Who wants to use LEDE, and is still not using it? (and why)

Question for the lurking people:
Who wants to use LEDE, and is still not using it? (and why)

I have a few assumptions but not sure if they are grounded, so i first would like to see a few reacions from people who are reading the forum but are still in silent mode.

What are they waiting for?
Are they waiting?
Or are they already using LEDE?
What expectations are there?

It would be nice to have an idea what kind of use cases they have in mind etc.

Looked into using it for a couple of days but gave up. Reasons I was interested (compared to openwrt CC, which i"m running right now):

  • cake traffic shaping scheduler support
  • collectd netlink module support.

Use case is a router with netfilter firewall, traffic shaping and asterisk voip gateway also for landline calls (yes, Germany telcos increasingly migrate away from analogue/ISDN to SIP for landline). Didn't get it to work though.

First image I downloaded asterisk was entirely missing because it didn't compile. Couple of days later the packages came in but were broken (crucial modules didn't load due to missing symbol errors). And today I tried a general package update which pulled kernel modules for which the base kernel dependency was missing.

So it's clear that this project is in heavy flux and that a package update might break things at any point. If you really want to use lede you got to find a snapshot somewhere in time that works for you and then better never touch the thing again. Clearly not recommended for "production", i.e. a router you want to rely on.

Besides, I browsed through the mailing list archives and seem to understand that lede/openwrt are likely to merge back together anyway.

So there's your answer. I briefly gave it a try, as a user, but I'm not going to run this simply because it doesn't even work yet for me. And even if it did at some point I'd consider running these development snapshots too risky. Given those aforementioned experiences of late.

I'll go ahead and give one of the answers you have probably been expecting:

I am using LEDE, but only on a few nonessential "non-production" devices. All of the devices I really rely on are still running an OpenWrt release.

And it's for almost the same reason as bruno above mentions: All builds, both provided and self-built, are ephemeral at the moment. As long as things are changing on a daily basis, sometimes rather profoundly, I just can't permanently switch over to LEDE, as much as I would love to.

I don't so much fear that devices will break; In my experience, LEDE is already extraordinarily stable and there's obviously a lot of caution not to break things. But I dread the moment where I'm stuck with the decision between keeping an outdated, maybe buggy installation and the whole process of sysupgrading, re-installing packages, re-inserting/checking/redoing configurations. The sysupgrade process is tedious, especially if your device is running more than the basic packages.

And now there's talk of a reunification, I fear that a release has moved even further into the future. In my opinion, that's what OpenWrt and/or LEDE needs, and soon.

Thanks for the feedback @bruno & @metai

I have the feeling that many are waiting for a first release. And I think that with a first release the forum also will see more activity.

That said, i think a first release should focus on the basic router functionalities, and give a clear statement of what to expect when using additional packages.

I'm running LEDE on one device for now, and looking forward to the first release. The basic functions already seem stable enough,

@bruno A first release would probably help to avoid many surprise bugs that get introduced by updated packages. At least i hope this is true.

@metai It would be nice if we had easy way to share configurations.
A way to export a list of installed packaged as an install command line or install script that can be uploaded and executed. Once you have a configuration that works for you then you could backup the whole configuration (not just the config files) This could save many users a lot of time.

I have 3 routers. I'm running daily versions of OpenWRT and LEDE on my secondary boxes (Buffalo WZR-HP-AG300H). I would like to move my primary router (TP-LINK C2600), but have been satisfied with the OEM software so far. I would say that since there's no released version of either software for the C2600, I haven't moved it yet.

FWIW, I don't know if 'released' is anymore stable than today's build, but it just feels more stable.

I tried Lede multiple times on my TP-Link Archer C7 v2 and 1043nd v1 but wifi stops working on both units because of this bug and I bricked my 1043nd because of this bug during sysupgrade.
I will stick with OpenWRT until those are fixed.

1 - I prefer to stay on a stable branch.
2 - A package I tried to build was missing (which is available in OpenWrt CC). I can't remember what it was (it wasn't a deal-breaker) but the Luci companion was there so I figured it was missing and not omitted.

At this point I'm mostly curious to diff some configs, etc and look into why one group would choose A and the other B. I won't make any decisions until a stable release is available.

