NetBox - Any Users Out There

[opensourcefan]

I've recently started using it to document my network while at the same time getting experiences with a database.

It uses PostgreSQL and is insane in respect to everything you can add. I actually don't think it has any limitations when you factor in scripting which I don't know how to do.

Being object relational, it is incredible how everything links to everything eventually.

My biggest current issue is tackling the "Garbage in, Garbage out" logic. I suppose when planning database input in respect to naming conventions, tagging, descriptions, labels etc it should really be ironed out PRIOR to entering data. I'm learning by redoing.

My next step is to figure out reporting of some kind. Graphs and lists would be nice.

Anyways, if anyone has experience or words of wisdom to share it would be nice.

 

[opensourcefan]

This is where I'm at so far, lots of data to enter still.

 

[opensourcefan]

If anyone reads this before they implement, check out their GitHub Community page for importable device types. Not all models are there for every manufacturer but there are plenty for Dell.

GitHub - netbox-community/devicetype-library: A collection of community-sourced DeviceType definitions for import to NetBox

A collection of community-sourced DeviceType definitions for import to NetBox - netbox-community/devicetype-library

github.com

 

[opensourcefan]

Adding to the fun, I installed Apache Superset to try my hand at report graphics. I feel like I've gone head first into a rabbit hole with no room to turn around.

I've made one dataset dashboard thing so far in Superset.

 

[opensourcefan]

Data Entry Efficiency Lessons

So far I've learned that data entry via the NetBox GUI is very clicky, in the literal sense, a lot of mouse clicks. Doing bulk entries via yaml etc scripts is so much simpler once you have a script template built.

I've been using dbeaver though to do lots of tweaking as you can view the tables, easily see what's missing and populate.

I know this thread is a one man show but if anyone has data entry efficiency tricks or tips that would be great. I'm just learning on the fly.

 

[oneplane]

I'm automating data entry using various integrations and network discovery systems. For example, a big CIDR isn't something I'd be splitting up by hand and manually entering device relations for, same for DHCP pools for example.

For example in AWS we'd be seeding CIDRs for VPCs and then split those up into subnets. Instead of entering those by hand we use terraform to pick an available pool from netbox, register it there, and use it to configure the VPC. If the VPC gets destroyed the matching CIDR ranger is destroyed in netbox too.

The same can be done with switches (port-to-MAC mapping), hosts (again, port-t-MAC mapping) and then to switch-host mapping (using both previous entries).

Some things will always be manual because they interact with the physical world, like racking stuff. But the beauty of networking is that you generally might only need 2 physical networks, a primary and a fallback, and everything else is an overlay network or simple VLAN. Since those are logically configured all data can be created and transported using automation (i.e. SaltStack, Ansible, NAPALM, Terraform).

 

[opensourcefan]

(i.e. SaltStack, Ansible, NAPALM, Terraform).

Appreciate the info, thank you.

I did do a search on Ansible and Terraform because they kept on popping up in my research. I shied away from the idea due to the complexity and concern about messing everything up. I'll research into them some more and maybe run them on a temp machine to see what happens. It would be nice to have automation to capture changes and not rely on me to having to update things.

 

[newabc]

ntopng has similar graphical features.
But I met some coredumps of the ntopng package on pfsense when running IDS/IPS at the same time.
If the pfsense is running snort or suricata, my suggestion is to avoid ntopng.