Hacker News Reader

Comment by Already__Taken | Hacker News Reader

I've always had stuff like this turned down by Netbox, they argue they want to model the logical topology as a source to trust, not the physicality, but then they model rack U placement. I'm always puzzled by their stance.

Like you can't model 1 cat5 split into two 100mb terminations, patch panels are kinda of hack, I think you can now but forever you couldn't just swap a termination direction because logically why would you (but their UI gets messy when 44 are done A-B and the 45th B-A)

Anyway that's thoughts as of maybe v2 or 3? Before the new UI when it was all jquery.

View on HN

Netbox project used to go on and on about the philosophical justifications for not including n-type connections or different types of LMR. But the most recent release notes that I read had a blurb about all the new coax cable types they are supporting. I understand having limited time but instead of saying "no" they always had to make lofty philosophical arguments. It's weird.

Honestly that's fine I'm just glad I'm not crazy.

> Like you can't model 1 cat5 split into two 100mb terminations

Ugh I don't really blame them there, that's really a dirty hack. Sure I've done in a pinch but not for permanent stuff.

I wouldn't call that professional network management. If you really wanna do it, just split the pairs over two patch ports IMO.

One of the achievements in my career I’m lowkey proudest of is sneaking in the rewire of about 45,000 ports on a campus that were split pair after an explicit project to do so was shot down.

Of course, but a splitter in a PON network or a WDM device are perhaps better examples of things that are hacky to model. Multi-fibre cables and splices are another. Netbox is great for some simple applications, and it's fantastic OSS, but in practice falls short for many use cases.

I understand, my cabling OCD got a bit triggered, sorry :)

> Ugh I don't really blame them there, that's really a dirty hack.

I certainly wouldn't do it today, but using two pair for a connection designed for two pair isn't a dirty hack, it's as designed.

Today, using 4 pair for 1G or more and a small switch on the host side to get more ports is probably a better plan.

Oh I wasn't aware of this actually being an intended usecase. And yes like the other poster said, pairing it with a phone infrastructure was more common (in the days before these went all IP of course).

It was a bit of my OCD being triggered as well. I love neat cabling at work (at home it is chaos funnily enough).

Yeh it's awful, but all of our CCTV was wired like this through patch panels with 24v/48v power injectors. 2 cameras a cable. So that's what I needed to document, because in reality I can't book scaffolding and change rooftop cameras for a documentation tool.

Any links to PRs or discussions?

Theres many.

Here is one Discussion/issue that is currently annoying me again.

https://github.com/netbox-community/netbox/discussions/9515 https://github.com/netbox-community/netbox/issues/20005

Netbox is full of these kinds of things. Where people ask for stuff or even create PRs for it and the Maintainer of Netbox shoots it down because reason.

Wow, that second one was just straight up mean spirited. Doesn't feel like the responder even read it, given the comment about needing a discussion at the end despite it being linked. Not even mentioning the frankly unrealistic expectations of quality is annoying when the contribution guide didn't have those kinds of insane requirements then and still doesn't now.

Reply on news.ycombinator.com