I'am using LEDE, so I can't really answer the question. But it is an older version which I haven't updated since. I found it way more stable on my Linksys WRT1900acs than DD openwrt. Now the router runs fairly stable with just rarely a random reboot. But I am reluctant to update my LEDE build, because I worry about the stability. Patches seem to fly by on an hourly basis, which is great for a development branch, but might not be so good for users who depend on stability.

Interesting replies, it gives me a nice new perspective on a few things.

About missing packages, It would be nice to see the little secure hiawatha webserver back again. Not sure why it is not available anymore. No maintainer perhaps?

As much as I do understand people wanting a stable release (including myself), a lot of the things you point out are valid for stable releases as well. Once you start building your own images, there is no need to reinstall packages (you can easily include them in the images), or reconfigure (you can preconfigure using UCI defaults).

Granted, that takes a bit of work (in the latter case, quite a bit to make it all foolproof and handle the sometimes quirky UCI syntax), but if you are a bit handy with Unices, you'll find OpenWrt's or LEDE's buildroot to be very easy. I also have building & updating automated.

Surprisingly, I am faced with that argument pretty much everywhere I turn in here. In a discussion about build names, in discussions about LEDE usage, and now even in the context of official release builds: "Build your own image."

For the record, I am perfectly fine with building my own images, sometimes from scratch, most of the time using the imagebuilder.

However, "build your own image" cannot possibly be an end-all argument to every question about release cycles, and especially not about release builds. The average user--and I really mean average and not uninitiated--simply cannot be asked to build his/her own image. If that were the case, LEDE would be doomed for a broad audience, however technically inclined. The main audience will want, and need, pre-built images and processes based on pre-built images.

It simply must be allowed to ask questions and critizise and trying to improve the general way LEDE is going without singularly getting the answer to "build your own image".

[quote="metai, post:11, topic:509"]
However, "build your own image" cannot possibly be an end-all argument to every question about release cycles, and especially not about release builds. The average user--and I really mean average and not uninitiated--simply cannot be asked to build his/her own image. If that were the case, LEDE would be doomed for a broad audience, however technically inclined. The main audience will want, and need, pre-built images and processes based on pre-built images.[/quote]
I wasn't arguing. I pointed that out because you were talking about re-installing packages/reconfiguring after upgrading. That's why I started rolling my own - that, and security updates on stable branches.

Thing is, I will switch to the stable branch when LEDE releases one, but I will keep running my own builds :slight_smile: .

I am not confrontational here, sorry if it came across that way. I just firmly believe that the cumbersome process of upgrading a system that has been customized beyond the defaults, be it release or snapshot versions, be it preconfigured with uci-defaults or not, is one of the weak points in the whole LEDE/OpenWrt system.

I have high hopes that at some point in the future the upgrade process will be streamlined, with some method of reinstalling packages and retaining (maybe even versioned) config files. Of course there are ways you can do that yourself right now, but again, they are cumbersome. Certainly not suitable for the uninitiated, and even bothersome to "advanced users" like I consider myself.

It is in fact one of the main issues I keep noodling in my head, planning to do a custom (and publicly offered) build for the My Book Live I love so much. I mean, I cannot possibly expect someone to upgrade to a later build and lose customized network connectivity, mount points and network shares unless you know exactly what you're doing and how to prepare everything. It's not a dealbreaker, of course it isn't, but it's something that could end up discouraging users from taking part.

Also keep in mind that until there is an official stable release, you mostly get a self selected audience, which is at least comfortable with using snapshots - starting from there, just building it from source isn't a far stretch (especially considering that luci isn't preinstalled in snapshots).

From reactions that i also see in other forum posts, I think having a first release will be the real start for most end users. Having enough stabillity for day to day use and no surprises.

So i hope there will be a first release not to long from now. I'm more and more convinced that a first release is the main thing that the end users are waiting for. Allthough i hope to get much more feedback on this topic. Just to get an idea who is waiting for the startshot . :slight_smile:

For basic routing LEDE seems to work just fine right now. (on my GLi AR150)
I'm in no way an advanced user, and i have no dev skills, i hope to use LEDE for several small projects. So i consider myself to be a creative enduser. I rely heavily on a stable build and packages that work more or less out of the box. And I will use LEDE more as a micro server, not so much as a router, or well... A hotspot that serves stands alone info pages is also a router.

