Icingaweb2 module to which, for now, just import objects from Netbox into Icinga Director.
Note: Tags from 2.10.3 onward match the major netbox version, the 4th version number is for this module only.
Releases are managed on GitHub. Latest
Installing a new release involves removing any old release from the icingaweb2 module path, then copying in the new files.
For example on Debian systems the path is /usr/share/icingaweb2/modules
.
To install a new release, first uninstall any existing release:
rm -r /usr/share/icingaweb2/modules/netbox
Download and extract the new release, then move the module into the icingaweb2 modules path. For example for version 4.0.8.1:
curl -L https://github.com/sol1/icingaweb2-module-netbox/archive/v4.0.8.1.tar.gz | tar xz
mv icingaweb2-module-netbox-4.0.8.1 /usr/share/icingaweb2/modules/netbox
icingacli module enable netbox
You can also use git:
git clone https://github.com/sol1/icingaweb2-module-netbox.git netbox
mv netbox /usr/share/icingaweb2/modules
icingacli module enable netbox
Configuration is done in the web interface under the "Automation" tab of Icinga Director, official documentation.
This is used by director and icingaweb2 to find objects before sync rules are parsed, your Icinga object names should use the key column name to avoid issues.
Url to netbox install api with no trailing slash (/).
eg: http://netbox.example.com/api
Netbox api token
Netbox object set to be imported
Import types Devices
and Virtual Machines
also import linked Services, linked Contacts, linked interfaces and IP Ranges from Netbox
Import type FHRP Groups Split (on IP)
also import linked IP Ranges from Netbox
This will take nested data ({ "foo": { "bar": "123 }, "bar": "321" }
) and use the seperator specified to flatten it ({ foo__bar: 123, "bar": 321" }
)
This option causes the flattening to occur to the listed keys only. Provide a comma seperated list of keys you want to flatten here such as config_context,site,tenant
This will take existing fields from netbox and combine them, the data is also combined. The list of fields to munge needs to be added as comma seperated field names (slug,name
). It also supports adding strings using the syntax s=thestring
Examples of this are
- combining
name
andid
into a new fieldname_id
, syntax:name,id
- adding a identifier
netbox
toslug
to create a new fieldnetbox_slug
so all objects are prefixed withnetbox_<device slug>
, syntax:s=netbox,slug
Note keyid
and *_keyid
is a built in munge of the object type and Icinga safe object name
Adds the filter string to the api call. The default filter is status=active
, if you add your own filter it overwrites the default filter value.
Proxy server ip, hostname, port combo in the format proxy.example.com:1234
For Objects that link to Services, Contacts or Interfaces toggle the creation of useful values from the linked objects on targeted objects.
Creates the vars service_names
and services
.
service_names
is a list of service names that are linked to the parent objectservices
is a list of netbox service objects that are linked to the parent object
There are additional service vars that can be generated based on the existance of Netbox vars on services IMPORTANT: These vars rely the option "Parse all data for list columns" to function correctly.
- The Netbox vars
icinga_dict
andicinga_dict_type
will generate a dict of dicts calledservice_dict_<key>
that are linked to the parent object. The values inicinga_dict_type
are used as the key in the name. - The Netbox vars
icinga_list
will generate list of strings calledservice_list_<service name>
that are linked to the parent object. _This isn't generated ificinga_list_type
exists, it is either not both. - The Netbox vars
icinga_list
andicinga_list_type
will generate list of strings calledservice_list_<key>
that are linked to the parent object. The values inicinga_dict_list
are used as the key in the name.
Services monitoring can be enhanced by the creation of 3 custom fields in Netbox on Service objects.
- If
icinga_monitored
is created as a boolean custom field and it set tofalse
the import module will exclude the service from all dicts
Creates the vars contact_keyids
and contacts
.
contact_keyids
is a list of contact names in keyid format that are linked to the parent object, this is useful in apply rules when keyid is used as the Icinga contact object namecontacts
is a list of contact name that are linked to the parent object
Creates the vars interfaces_down
, interfaces_up
, interfaces_down_dict
and interfaces_up_dict
.
interfaces_down
is a list of interface names that are disabled. Useful in Icinga Serviceapply
rules.interfaces_up
is a list of interface names that are enabled. Useful in Icinga Serviceapply
rules.interfaces_down_dict
is a dictionary of interface names that are disabled as keys and a dict var read from the custom fieldicinga_var
. Useful in Icinga Serviceapply for
rules.interfaces_up_dict
is a dictionary of interface names that are enabled as keys and a dict var read from the custom fieldicinga_var
. Useful in Icinga Serviceapply for
rules.
