• 0 posts
  • 91 comments
Joined 3 years ago
Cake day: July 23rd, 2023
  • This one is at least as informative as the default. Sometimes I’ll be troubleshooting an application where the developers got a little too creative with error messages and I’ll get an “Error: Oopsie woopsie you broke the website!” And I’m like JUST GIVE ME AN ERROR CODE I’M THE ADMINISTRATOR AND I NEED TO ACTUALLY FIX THIS!

  • Most big LLMs will pass the math off to a more typical service that will solve the math problem deterministically and then pass the result back to the LLM to include in its response.

    So yeah LLMs can’tactually do math but from a user perspective they can.

  • I avoided tailscale for so long because I was already using wireguard and I didn’t know you could self-host with headscale. But once I started using it with headscale the mesh design really is a big improvement to usability. I don’t miss having to carefully manage my config files and ip route rules.

    I need to get setup with app connectors and then I think it’ll finally be a high enough wife-usability factor for me to remove some things I still have exposed over the internet.

  • DERP is the service that actually relays packets between tailscale connected devices when they are crossing a NAT (leaving one private network and going across the internet to another private network).

    If you host headscale (the self-hosted community version of the tailscale control plane) and use it with tailscale, by default it will still use the public Tailscale DERP servers. Your traffic is still encrypted and not visible to them, but it does still rely on part of their centralized architecture even though you are hosting the control plane yourself.

    That being said, you can just use the embedded DERP that ships with headscale, although there are some other considerations when doing that because it will need to be publicly on the internet, probably with a proper domain name and publicly trusted certificate.

  • You can self host the control plane for Tailscale using a community project called Headscale. I use that along with Headplane which gives you a nice admin web UI.

    Then you just use the tailscale client on devices like normal but you authenticate new clients with your endpoint instead of the centralized one.

  • They are. Lots of motherboards still include these. There are a lot of special PS/2 input devices that are still around in business/industrial settings and gamers sometimes need them for use with flight sticks, steering wheels, mechanical keyboards etc.

    Usually it’s a combo port now instead of a separate port for keyboard/mouse.

  • Definitely do not do tapes.

    I’d also recommend Backblaze. Their S3 compatible storage is pretty affordable. I backup to a Kopia repo and then replicate to Backblaze nightly.

    Tapes require so much more work to keep up to date and mght not even be cheaper over time.

  • I have gotten a couple meetings to be something we ‘skip by default’ where we keep it on the calendar but someone only starts the meeting if they actually need something.

    Dramatically cut down on meetings without any problems so far. Now it’s just occasional and way shorter because we get straight to business and then drop the call.

  • At the tail end of my last job I was saddled with a massive project to migrate a client to a new version of an application. We did this by standing up the new version, copying over their current data, asking them to test it and then cutting over when they were ready. This was a huge undertaking because most clients had one or two environments but my client had 18 different environments so the workload was way higher and everything took way longer.

    On top of the scope they also took updates to these environments almost every night which meant it was a full time job just to keep things in sync, setup a testing window and then try to get them to approve the new state of things.

    I was already burnt out before this all started, but thanklessly maintaining 18 non-production environments by myself for an application that no one could commit to testing or cutting over was driving me insane. I felt such a weight lifted off my shoulders when I quit. It came at the end of months of stress and wasted effort. I couldn’t imagine a reality where anyone else would put up with that work or have a better chance of success.

    Anyway I caught up with some coworkers and asked if that project ever got done. Apparently it got passed to a small team of three to manage and after getting jerked around for months themselves the whole thing fell apart.

    So glad I didn’t waste any more energy on that shit.

  • If you mean that you are using Proton VPN on your Raspberry Pi to mask your downloading traffic, then no that same VPN will not help you access services like Jellyfin on your home network while you are remote.

    Instead you’ll want to use something like Tailscale (or Wireguard). You run it as a service on your home network and it then becomes your own VPN that you (or others) can use to connect to your home network when you are remote.

    You could run Wireguard on the same RaspberryPi that you use for downloading but I would recommend against it assuming that you’re running Proton VPN right on the host itself (and not inside a container).