The My Book project seems very interesting to me. How hard will it be to have that working as a secure little webserver?

I'm not using LEDE because I'm waiting for a stable release.

I don't want to have any hangups or issues. I just want to install and have a trouble-free experience. I don't have time currently to be tinkering and experimenting.

Yes, I think there are a lot of people in your situation. You want the best, but you're not able/willing to spend the time farbling around with experimental software.

The good news is that we are finalizing the "release machinery" (Draft of Release Plan published) that we'll use to generate our long-term stable releases, so we're well on the way to getting there.

This is why it is great to have small boxes like the AR150.

I'm no developer, i really consider myself to be an mixed CLI/GUI end user, So i also wait for that first release before i do more serious things with it. I think I will never be a developer.

I also think that there is a void between the end users and the developers. If i had to give it a name i would call it the configuration void.

Once there is a stable release then the end users can do stuff with configurations. It would be nice to have something going in that section. Where the focus is a bit more on the end user and a bit less on the developer stuff. Many endusers have great ideas but don't have dev skills.

I'm curious what will happen once that gap starts to close.
Ubuntu became the bridge in that gap for Debian somehow.

Not sure what i'm saying now, just an observation, there is probably more to it. Evolution will bridge all gaps eventually. Should we help evolution? Can we? What has to be done? And who can/will do it?

What can end users with no dev skills do to bridge that gap. What skills do they have? how do we know?
Im sure there are bloggers, vloggers, hackers, tinkerers, who can make tutorials etc. As an addition to the wiki. We have this ecosystem of users, and how can we mix the users into this soup of the evolution?

Without expecting them to become developers.

Is there a platform within the community for these people? Or even outside of the community that still can be a benefit for the community? What do we have what do we need, what lies inbetween that has overlap? and hoe can we close those little gaps. So that the DEV's have their freedom to think on the outer side of the inside of the box. And that the end users can do their thing on the inner outside of the box and gain some more insight of the inner workings of the box.

I more and more like the penguin in the box logo. The LID is off the box, everyone can look inside. but for an end user it is not always easy to make sense of what's cooking on the inside.

The skillset of the enduser can be very different, builders, housewives, teenagers, you name it they ALL use routers for MANY hours each day. And they simply cant become devs, maybe some can, but this world still needs farmers cooks painters etc. So i don't think the devs can expect from the endusers to get involved, some will some won't. As long as we are inviting enough and never ever push them away then some may become interested and discover the world on the inside of the box.

So It would be for instance an idea to make a forum section for the void. Don't narrow it down to anything but make a pond, feed the curiosity, and lead slowly teach people little details on the end user level. Think GUI users, who have absolutely no networking experience. And then 'hope' that they start to help to document the GUI, and make videos, write cool blogs how to NOT use a router. The more they write about LEDE the more LEDE awareness in general and the less that remains unknown.

It all benefits.

But there has to be a platform for that part of the community to live on and evolve into system.

Yup. You ARE that evolution. People like you (and me) feel the urge to bridge that gap between the understanding of what's possible, and making it so a) that "ordinary" people can do it, and b) how they can do it.

We're the people who write the HOWTO documents that describe feature/package, tell why it's useful, then describe how to make it work.

Long term, our routers will know how to do these most common (most useful) functions themselves. Already, there's a router (IQrouter) that adds auto-setup and auto-configuration for SQM code onto a base OpenWrt to minimize bufferbloat/latency/lag. The Homenet project is working hard to create auto-configuration for an entire multi-router setup with all the trimmings so that you can just unbox a router, put it in your spare room, and it'll configure itself to your existing network.

But for now, we are the ones who figure out how to do the manual configuration of these amazing facilities, prove them out in all kinds of network environments, and make the world a little better. I'm glad you're on the team.

I'm not using LEDE because in the beginning there were a lot of issues with ath9 which caused the connection to drop, this has been resolved. But with the latest builds I experience some DNS issues that Dnsmasq is not always resolving fine which causes time outs. In the logfiles I cannot find any lead on this. So for now I'm using Chaos Calmer with the latest updates until LEDE is released as stable.