| 6 Apr 2024 |
 K900 ⚡️ | The more correct way to do it would be through a bootloader script, yes | 09:48:43 |
| K900 ⚡️ | But the important thing here is that you can't just symlink NixOS init to the right place and expect it to work | 09:49:01 |
| K900 ⚡️ | Because normally activation happens before that | 09:49:06 |
| K900 ⚡️ | So you have to have your own thing that will run the activation script and then run the stage2 init | 09:49:24 |
 steveej | FWIW, this works imperatively on the system:
set -xeEu -o pipefail
result="$(nix build --refresh --tarball-ttl 0 github:holochain/holochain-infra/workorch-zos#nixosConfigurations.tfgrid-devnet-vm0.config.system.build.toplevel --print-out-paths --no-link)"
ln -sf "${result}"/init /init
| 09:49:30 |
| steveej | i'd be ok putting this anywhere appropriate. but i want to prevent running it anywhere else other than on the actual target system, for probably obvious reasons 😆 | 09:50:47 |
| steveej | this is just the update use-case. i've already got something for the initial system assembly | 09:51:07 |
| steveej | i'm looking through the shim code you've sent me and am intimidated by wrangling with /dev entries :-) is this used as a PID1 shim? | 09:53:34 |
| steveej | either i'm oversimplifying or my use-case is actually simpler than in WSL | 09:54:17 |
| K900 ⚡️ | The /dev stuff can be ignored | 09:55:47 |
| K900 ⚡️ | It's a workaround for a stupid WSL bug | 09:55:50 |
|  @tumble1999:matrix.org joined the room. | 10:11:52 |
| steveej | this works (obviously it's not a noop anymore)
boot.loader.external.installHook = pkgs.writeShellScript "noop" ''
${pkgs.coreutils}/bin/ln -sf "$1"/init /init
'';
| 10:37:57 |
 adamcstephens 🐝 | The lxc container does something similar | 11:29:02 |
|  Sammy (It/Its) joined the room. | 13:05:26 |
 raitobezarius | My googlefu is failing me but is anyone aware of some studies on NVMe/SATA SSD being shared between N VMs and expressing their inner performance characteristics like speed and IOPS off the original device? | 18:22:07 |
| raitobezarius | With some synthetic workloads or something | 18:22:16 |
| raitobezarius | I am also curious if it talks about virtio single SCSI driver and takes into account IO threads, etc. | 18:22:59 |
| 7 Apr 2024 |
|  Dragos-Andrei Rotaru (dragosh_developer) joined the room. | 11:49:47 |
| 8 Apr 2024 |
|  Levi joined the room. | 03:20:38 |
| Levi changed their display name from levigross to Levi. | 03:38:38 |
|  Saturn joined the room. | 06:01:33 |
| K900 ⚡️ | https://lore.kernel.org/all/ZhBRSM2j0v7cOLn%2F@oracle.com/T/#u | 11:21:55 |
| K900 ⚡️ | Woo new dtrace | 11:21:57 |
| raitobezarius | wooooooow | 11:22:42 |
| raitobezarius | nixpkgs when | 11:22:55 |
| raitobezarius | i so want dtrace | 11:23:25 |
| K900 ⚡️ | I love how their main branch is dev | 11:25:15 |
| K900 ⚡️ | And the 2.0 branch is devel | 11:25:19 |
| K900 ⚡️ | Just in case you get confused | 11:25:22 |
