Another thing that doesn’t work anymore after blocking network traffic from my Android phone: Some push notifications.

I run a Matrix server for our family. I use “FluffyChat” on my phone. Traffic from the phone to my Matrix server is allowed and chatting in FluffyChat works.

But I don’t get any notifications anymore on new messages.

So, what’s going on here? Does FluffyChat, which only really needs to talk to my own server, rely on some cloud service for notifications? Seriously? 🤔 How does that work, does this cloud service see all my notifications or what?

Anyone around who did app development on Android? Can you shed some light on this?

(The old device I’m referring to was a handheld device. It was not built into the car and was not running 24/7.)

I’ll make an experiment: I’ll keep blocking all the phone’s internet traffic and then we’ll see how bad the GPS performance will get in a couple of hours/days. 😅 (If I got it all wrong and it still works fine, that’d be great!)

@prologic Regarding the static URL: The hostname at least is a CNAME record and resolves to something at cloudfront. What I meant was that I, as a user, cannot configure this URL anywhere in the Android UI. 🤔

@prologic Hmm, have you used a GPS device 15, 20 years ago? I had one in my car. It would take a long time until it got a first “fix” of your location. That’s because it can take up to 12 minutes until you have gathered all the data directly from the satellites. These days, GPS trackers on smartphones get a fix within seconds, maybe 30 seconds tops, because they get pre-seeded with (approximated) satellite positions via A-GPS.

We also not only have the USA’s GPS these days but also other satellite systems like the EU’s Galileo or Russia’s Glonass. A-GPS helps you get “in contact” quickly with more satellites, which enhances the precision quite a lot.

So, yeah, you can use it without A-GPS. But it would be very annoying and imprecise. I bought a new phone last year and A-GPS was broken on that one (I saw no internet traffic at all), which made it basically useless, to the point where I wouldn’t want to use it at all. I sent it back and bought another model.

To my knowledge, the only way to use GPS without something like A-GPS is to have it turned on all the time, so you get regular updates directly from the satellites.

One thing I’ve learned from locking down my Android phone (see #pknsrda):

The data for assisted GPS does not come from Google or, better yet, A PUBLIC SERVICE, but from a server hosted by the hardware manufacturer. Without regularly fetching fresh A-GPS data, the GPS performance is much worse (https://en.wikipedia.org/wiki/Assisted_GNSS).

This means that the hardware manufacturer has (more or less) direct control over whether I’m able to use GPS or not. This isn’t an Android setting, it’s buried deep within the device, no way to change the URL. If that manufacturer decides one day to cut me off, for whatever reason, or goes bankrupt or whatever, then I’ll have to buy a new phone.

And of course, this data transfer is encrypted as well, so I don’t know what my phone sends to those servers.

All this smartphone business is such a clusterfuck. I should have never bought one of those things.

@prologic Thanks. I expected this to take much longer. 😅

Whoohoo, it’s fixed. 🥳 Now I can look at funny pictures again!

I took the opportunity to remove some dependencies on the internet from my workflow. Actually, outages like these are healthy.

This is going to take a while … See ya in a couple of days/weeks.

I have a day off, national holiday.

What happened so far:

• Internet outage since early in the morning. Still going on.
  
• Unable to reach a human being at my ISP, so I hope they mean it when the computer voice says “we know it, we’re on it”. 🤣
  
• systemd (PID 1) crashed. Might be partially my fault, but meh.
  

I take this as a sign to not do any computer stuff today. 🤣

@prologic Doesn’t matter. Far, far away! From everything! That’s where I’d go. 😂

Ran a few tests.

Copying data from the NAS’s encrypted ZFS pool to the USB disk’s encrypted btrfs runs at ~20 MByte/s. That is for a single 1 GB file of random data. Cold caches, sync included.

That same USB disk with the same btrfs can sustain ~75 MByte/s when I use it on my workstation (i7-3770).

And indeed, the aes flag does not show up in the output of lscpu on the NAS.

I’ll try to tweak some things about this, but it might be time for an upgrade … 🫤 (Or I’ll have to re-think the entire thing somehow.)

@mckinley It’s probably a bit faster, but not much. Maybe 20-30 MByte/s (I watched one 40 GB file being copied and it took 20-30 minutes or something like that.)

I need to optimize this. 🥴

The “annoying” thing about hardware these days is that it basically keeps working “forever”. At least much, much longer that you’d expect.

Now that I think about it … I only remember one PC of mine actually dying because of a hardware failure – and that was probably because I did too much overclocking. 😂 If it wasn’t for changes in software, I could probably still use them all. I mean, why not, my Pentium 133 still works and I use it for gaming regularly.

So … my little NAS probably won’t die any time soon. Hmmm.

@mckinley Not really sure, to be honest. Probably a couple hundred GB … ? 🤔 With the changed data, it might be half a TB to transfer? I’m just guessing.

Let’s see how it goes next time. I don’t expect to add much data any time soon. (On the other hand, I’ll swap the USB disks for the next run, so it’ll take the same ~9 hours, again. Meh.)

I think the solution is to have less data. 😈

@lyse@lyse.isobeef.org Yeah, only ~30 of the ~133 feeds I’m following have had a twt in the last month … 56 in the last year. Some had their last twt in 2016. 🫤

@prologic From the DOM? That can’t be right. 😳😳😳

@prologic It always fetches the canonical feed URL and, when it can’t find the latest twt hash (that it saw in the previous run) it traverses the archived feeds until it does find it. Something along those lines.

I just got one such notification:

Date: Tue, 07 May 2024 15:56:01 +0200
From: me@pinguin
To: me@pinguin
Subject: [regularly] jenny

Fetching archived feed https://twtxt.net/user/prologic/twtxt.txt/1 (configured as prologic, https://twtxt.net/user/prologic/twtxt.txt)
Fetching archived feed https://twtxt.net/user/prologic/twtxt.txt/2 (configured as prologic, https://twtxt.net/user/prologic/twtxt.txt)
Fetching archived feed https://twtxt.net/user/prologic/twtxt.txt/3 (configured as prologic, https://twtxt.net/user/prologic/twtxt.txt)
Fetching archived feed https://twtxt.net/user/prologic/twtxt.txt/4 (configured as prologic, https://twtxt.net/user/prologic/twtxt.txt)
Fetching archived feed https://twtxt.net/user/prologic/twtxt.txt/5 (configured as prologic, https://twtxt.net/user/prologic/twtxt.txt)

Now, your feed did not get archived, as far as I can tell. So why am I getting this then? Have you edited a twt just now? That would explain it. 😅

@prologic Strip it from what? From requests being sent to the server? That’s always been the case, afaik. 🤔

@prologic Huh? What does that look like in Chrome? 🤔 (I only have Chromium.)

@prologic My client tells me when it fetches archived feeds. That’s all.

@prologic Ah, yes, that’s better! 👍

(Hmmm, I think I could add the time of the last twt to the output of jenny -l. 🤔 Currently it only shows the last successful retrieval time.)

@prologic Every now and then, I get a notification about Yarn feeds getting archived/rotated. 😅 Appears to work without issues. 👍

@prologic One minor detail: The Makefile wants to run date -Is, which doesn’t exist on OpenBSD. Not sure how relevant this platform is for you, though. 😅

I haven’t come up with a portable solution yet. date '+%FT%T%z' is the closest approximation that works on both GNU and OpenBSD, but it doesn’t include a colon in the time zone offset, so it’s 0200 instead of 02:00. 🤦 I’m not sure if this is ISO8601 compliant. And it’s still not POSIX. 🤦 Well, I tried. 😂