| 7 Jun 2024 |
 raitobezarius | Buffers? | 21:07:58 |
| raitobezarius | Or TCP Nagle idk | 21:08:36 |
 netpleb | Ok, thanks for your input. I do not have much experience with sshfs either. Maybe sshfs over the internet is just super slow like that. I read somewhere that people get 30-50MB/s when using sshfs on their lan, but maybe it does not work very well on wan | 22:04:57 |
| 8 Jun 2024 |
 Linux Hackerman | In reply to @joerg:thalheim.io
But using the nixos configuration instead of fstab the disko configuration, not the nixos configuration, right? | 11:52:39 |
| Linux Hackerman | last I checked disko configuration doesn't provide the full range of functionality you can get via writing fileSystems yourself, especially stuff relevant for impermance-style setups | 11:53:20 |
 Mic92 | In reply to @linus:schreibt.jetzt
the disko configuration, not the nixos configuration, right? The nixos configuration that has the disko configuration included | 11:53:21 |
| Linux Hackerman | like if I want to have / be a subdirectory of the filesystem, there's no way to represent that in disko | 11:53:55 |
| Mic92 | Linux Hackerman: I think there is an open issue, I haven't responded to, but there is https://github.com/nix-community/disko/blob/master/lib/types/nodev.nix | 11:54:19 |
| Mic92 | In reply to @linus:schreibt.jetzt
like if I want to have / be a subdirectory of the filesystem, there's no way to represent that in disko Not sure what this would look like for other filesystems or how you mount this in NixOS. For zfs we have an impermanence example: https://github.com/nix-community/disko-templates/blob/main/zfs-impermanence/disk-config.nix | 11:57:01 |
| Linux Hackerman | In reply to @joerg:thalheim.io
Not sure what this would look like for other filesystems or how you mount this in NixOS. For zfs we have an impermanence example: https://github.com/nix-community/disko-templates/blob/main/zfs-impermanence/disk-config.nix using the X-mount.subdir= mount option | 11:57:25 |
| Linux Hackerman | I guess that could work with nodev | 11:57:30 |
| Mic92 | I think we also have mountoptions for other filesystems | 12:01:02 |
| 9 Jun 2024 |
 Qubasa | 
Download 1000006011.png | 01:55:22 |
| Qubasa | Mic92: osdev.org seems to have a similar journey to the nixos.wiki | 01:56:04 |
| Mic92 | Forking wikis is hard. | 04:54:41 |
| Mic92 | If we could have better mechanisms to then sync changes back. | 04:55:11 |
| Qubasa | that sounds like a chatgpt task :) | 10:23:45 |
 K900 | Please no | 10:24:35 |
| 10 Jun 2024 |
|  NixOS Moderation Bot unbanned  jonringer. | 00:17:07 |
| 11 Jun 2024 |
|  @networkexception:chat.upi.li changed their display name from networkException to networkException (moving to @networkexception:nwex.de). | 18:32:31 |
| 12 Jun 2024 |
|  networkException joined the room. | 21:44:37 |
| 13 Jun 2024 |
 dnr | In reply to @netpleb:matrix.org
I was thinking of trying it as a native btrfs backup method but where the remote backup server does not ever have access to the data. It simply exposes a block device that the client can setup with luks+btrfs. I have done this! I used a scheme like this for personal backups for a while, using btrfs send/recv over luks over iscsi over tcp over zerotier over slow home internet connections. It took a lot of tuning for reasonable performance, but after that it was actually more stable than I expected. Which is not to say that it was very stable.. I had low expectations :) It never got corrupted or badly wedged, but I decided it was just too risky for the case of backups, which are supposed to de-risk things. Local filesystems just aren't mean to run over high latency links. | 11:39:53 |
| dnr | (Looking at the old code again, wow, this was crazy. I had to write an adaptive pipe with a PI control loop to limit the block requests in flight. Also a dm-cache layer so it didn't go back and forth for metadata so much. Fun project but yeah, crazy. I can share the code if anyone wants.) | 12:04:04 |
| 14 Jun 2024 |
| raitobezarius | Does anyone know of a way to parse IPv6 addresses in textual form with a zone identifier with the C library? | 21:15:44 |
| raitobezarius | inet_pton doesn't seem to fit the bill | 21:15:48 |
| netpleb | In reply to @dnrix:matrix.org
(Looking at the old code again, wow, this was crazy. I had to write an adaptive pipe with a PI control loop to limit the block requests in flight. Also a dm-cache layer so it didn't go back and forth for metadata so much. Fun project but yeah, crazy. I can share the code if anyone wants.) Very interesting! Thanks for sharing your experience. I have also gotten further in my own testing of it and have been surprised how well it worked, or even that it worked at all (though I am not using it for anything real yet). I may ping you at some point in the near future if I end up pursuing this any further. | 22:46:01 |
| 19 Jun 2024 |
|  Zach joined the room. | 21:24:18 |
| 21 Jun 2024 |
| Linux Hackerman left the room. | 14:07:00 |
| 22 Jun 2024 |
|  nyanbinary 🏳️⚧️ joined the room. | 01:21:24 |
|  thaumavorio joined the room. | 09:22:28 |
