Old thin clients and mini PCs are great for this. Many either have a half sized PCIe slot or can take a second network interface using the WiFi m.2 slot and a 3d printed bracket to mount the nic port.

You’re best off splitting the routing and WiFi tasks into separate hardware. Buy yourself a used ruckus unleashed r550/650 or r510/610 depending on how much you want to spend for wifi then run routing on whatever hardware is fit for purpose. I usually slap OPNsense on something like a dell/wyse 5070 j5005 mini PC, any mini PC with a PCIe slot will allow you to build a 1/2.5/10GbE router with open software. Chinese N100 router boxes are cheap now too, or you could reuse an old mini PC of some kind.

I don’t like rolling my own router using arm boards anymore, router distro support for them is unreliable and j5005 pulls <10W anyway.

Cyber security eh?

I get spam like this on every single messaging platform eventually. Just block and report it and get on with your life.

Drill a small hole and you can scratch it

Happy Cranksgiving 🦃

I’m not sure how to get the N from session history, nor how to check my session history…

journalctl --list-boots will list all sessions stored in the journal.

The output is from yesterday, when the device stopped working correctly.

I’m not familiar with linux kernel, but I can see there is definitely something wrong…

The HDD (old) is attached to a USB hub (new), I tried switching port of the hub but the same issue happened again, if I try to mount it with sudo mount /mnt/2tb, it says it is already mounted:

Those messages tell you what’s happening, there’s an unrecoverable error on the USB bus connecting the hard drive which is causing filesystem errors when writes fail. Diagnose that, lose the hub first and directly connect the drive to the pi, then try replacing the cable that attaches the drive if the error still occurs. I’d also check with people in the rpi community in case there are any known issues with USB on your model. There may be some pi specific USB firmware things you can do to increase reliability.

You can also try disabling UASP for the drive in case BOT transfer somehow stabilizes the connection. You’ll lose performance but that helps with some USB storage bridges.

Some USB storage bridges are just unreliable under Linux and crash under load, your last option is to buy another drive enclosure that’s tested and known to work correctly. I went through like 5 USB/NVMe enclosures looking for one that worked properly, that whole space is a compatibility mess.

Don’t just look at sdb hits in the log. Open up that entire session in journalctl kernel mode (journalctl -k -bN where N is the session number in session history) and find the context surrounding the drive dropping and reconnecting.

You’ll probably find that something caused a USB bus reset or a similar event before the drive dropped and reconnected. if you find nothing like that try switching power supplies for the HDD and/or switching USB ports until you can move the drive to a different USB root port. Use lsusb -t and swap ports until the drive is attached beneath a different root port. You might have a neighboring USB device attached to the bus that’s causing issues for other devices attached to the same root port (it happens, USB devices or drivers sometimes behave badly.)

Always look at the context of the event when you’re troubleshooting a failure like this, don’t just drill down on the device messages. Most of the time the real cause of the issue preceded the symptom by a bit of time.

Honestly, unless the community can all get behind one set of recommendations it’s going to fall on whoever’s running the instance.

I’m really hoping we see some kind of server side community bundling sometime soon because that would make it possible for users to curate bundles of communities themselves and people could just subscribe to those if building a subscription list is too much effort. Present new users on an instance with the top 5 or 10 bundle feeds and let them explore those while they build their own subscription selection.

EDIT: fuck, it just clicked me why you said so. Discord might be a good place to advertise Lemmy, indeed - good catch.

I just drop Lemmy post links the way I used to drop reddit links and answer questions if someone asks about Lemmy. That’s really all it takes to draw interested people in over time.

I drop links to interesting Lemmy posts into discord and my other communication channels in moderation and it’s hooked a couple of handfuls of my friends enough to make an account and join. Reddit spread because people linked to useful communities there often, we can do the same thing here with time and effort. We’re not going to see explosive growth the way a commercial enterprise would, we don’t have the marketing budget or the salaried people paid to grow the community, we have to do it ourselves slowly and consistently like we have been. Make useful posts, link people to the site, do what you can to reduce the barrier to entry for new folks, etc etc.

The barrier to entry on Lemmy is pretty huge for new people. You have to sort through anywhere between dozens and thousands of instances (after you figure out what an instance even is) then pick one that hopefully aligns with your values (and if you guessed wrong you have to uproot yourself and start all over again.) After that you can either drink from the firehoses (/c/all or /c/local) or if that’s too overwhelming or full of porn or anime vore furry fan art then you need to sit down and commit to hours or days of manually building your own subscription feed from the tens of thousands of available communities. Then, finally, if you stuck it out through all of the above you can start interacting with the rest of the Lemmy community and be welcomed by the well intentioned and constructive posters of instances like .ml. It’s a lot to ask of someone who wants Reddit but without spez or Nazis and not that many people are motivated to persist through all of it.

If I had to suggest one way to reduce that barrier to entry I’d suggest building a default set of subscriptions (maybe take all of the trending communities from the last 12 or 18mo and put them in a feed) so that new people can start participating more quickly. I think that would reduce the “fuck this I give up” rate considerably.

A lot of communities moved to Discord, and I mean a lot of them. I don’t think that was the right call, Discord and Slack are absolutely the wrong tools to maintain a forum, but it was easy and familiar so they hoovered up a lot of communities.

I stood ready to pay and then the cashier said “now I just need your phone number and you can pay”. Hold up. What. I did not expect that, I honestly had a burst of anger inside me (never gonna take it on a cashier, they are just doing their job). I asked nicely why do I need to give my phone number and I was told that to register me as a member so I can get the discount.

“Sure thing, It’s +XX 111 222 3333” Just give them garbage.

I’ve had the idea for a while to use an LLM to gather metadata about books for me as well as generate tag lists for themes, plot, writing style, etc for everything in my ebook library. You could also generate non spoiler plot summaries and produce recommendations for similar books.

Why are we tolerating this criminal behavior by corporations?

Because it’s done in the open and it’s accepted as part of the cost of the device. This is an expected consequence of our adtech surveillance economy where devices are now subsidized because they can harvest data about you, your usage and your behavior to sell on an ongoing basis. We’ve been screaming about these sorts of practices since the late 90s and consumers have just blithered right along with every new and creepy intrusion because they get cheap things and don’t think about the real costs or consequences. And so … Here we are.

I’m just waiting for them to add a sideband channel to some LoRa network so they can exfiltrate data even when their devices are “offline”

Interesting that the one has such large capacitors in it. I imagine that is as last-ditch effort to keep the board powered long enough to finish flushing all of its caches in the event of a power failure.

That’s exactly the point of power loss protection (aka PLP.) As a side effect of not needing to wait for a flush after a write synchronous write workloads are dramatically faster on enterprise drives with PLP.

Edit: To add a bit of detail - you don’t need to wait for a flush after a synchronous write with PLP because the drive firmware can lie and immediately return from a flush call because there’s enough backup power to complete that flush if the power were cut.

Why don’t you volunteer to be the test case Larry?

Yes, the machine that stays off 363 days of the year is such a security risk to my home network 🙄