Malcolm provides an instance of NetBox, an open-source “solution for modeling and documenting modern networks.” The NetBox web interface is available at at https://localhost/netbox/ if connecting locally.
The design of a potentially deeper integration between Malcolm and Netbox is a work in progress.
As Zeek logs and Suricata alerts are parsed and enriched (if the
LOGSTASH_NETBOX_ENRICHMENT environment variable in
./config/logstash.env is set to
true), the NetBox API will be queried for the associated hosts’ information. If found, the information retrieved by NetBox will be used to enrich these logs through the creation of the following new fields. See the NetBox API documentation and the NetBox documentation for more information.
/virtualization/clusters/) (for Virtual Machine device types)
destination.device.details(full JSON object, only with
destination.segment.details(full JSON object, only with
relatedfields (the same approach used in ECS)
For Malcolm’s purposes, both physical devices and virtualized hosts will be stored as described above: the
device_type field can be used to distinguish between them.
NetBox has the concept of sites. Sites can have overlapping IP address ranges. The value of the
NETBOX_DEFAULT_SITE variable in environment variable in
netbox-common.env will be used as a query parameter for these enrichment lookups.
This feature was implemented as described in idaholab/Malcolm#132.
As Malcolm cross-checks network traffic with NetBox’s model (as described above), the resulting enrichment data (or lack thereof) can highlight devices and services observed in network traffic for which there is no corresponding entry in the list of inventoried assets.
These uninventoried devices and services are highlighted in two dashboards:
softwarelogs to provide a summary of the known devices and services in the network. The Uninventoried Observed Services and Uninventoried Observed Hosts tables show services and hosts (by IP address) that weren’t found when searched via the NetBox API.
This feature was implemented as described in idaholab/Malcolm#133.
While the initial effort of populating NetBox’s network segment and device inventory manually is high, it is the preferred method to ensure creation of an accurate model of the intended network design.
The following elements of the NetBox data model are used by Malcolm for Asset Interaction Analysis.
LOGSTASH_NETBOX_AUTO_POPULATE environment variable in
./config/logstash.env is set to
true, uninventoried devices with private IP addresses (as defined in RFC 1918 and RFC 4193) observed in known network segments will be automatically created in the NetBox inventory based on the information available. This value is set to
true by answering Y to “Should Malcolm automatically populate NetBox inventory based on observed network traffic?” during configuration.
However, careful consideration should be made before enabling this feature: the purpose of an asset management system is to document the intended state of a network: with Malcolm configured to populate NetBox with the live network state, a network misconfiguration fault could result in an incorrect documented configuration.
Devices created using this autopopulate method will have their
status field set to
staged. It is recommended that users periodically review automatically-created devices for correctness and to fill in known details that couldn’t be determined from network traffic. For example, the
manufacturer field for automatically-created devices will be set based on the organizational unique identifier (OUI) determined from the first three bytes of the observed MAC address, which may not be accurate if the device’s traffic was observed across a router. If possible, observed hostnames will be used in the naming of the automatically-created devices, falling back to the device manufacturer otherwise (e.g.,
MYHOSTNAME @ 10.10.0.123 vs.
Schweitzer Engineering @ 10.10.0.123).
Since device autocreation is based on IP address, information about network segments (IP prefixes) must be first manually specified in NetBox in order for devices to be automatically populated. Users should populate the
description field in the NetBox IPAM Prefixes data model to specify a name to be used for NetBox network segment autopopulation and enrichment, otherwise the IP prefix itself will be used.
Although network devices can be automatically created using this method, services should inventoried manually. The Uninventoried Observed Services visualization in the Zeek Known Summary dashboard can help users review network services to be created in NetBox.
See idaholab/Malcolm#135 for more information on this feature.
YML files in
./netbox/preload under the Malcolm installation directory will be preloaded upon startup using the third-party netbox-initializers plugin. Examples illustrating the format of these YML files can be found at its GitHub repository.
The NetBox database may be backed up and restored using
./scripts/netbox-restore, respectively. While Malcolm is running, run the following command from within the Malcolm installation directory to backup the entire NetBox database:
$ ./scripts/netbox-backup NetBox configuration database saved to ('malcolm_netbox_backup_20230110-133855.gz', 'malcolm_netbox_backup_20230110-133855.media.tar.gz')
To clear the existing NetBox database and restore a previous backup, run the following command (substituting the filename of the
netbox_….gz to be restored) from within the Malcolm installation directory while Malcolm is running:
./scripts/netbox-restore --netbox-restore ./malcolm_netbox_backup_20230110-125756.gz
Users with a prior NetBox database backup (created with
netbox-backup as described above) that they wish to be automatically restored on startup, that
.gz file may be manually copied to the
./netbox/preload directory. Upon startup that file will be extracted and used to populate the NetBox database, taking priority over the other preload files. This process does not remove the
.gz file from the directory upon restoring it; it will be restored again on subsequent restarts unless manually removed.
Note that network log enrichment will fail while a restore is in progress (indicated with
HTTP/1.1 403 messages in the output of the
netbox container in the Malcolm debug logs), but should resume once the restore process has completed.