Neo

https://social.hacktheplanet.be/@neo

Cyberpunk | Programmer | Ruby on Rails veteran | Nix user | Sysop | Mr. Fusion maintainer for the MiSTer project

Website: https://michaelsmith.be
GitHub: https://github.com/michaelshmitty
Location: Antwerp, 🇧🇪 Belgium
MiSTer FPGA: https://github.com/MiSTer-devel/Main_MiSTer/wiki

Public Key

npub1acc2dnywrwkvaydc7uy3a3ztfh7j8aqg9mwnz4ec09hppnnlg8msc9v90a

NIP-05 Address

neo_at_social.hacktheplanet.be@momostr.pink

Profile Code

nprofile1qqswuv9xej8phtxwjxu0wzg7c395mlfr7syzahf32uu8jmsseel5racsdh45f

Author Public Key

npub1acc2dnywrwkvaydc7uy3a3ztfh7j8aqg9mwnz4ec09hppnnlg8msc9v90a

Show more details

Last Notes

2024-11-15 22:52:49

- reply

I think #nostr comes close. But it appears to still rely on relays for passing along messages.

2024-11-15 22:51:19

My dream is still to have a completely serverless, distributed, peer to peer social media platform experience. Where all the data is just kept on your device. I imagine that’s what #solid is going for?
https://solidproject.org/about

2024-11-13 00:49:32

- reply

indeed, ssh-agent support would be a great addition to age.

2024-10-28 22:11:57

- reply

Sir, have you heard about #Nix? I think drinking is allowed but there’s other stuff you have to put up with though. YMMV. /s

2024-08-06 09:10:34

https://world.hey.com/dhh/cookie-banners-show-everything-that-s-wrong-with-the-eu-3bfcf03d

2024-08-04 22:48:28

- reply

mailman worked well for me when I ran a few lists 5-7 years ago. I would recommend it. It’s very mature and has a lot of options. It’s been around and knows a lot of the quirks. I would not go as far as to call it painless, but ymmv. I believe there are hosted mailman services too that take away some of the churn.

2024-08-03 16:58:23

- reply

heel mooi. Veel bugs!

2024-08-03 07:26:01

- reply

ooow!
https://social.hacktheplanet.be/system/media_attachments/files/112/896/898/116/801/642/original/3862a084c1bfa24b.jpeg

2024-08-02 12:42:07

Rien Poortvliet, die kon wel illustreren zeg!
https://social.hacktheplanet.be/system/media_attachments/files/112/892/479/107/223/293/original/d0be29a39a087f32.jpeg

2024-08-02 09:15:23

Hackers op Tiengemeten! @npub1whe…jcnz
https://social.hacktheplanet.be/system/media_attachments/files/112/891/665/280/737/617/original/3aa04bf03eef9b17.jpeg

2024-07-31 22:54:49

Tss, augustus al!

2024-07-22 16:09:50

- reply

Microsoft is at fault here too for digitally signing a driver than can dynamically execute arbitrary code.

2024-07-22 14:22:24

- reply

it's bad. Crowdstrike have a driver running in ring 0 on the CPU, the highest privilege level. Imagine what it can do for a second. These drivers are required to be signed by Microsoft. To prevent having to go through the signature process every time they update the driver they have it execute dynamic code downloaded in userspace through updates. That is absolutely insane. Reference: https://youtu.be/wAzEJxOo1ts

2024-07-22 13:21:53

@npub1qda…wevs the balls on this company haha
https://go.theregister.com/feed/www.theregister.com/2024/07/22/windows_crowdstrike_kernel_eu/

2024-07-19 21:13:09

- reply

forcing the use of UUID's appears to have fixed it! So even though the hosting provider supposedly allowed me to configure /dev/sdX to always assign to the same device, this is in practice not true. UUID's it is! Perhaps maybe labels for better portability. Thanks for your help, really appreciate it!

2024-07-19 21:01:59

- reply

of course, I totally forgot about that. Great suggestion, thanks!

2024-07-19 20:56:48

- reply

I think the reason they recommend using device names over UUID's in the guide (https://www.linode.com/docs/guides/install-nixos-on-linode/#rewrite-device-identifiers) is because you can make an image that you can make many VPS instances of without having to configure the UUID for every new instance. But for my use case that's something I can live with if this fixes the problem.

2024-07-19 20:53:38

- reply

I will try it though, because check this: https://social.hacktheplanet.be/@neo/112815135569328413

2024-07-19 20:52:36

- reply

I will try this after all. Because from the latest screenshot it shows that it's trying to boot from /dev/sda which it recognizes as the swap partition! Even though the Linode configuration clearly sets nixos as /dev/sda and swap as /dev/sdb. Compare the two screenshots.
https://social.hacktheplanet.be/system/media_attachments/files/112/815/134/957/263/734/original/82f7164a9a360bac.png
https://social.hacktheplanet.be/system/media_attachments/files/112/815/135/397/539/393/original/d7103557c8a0e755.png

2024-07-19 20:04:15

- reply

what it feels to me is that QEMU does not have the disk “physically” available to the VM yet. It sounds weird though. Why would a virtual device not be available yet? It’s not like it’s spinning rust that takes it sweat time to become available like back in the 90s!

