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.

Better to figure this out late than never I guess. Wasn’t this obvious 10+y ago though? Facebook has always been a predatory propaganda firehose.

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.

When he started pushing his own tweets and Republican influencers to the top of my feed I quit the same day

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.

200% markup doesn’t matter when you’re billing it straight to the client anyway 🍻

Emacs users be like

It would have cost you nothing not to post this

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.

Look at MIT and UC Berkeley’s CS curricula and start tackling things that you haven’t covered. They’re both available freely online and you might still be able to find video recordings of Cal’s lectures somewhere (they recorded every class for students who weren’t present or had difficulty understanding in real time until 2015 or so but were hit with an ADA accessibility lawsuit because they weren’t captioned or something.)

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.

“it appears that the GOP’s leadership may have played themselves into making the right call for the right underlying reasons (the censorship powers), but based on a near total misunderstanding of how the world actually works.”

So business as usual for them

They also have some fun policies like allowing employers to pay to moderate their reviews. So yeah, they’re the Yelp of employer reviews these days

Sounds like a job for account sharing with other people

I mean, global warming us slowly into extinction is a pretty AI way to go about it

If: you’re a starred contact and call twice within 10 minutes and I happen to have the phone at hand and I’m pretty sure you have something important to say I’ll probably pick it up.

That happens about once or twice a year. We invented voicemail so we can speak when it works well for both parties.

Ruckus APs and Opnsense have been solidly reliable for me for 5y now. No random fucking with unifi bugs (like having my WPA enterprise SSID punting users out onto the management vlan at random instead of the Kerberos assigned VLAN for that user, thanks unifi) and fantastic wireless coverage has me completely satisfied with my infra choices. Also, Ruckus unleashed handles controller duty on the primary AP rather than requiring a management container, that’s also a plus.

Read the fine print too, a lot of these “mental health counseling services” have carve outs that stipulate that they can pass along data about your sessions to third parties (like your employer.) Most of them aren’t health care professionals so health care law doesn’t apply.