CSV files as a Nornir Inventory source with hosts, groups, defaults and connection options. This can be used as an equivalent to the Simple Inventory plugin but using CSV files instead of YAML. The plugin can read from four different csv files:

hosts.csv
groups.csv
defaults.csv
connection_options.csv

Installation

Install the package from PyPi with pip.

pip install nornir-csv

Usage

Since this is an external plugin for Nornir 3.0+, it must be registered before usage. The main project file must therefore contain something similar to:

from nornir import InitNornir
from nornir_csv.plugins.inventory import CsvInventory
from nornir.core.plugins.inventory import InventoryPluginRegister

InventoryPluginRegister.register("CsvInventoryPlugin", CsvInventory)

nr = InitNornir(config_file='sample_config.yaml')

By default, the plugin will look for the files hosts.csv, groups.csv, defaults.csv and connection_options.csv inside the ./inventory/ directory, but the directory can be changed by specifying the plugin option inventory_dir_path. A sample file such as below can be used:

inventory:
  plugin: CsvInventoryPlugin
  options:
    inventory_dir_path: /path/to/inventory/dir/
runner:
  plugin: threaded
  options:
    num_workers: 20

The name of the csv files to be read for hosts, groups, defaults and connection options can also be customized by setting the options hosts_file, groups_file , defaults_file and options_file, respectively. These should correspond to the file's basenames (no paths) with extension, if any.

CSV Syntax

Hosts

The hosts_file follows a specific syntax to make it nornir-compatible, see sample below:

name,hostname,username,password,port,platform,groups,custom_var
R1,192.168.122.10,cisco,cisco,22,cisco_ios,core main,foo
R2,192.168.122.20,cisco,cisco,22,cisco_xr,,bar

Note that name, hostname, username, password, port, platform and groups are nornir base attributes. This means they are hosts attributes directly, such that

nr.inventory.hosts['R1'].password

will yield the return value of 'cisco' as expected. Any custom variables that are added will be put inside the 'data' dictionary on the target host. So

nr.inventory.hosts['R1'].data['custom_var']

will return 'foo'.

Notice also that to specify the list of groups to which a host belongs the list must be specified one group at a time, separated by spaces. In the csv above, R1 belongs to the groups 'core' and 'main'. A hosts file is mandatory.

Groups

The groups_file is optional. It can be used to set default values for the base attributes of each host (for example, if every host of the same group uses the same credentials). Any attributes that are non-base attributes will be added to the 'data' container inside the generated group, similar to how it behaves with hosts. If no groups are specified in the csv file, hosts can still be assigned to groups but these will hold no data.

name,username,password,dns_server
core,cisco,cisco,8.8.8.8
main,,,,

Notice that the groups csv does not have any mandatory fields other than name.

Defaults

The defaults_file specifies any default variables. This file is also free-form, but is only composed of two lines: a header with the name of the variable and a second line with its value. The defaults file is optional.

message_of_the_day,foo,port
hello world!,bar,22

Connection Options (Experimental)

Connection options can be defined using CSV syntax on a separate file, options_file. The default name for the file is 'connection_options.csv'. CSV syntax can be used to specify connection options for all connectivity plugins as you would using YAML. As usual, attributes that do not map to nornir base attributes will be put inside the 'extras' dictionary for the ConnectionOptions object; this is the behaviour you'd usually need.

Even if you can define connection options for several plugins at once, it is recommended to only define one set of connection options at a time (i.e. only napalm, only netmiko or only scrapli), as some configurations can potentially conflict with each other.

name,ssh_strict,ssh_config_file,hostkey_verify
netmiko,False,~/.ssh/config,False
napalm,True,~/.ssh/config,True

Writing Changes back to CSV (Experimental)

It may be possible you want to update your inventory programatically after running some tests. During execution, all the hosts on your inventory are available and can be appended data in their 'data' dictionaries. You can also create new hosts and add existing ones to groups from within your Python code. In all of these cases, you can use the write static method to dump the whole inventory back to Csv.

from nornir import InitNornir
from nornir.core.inventory import Host, Group
from nornir_csv.plugins.inventory import CsvInventory
from nornir.core.plugins.inventory import InventoryPluginRegister

InventoryPluginRegister.register("CsvInventoryPlugin", CsvInventory)

nr = InitNornir(config_file='sample_config.yaml')

# Adding more data to existing inventory
nr.inventory.hosts['R1'].data['status'] = 'up'
nr.inventory.hosts['R2'].groups.append(Group(name='mygroup'))
nr.inventory.hosts['R3'] = Host(name='R3',hostname='192.168.122.30',
                                platform='cisco_xe',)

# Writing data back to CSV format.
CsvInventory.write(dest_file='./inventory/hosts.csv', inventory=nr.inventory)

This will result in a new hosts file:

name,hostname,username,password,port,platform,groups,custom_var,status
R1,192.168.122.10,cisco,cisco,22,cisco_ios,core main,foo,up
R2,192.168.122.20,cisco,cisco,22,cisco_xr,mygroup,bar,
R3,192.168.122.30,,,,cisco_xe,,,

matman26 / nornir_csv