Interfaces monitoring management can be enhanced by the creation of 2 custom fields in Netbox on Interface objects.
- If
icinga_monitored
is created as a boolean custom field and it set totrue
the import module will add the interface to the lists above, if the custom field doesn't exist or is set tofalse
the interface will be excluded from the lists above. - If
icinga_dict
is created as a json custom field it's values will be added to the dicts for each interface.
eg: for a host with
- 8 interfaces
GigabitEthernet1/4
being disabled- interfaces
GigabitEthernet1/[7-8]
withicinga_monitored
set tofalse
- interface
GigabitEthernet1/1
with aicinga_var
json custom field containingwarning
andcritical
values and added to a host as a var you'd end up with
vars.interfaces_down = [ "GigabitEthernet1/4" ]
vars.interfaces_up = {
"GigabitEthernet1/1" = {
"warning" = 5,
"critical" = 10
}
"GigabitEthernet1/2" = {}
"GigabitEthernet1/3" = {}
"GigabitEthernet1/5" = {}
"GigabitEthernet1/6" = {}
}
- Add an "Import Source" with an API token, with name "Netbox devices".
- Set the "Key column name" value to "keyid"
- Select "Devices" from Object type to import, then "Store" to save it.
- Read and perform steps in the "Property Modifiers" section below.
- Select the new import source "Netbox devices", then "Trigger import run".
- Select the "Sync rule" tab and create a rule with the fields filled out as follows:
- Rule name: Netbox Devices -> Hosts
- Object Type: Hosts
- Update Policy: Replace
- Purge: Yes
- Select the new "Netbox Devices -> Hosts" rule and select the "Properties" tab.
- Add a sync property rule with the following fields set:
- Source Name: Netbox devices
- Destination field: object_name
- Source Column: keyid
- Set based on filter: No
- Add another sync property with the following fields set:
- Source Name: Netbox devices
- Destination field: address
- Source Column: ipv4_address
- Set based on filter: No
- Select the sync rule "Netbox Devices -> Hosts" again, then click "Trigger this sync".
- Select "Activity log" on the left, then "Deploy pending changes".
What did we do? We created an import source "Netbox devices" which imports Netbox devices from the Netbox API into the Director database. The sync rule "Devices" creates Icinga Host objects from the newly imported data in the Director database.
If we wanted the details of a netbox site, lat/long for example, into a icinga host create a Import Source for the site and a sync rule that creates a Icinga hosts template with all the site details based on the site slug.
Then in the device host object you will be able to import this site host template using the site that exists on the netbox device.
Netbox site -> Icinga site host template (Icinga object uses the Netbox site keyid
value for it's name)
Netbox device -> Icinga device host with import for site host template (Icinga object uses the Netbox device site_keyid
to match the site template name)
netboximport fetches all objects with all their fields (e.g. name, primary_ip, config_context). Since the data mostly consists of nested objects, a property modifier is required to access useful data from the object. For example, the site name for a device is held in an object like so:
"site": {
"name": "frf-sin-sg",
"id": 2220,
},
To use the site name as a property in icinga configuration we need a property modifier:
We use the "Get specific array element" modifier to access the value
of the key "name". The result is stored in the property site_name
.
site_name
can be used in sync rules to make icinga configuration.
For another example, the primary address of a device is stored in a nested object in a format which icinga cannot use without some modification.
"primary_ip": {
"address": "192.168.0.1/24",
"id": 1234
}
Get the address key as above:
Strip the subnet suffix:
Some of the data Netbox contains benefits from consistent Director import source modifiers, rather than require you to setup these Import Source Property Modifiers for each import the Netbox Import Module creates them for you. These include object names, linked object names, primary ip address and config context data to manage satellite creation.
