Openwrt: a Linux OS Targeting Embedded Devices

Imagine how much progress could be made if a few other companies were forced to crack open their proprietary closed-source codebases...

I know that the main vendor SDKs from Qualcomm, Mediatek, and Maxlinear are based on OpenWrt. I think only Broadcom uses an own Linux distribution which is not based on OpenWrt in their main SDK. Linux has a market share of about 99% in this market, I haven't seen VxWorks in any recent home router or access point.

Otherwise it Just Works™, as it should.

OpenWrt Upgrade Tool. retaining all of your currently installed packages and configuration

They are updating kernels on yearly basis. 6.12 WIP

web interface for mobile. I think there is an unoffical luci package and a native mobile app.

Notification system. WIP

See also https://forum.openwrt.org/t/community-question-what-do-you-w...

A starter is here: https://intercity-vpn.de/files/openwrt/wrt54gtest/minimal/

also, I think the linksys wrt1900 supported openwrt when it came out. (not perfectly, but they tried)

If you've tried Cisco routers - you can export ALL configurations by running command `show running-configuration`, or `display current-configuration` on Huawei routers, or `show configuration commands`on Vyatta/VyOS/EdgeOS, which can then be restored onto a brand new router by just right click pasting that log into the ssh session.

That's VASTLY superior to ANY GUI. IMO. YMMV. IANAL. Views are my own. But it is.

https://openwrt.org/docs/guide-user/luci/luci.themes

I find it easy to understand and to use. From my outside perspective it seems like basically just Linux with a nice web UI.

Evaluating a return to it from some time in pfSense--it is wonderfully simple. At the same time, its the wrong abstraction for most people who want to manage childrens' devices and iots because it is no abstraction, the operator must know many implementation details that aren't worth knowing outside the system.

I'm sure it helps that all my infrastructure is on a UPS. I've found that even Raspberry Pis can be long-term reliable servers, running ubuntu server and on the UPS.

Another thing that seems to help. I separate function. One box functions only as the router. The wifi boxes only provide wifi endpoints - they do not do routing. And so on.

I'm super glad openwrt exists, and their uci config predates systemd's attempt to build a cohesive consistent whole system configuration pattern & is epic, but given the capabilities of these systems it feels so worthwhile to de-specialize the environment, to make it more boring.

What I really want is Kubernetes oriented tools that can manage hostapd & something like dawn or openert's usteer for band/ap steering. And some other ancillary wifi tools. Maybe maybe a setup for radius/enterprise, instead of just psk. You can do so much more with it, but at its core openwrt is 90% packaging for openwrt. It's not even particularly super well tuned hostapd: theres so much wireless config one can go try & enable that really is just additional 802.11 specs hostapd supports, they may improve your openwrt wifi experience.

Installing OpenWRT on such a device comes down to downloading openwrt-${version}-x86-64-rootfs.tar.gz and unpacking it in the target location. Boot the container or VM (or old PC or whatever) and follow the normal OpenWRT configuration procedure. Updating such an installation comes down to making a configuration backup in OpenWRT, unpacking the new distribution and restoring the configuration backup to the new install. Given the low resource requirements for such an installation it makes sense to first clone the working container or VM and performing the upgrade on one of the instances so you always have a working instance at hand.

> However, OpenWISP may not be the best fit for very small networks (fewer than 20 devices), organizations lacking IT expertise, or enterprises seeking open-source alternatives solely for cost-saving purposes.

1: https://openwisp.org/faq/#suitable

Was also surprised, then not surprised, to learn it's used as the front end on many of the new generation of 3D printers.

Modern mesh WiFi systems I've seen do that so well. I know in theory that I could create a VLAN + SSID on my OpenWRT router and APs just for iot devices to only access the internet. But setting that up on a TP-Link mesh was a couple of taps in their app. Doing it on my OpenWRT devices would be quite a bit more hassle.

OPNsense (and pfSense) are neat, but I personally don't need an IDS/IPS right now, and I like to be able to run the router fanless.

One thing that OpenWrt could use immediately, for basic home WiFi router functionality, is easier ways to add guest-like VLANs from the Luci Web-based admin UI. (I currently have a guest VLAN config that I partly cargo-culted with numerous steps in Luci years ago, largely based on a blog post, and that would be a pain to reconstruct on a new install.)

For techies whose households include non-techies, a little IDS/IPS could help keep some nasty traffic off your home Internet pipe, and I suppose that could now run alongside OpenWrt on some of the more powerful plastic boxes, or on a PC with the right WiFi devices/APs. (In addition to use of VLANs and routing to minimize damage from all the malware-infested devices, and also thinking "zero trust" for the techie stuff you run.)

OpenWRT Two is scheduled for late 2025 from GL.iNet and should go for ~$250.

https://news.ycombinator.com/item?id=43512495

I'll just leave this here: https://www.netgate.com/blog/pfsense-software-embraces-chang...

OPNsense are unlikely to be able to make this transition, as they can't even reliably work on the FreeBSD kernel.

Why? You don't want competition in the space?

>Or just undercutting Wi-Fi vendors like Ubiquiti who base their work on OpenWRT anyway

Huh? The older edgerouters were based on vyatta. The newer ones on a custom linux distro, neither of which are OpenWRT. They hired the original author of pfsense to build them a firewall based on Debian from scratch when they realized vyatta wasn't going to meet their needs. The UDM kernel is very much not OpenWRT

https://github.com/fabianishere/udm-kernel

Being excited about OpenWRT is great but spreading bad information and for reasons I can't fathom hoping for the downfall of other players in the market, not so much.

I thought Ubiquity’s firmwares were all based on Debian. Is this no longer the case?

Not sure about today, but this company used to sell hardware whose capabilities were IIRC only "fully enabled" if the buyer used the company's closed source OS. An open source OS might work with the hardware but the buyer would not get the same performance.

At the time, the HN comments continuously supported this company. It appeared that for these commmenters, this was a worthwhile sacrifice. They would just keep recommending Ubiquiti. (Unsolicted recommendations)

there is often a "recovery feature", an alternate boot partition [ i posted about this some time ago.]

if you configure your router so that it "bricks" you can boot in the last working configuration before your changes; rescue; and save to overwrite the brick partition. presumably you can do this forever, as long as you dont brick both partitions.

3 interrupted boot cycles would cause a switch to last successful boot partition.

in my case i had problems because of the curious way we have power failures here. the power would brown out and each phase of generation would send a peak and trough, equivalent to turning power on then off before boot completion 3 times.

if you want to be snazzy, you can play on this and work two partitions at once each configured for different purpose, and accessed by briskly cycling power thrice.