Collecting statistics from LEDE devices (telemetry)

For your info, I see LEDE are packed with a lot of VPN packages available in those oppressive countries, in some underground website. They don't need to visit this website, because if they are not using VPN, probably they can't reach us.

Telemetry traffic in 10000 hits to a single site will get flag in their national firewall. Matching with VPN traffic will lead to more people get caught and send to jail.

Not about those countries, but in general, I am not sure how to best balance the development needs, to the privacy of users. If it has to be included, better make it explicit to everyone there will be data collection, i.e. on download they must agree they know these data will be sent in. Those using Image Generator should be alert they could take the telemetry features away if they are building for users in oppressive countries.

I hope LEDE will not fall into the alert list of info security professional group as leaking privacy, because it hurts the project in long term basis.

2 Likes

How about a field to identify LEDE community builds?

Okay, good to see that it's within scope

NO!!!! If LEDE starts to collect any user data then fuck it! I will stop using that shit!!! It's my private router. LEDE is free and open source it should keep my privacy! Is this why they divided from openwrt? LEDE wants to use telemetry and openwrt team did not? Why the hell would is this question even doing here? This is linux, free and open source linux OS must keep our privacy!!! This is the only OS that does this. MS, Apple and GOOGLE are big cocksucker companies. They do not respect user privacy! Do not be like that! You are better then that.

  1. Mind your language.
  2. There are reasons pro, and reasons con. The radical denial of any positive aspects of statistics, combined with the continued use of open source software, gives your attitude a bitter aftertaste.
    Exaggerated example (but not so far from reality): "Hey, are you sleeping? I need $mypersonalroutermodel supported urgently! Until the end of this month!!! [...] What? You want to collect anonymous statistics to improve this software? NO WAY! You have one week left to get my device supported, you hear me???"
    The funny thing with that: By this strict denial you confirm, that you do not trust the people creating the software, while you still trust the software created by those people.
  3. If there will be some statistics collecting ever, I guess it will be a manually to install package with manually to send data. No reason at all to be upset or use strong language.
2 Likes

If my privacy is price for this free software then it's not free and i am not willing paying that price for it. That's all i wanted to say. Agree i overreacted with my speech though.

Hi all, as the last response is 10 month ago I'm not sure if this went anywhere. If the two issues of implementation and dialing home are still unsolved, I'd like to propose an idea.

For some time I've worked on a backend which creates firmware images on demand and tells you if your current running firmware (opt. incl. packages) is outdated.

To do so a Luci App gathers device information and ask the server if any upgrade is available. To increase performance the requests are cached on the back-end, only containing distro/version/target/subtarget/[packages].

If an upgrade is available, the user can request the image which the server builds within a few seconds, containing all previously installed packages. If desired the firmware upgrade is downloaded and installed with another click, the Luci app then informs the user about a successful restart (done by trying to reconnect to ubus).

This is the current state, however I see this fitting perfectly fine in the desired telemetry data. Users actively and consciously send device specific data to the upgrade server. The stored upgrade checks and build requests let you easily create some statistics, here just a very trivial example.

Lastly, the Luci app could request the user to kindly report a successful upgrade, so after reconnecting to ubus, the app could offer a button like "Report successful installation".

I've no experience with any message broker, my Luci app skills are very limited and I'm no data scientist, so I'd be happy for anyone helping me with improving the performance, interface or data visualization.

The question is: what is the benefit of the stats for the OpenWrt community, how will it help to make OpenWrt be better?
Collecting data about the router (software) i use is just another piece of information about my preferences - stored on OpenWrt servers but as well as all corresponding metadata on my cable modem, my ISP, de-cix in Frankfurt, and more you could imagine...
Compared to all the other information of metadata I left there with just posting this Message it is a really small piece.
For the super paranoids it is already too much - for me, as a regular paranoid I could live with it :nerd_face:
But I would also like to see the benefits of the stats - on top of the possibility to enjoy looking at them.

Edit: Ok, looking at http://anon.groov.pl/ I must admit that just looking at the stats is awesome. I would like to see the same for OpenWrt. I vote for "LEDE should collect statistics by default".
But still some other countermeasures could be taken after evaluating the usage. E.g. devices with a wide spread usage might be more supported by e.g. providing build scrips. Or other useful stuff... And target profiles without usage could be dropped...

I don't have a real objection to optional statistics. You could simply just copy the way a Tomato build by Shibby I used to run did it, http://tomato.groov.pl/?page_id=314 It's a bash script.

You can see the output of the current stats in action. http://anon.groov.pl/

For privacy concern, the script doesn't send mac # or ip #, it sends the MD5SUM instead.