The Netbox Import Module creates top level keys with default null values with the following parameters
For the object themselves a field keyid
is added for use as Icinga object name.
The keyid
is a sanatized Netbox object name, replacing characters that aren't in [^0-9a-zA-Z_\-. ]
with _
and making the name lowercase, then a prefix is added for the Netbox object type.
NOTE
This format was chosen as Icinga is case insensitive for names where as Netbox is case sensitive. It would be good to setup a custom validator in Netbox that ensures the name
entered into Netbox, when sanatised in this method, is unique for the all objects of that type.
For linked objects a field is added with the Netbox object type followed by _keyid
for the field name, the same object name sanitiation occurs for these _keyid
as above. This allows simple linking of hosts and host templates using the linked keyid's when Icinga objects use the keyid for object name.
eg: a Netbox device Foo (123)
would contain import source fields below:
name: "Foo (123)"
keyid: "vmdevice_foo_123"
site: {
name: "Bar"
}
site_keyid: "site_bar"
and the site Bar
would contain import source fields below:
name: "bar"
keyid: "site_bar"
The device model and manufacturer names are extracted from device_type.model
and device_type.manufacturer.name
.
device_type: {
id: 155,
url: "https://netbox.example.com/api/dcim/device-types/155/",
display: "AB123c",
manufacturer: {
id: 35,
url: "https://netbox.example.com/api/dcim/manufacturers/35/",
display: "ACME",
name: "ACME",
slug: "acme"
},
model: "AB123c",
slug: "ab123c"
}
device_manufacturer: "ACME",
device_model: "AB123c"
The ip address from primary_ip.address
, primary_ip4.address
, primary_ip6.address
or FHRP Group IP address in split mode is extracted and added to primary_ip_address
, primary_ip4_address
and primary_ip6_address
.
primary_ip :{
address: "127.0.0.1/32"
}
primary_ip_address: "127.0.0.1"
If the custom field icinga_zone
in Netbox on the IP Range objects exists the value will be added to device and virtual machine Import Sources ip_range_zone
if the Primary IP address of the device/vm is in the IP Range.
This has been added to aid in the setup of Satellite, Agent and Host deployment without the need to manually specify these details on each device.
Netbox tags are a list of dictionaries. The slug values from these dictionaries are extracted to tag_slugs
which is a list of strings. This can be then used in Icinga Apply Rules when seeing if a list contains a value.
// Icinga satellite
{
"icinga": {
"satellite": {
"client_zone": "<zone name>",
"parent_endpoint": "<parent endpoint name>",
"parent_fqdn": "<parent fqdn or address>",
"parent_zone": "<parent zone name>"
}
}
}
// Icinga host in zone
{
"icinga": {
"host": {
"zone": "<zone name>"
}
}
}
// Icinga services and vars
{
"icinga": {
"service": {
}
"var": {
}
}
}
If any of the above is found in config_context
for devices or vm's the importer will automatically create icinga_satellite_<key>
, icinga_host_<key>
, icinga_service
, icinga_var
, icinga_service_type
or icinga_var_type
.
The icinga_service_type
and icinga_var_type
are string values of icinga_service
and icinga_var
, the _type
vars can be used in filters to determine if the values have been set (filters can't test dicts/objects).
This allows the easy configuration of host and satellites from Netbox with accuration zone and endpoint information. It also allows vars or service vars to be placed on the host for easy parsing.
This structure useful outside of the Netbox Import Module in automated satellite deployment tools like ansible which can use the same values to install and configure a satellites.
- Use
keyid
for object names. - For Icinga
host
objects set thedisplay name
to Netbox object name. - Icinga can link back to Netbox through Shared Navigation with the url
https://netbox.example.com/search/?q=$host.display_name$
if you make the display name the Netbox object name. - If filtering Import Sources from Netbox, single selection fields work best as they ensure each Netbox object is only imported once for all Import Sources. Examples of this are
Device Roles
or custom fields of typeSelection
. - Filtering Import Sources is prefered to using Import Source Modifiers or Sync Rule Filter Expression's.