2024-07-19 19:59:28

- reply

I’ll consider that. But Linode actually recommends to use the sdX as the drive names in their official NixOS guide (https://www.linode.com/docs/guides/install-nixos-on-linode/#rewrite-device-identifiers) because your defined them fixed in their hosting panel configuration for the VPS anyway. Thanks for thinking along, really appreciate it!

2024-07-19 19:45:19

- reply

yes I must have missed that. Do you have any references? By "booting me of the project" do you mean it won't boot properly because I'm not supporting trannies? Surely they can't such a thing? That would be crazy.

2024-07-19 19:40:59

- reply

And here's the same problem rebooting just now on Linode with `boot.initrd.systemd.enable = true`. I have enabled `boot.initrd.systemd.emergencyAccess` now so I'll poke around.

2024-07-19 19:37:21

- reply

why not?

2024-07-19 19:32:11

- reply

Screenshot of the #nixos boot problem on a VPS at Hetzner. Booting with systemd in initrd gives a similar error.
https://social.hacktheplanet.be/system/media_attachments/files/112/814/817/014/632/689/original/be7bad06b9346571.png

2024-07-19 19:25:34

- reply

I can’t be the only one running into this with #nixos on Hetzner and Linode. Unfortunately neither providers support #nixos officially (yet), so I’m on my own. It’s blocking me to use #nixos in production for a project. Unless I don’t rely on auto reboot after upgrades. But even if the server does not come back up on a manual reboot, it’s still annoying.

2024-07-19 19:21:58

- reply

helped me debug the problem and suggested booting #nixos with `boot.initrd.systemd.enable = true`. This appeared to work, unfortunately the problem still occurs. It feels like there’s some kind of timeout condition at play, perhaps related to the QEMU system that these VPS use?

2024-07-19 19:18:31

- reply

Choosing reboot from the initrd error prompt doesn’t fix it. But restarting the VPS through the hosting dashboard, on Hetzner the equivalent of ctrl+alt+del fixes the problem and boots the server.

2024-07-19 19:17:05

I’m debugging a problem I’ve had for many months, possibly years using #nixos on VPS with Linode and Hetzner where the server will not come back up after a reboot. Mostly after an automated upgrade but I’ve been able to reproduce it by rebooting manually as well.
The problem is that sometimes the first stage cannot find the block device to mount and just hangs on an error message. I’ll share a screenshot later.

2024-07-19 14:54:58

What is this #crowdstrike thing you speak of? /s
#linux

2024-07-19 14:44:45

- reply

that makes sense, thanks for your response. Maybe it’s another one of those paradigm shifts that #nix brings and I need to look at it from a different viewpoint than when I was using Debian. I already noticed that going from NixOS 23.11 to 24.05 does not feel the same as going from Debian 11 to Debian 12. With NixOS is much more seamless and feels much more solid. Plus it’s not a replacement so you can easily roll back. Another win for #NixOS

2024-07-19 12:04:40

- reply

thanks, I read the discourse threads. I’ll link them here for reference.
https://discourse.nixos.org/t/interest-in-a-nixos-lts/4705
and
https://discourse.nixos.org/t/issues-when-suggesting-nixos-within-corporate-environments/28629/3
and
https://dataswamp.org/~solene/2024-04-27-nix-internal-crisis.html#_Extra:_Why_did_I_stop_using_Nix?

2024-07-18 18:48:00

- reply

the context of course is important too, NixOS is such a general distro with a wide variety of uses. I was considering the LTS concept for servers. I’m fine with half year release cycles for desktop. I may not be familiar enough with the reasoning behind half year cycles. If someone has more on this, I’d love to hear!

2024-07-18 18:44:55

- reply

I find it a bit too fast, but that’s coming from Debian for 20 years, which arguably has one of the slowest release cycles. It’s more that on release of the new NixOS version, the previous becomes unmaintained if I understood correctly. I think there is no concept of security fix backports to previous releases NixOS, or is there?

2024-07-18 14:13:05

Considering looking into a #nixos LTS. Coming from Debian I really miss the slow paced, solid release cycle of #Debian compared to #NixOS.

2024-07-02 17:50:11

If you like DHH’s Omakase developer setup #Omakub for #Ubuntu and you’re into #nix and #nixos or want to try it out, check out #Omakix a port of the project I made using Nix and NixOS. It even has an installer ISO you can use in a VM or on bare metal.
https://github.com/nixup-io/omakix