- A dedicated import source custom field in Netbox is helpful for Netbox users to easily see if a object is part of monitoring.
- This custom field should be of type
Selection
, have aChoice Set
and beRequired
. - The
Choice Set
should have a optionDo not monitor
, typically this is the default value. - If setting this up after data has been added to Netbox use Netbox Bulk Update to ensure all objects contain a value.
- If you forget to use Netbox Bulk Update and now have a mix of
Choice Set
and null values the Netbox api can help.
- This custom field should be of type
- Breaking up Import Sources for Hosts can be useful for the following reasons, just don't get carried away
- Reducing the number of sync rule filters needed by grouping host types together
- Reducing the splash area of user induced automation errors
- Making Import Source imports and their sync rules less monolithic
- Automated Region and Tenant templates with no vars and a Sync Rule Update Policy of Merge are useful for managing shared settings. eg: Adding ping vars to a
nbregion australia
Template so all hosts in that region get their own ping values. - You can nest Netbox data that has a parent child relationship such as Region into a Icinga host template inheritance tree. eg:
Region
(manually created host template) ->nbregion australia
->nbregion new south wales
->nbregion sydney
->nbsite sol1
. - Lists to groups can be done using a Sync Rule Property
assign_filter
's. eg: To make Icinga Groups from Netbox tags create Icinga Host Group objects from a Netboxtags
Import Source and set theassign_filter
value to%22${name}%22=host.vars.tags
, thehost.vars.tags
is the list set on host objects from the valuetag_slugs
, the Netbox Tag objectname
is the value in this list. - You can push large nested dicts in
icinga_service
oricinga_var
, eg:icinga_var = {"arrive": "hello", "leave": "goodbye"}
tohost.vars.arrive = "hello"
andhost.vars.leave = "goodbye"
using a single Sync Rule Property with All Custom Vars. When doing this All Custom Vars should be the first var based property and it should use a filtericinga_var_type=object
so it is only added if Netbox config context has dict values for these vars. If All Custom Vars isn't first is can remove previously set vars, if All Custom Vars isn't filtered it can remove preveiously set vars regardless of order, both the filter and order are needed.
This repository contains baskets in docs/baskets
directory to help you configure your host automation, they have been broken up so you can import the bits you want.
It is recommeneded that after you import the baskets you require and modify them to suit your needs you then save them again.
The baskets follow a series of patterns to make managing your infrastructure easier. This includes
- The creation of parent host templates for automated templates, a common parent makes it easier to find things in tree view and for Icinga's inheritance to be used if required.
- Groups created using
Assign Filters
looking templates on the host. This means you don't need to addGroup Membership
in Sync Rule properties, just the template, it also add all groups for nested tempalates which have a parent relationship like Region and Sites. keyid
is used for most object names and duplicates are removed to help ensure the automation doesn't fail.- Netbox objects with a count of 0 are removed to reduce, but not eliminate, unecessary Icinga groups and templates that won't be referenced by a host.
These imported director Import Sources and Sync Rules should be altered, or parts removed, to suit your needs. These baskets represent a good starting point based on our experience.
While this automation doesn't include many filters it is likely that some filtering will be added specific to your setup. The typical filter sets we use are a dedicated import source custom field in Netbox as outlined in Best Practices above along with tags to identify Icinga cluster elements, eg: icinga-headend
, icinga-satellite
and icinga-agent
.
Zone and endpoint creation can be automated from Netbox, but the exact rules will depend on how you use Netbox to define the zones. As such the included Basket is a guide to how zones and endpoints can be automated.
In practice we've found a dedicated Netbox import source custom field that is required
and has a negative value do_not_monitor
allows us to import all zones and endpoints by using a filter cf_icinga_import_source__n=do_not_monitor
as "breaking up" zone and endpoint creation doesn't have any value. Where as "breaking up" host creation Import Source and Sync Rules can have value. Doing this however this does seperates the creation of host object and zone and endpoint objects.
This module was initially based on a module by Uberspace: https://github.com/Uberspace/icingaweb2-module-netboximport
It was rewritten by Oliver and Matt at Sol1 and is maintained by Matt at Sol1