| 11 Mar 2025 |
 ElvishJerricco | (unless the decompressed size is huge) | 18:16:51 |
| ElvishJerricco | wait not it doesn't decompress | 18:17:05 |
 aloisw | Systemd-boot does not decompress the initrd and I think the kernel alread | 18:17:07 |
| ElvishJerricco | * wait no it doesn't decompress | 18:17:09 |
| aloisw | * Systemd-boot does not decompress the initrd and I think the kernel already has virtual memory at the point it does. | 18:17:18 |
| ElvishJerricco | regardless, the bug on nixos's side is how the hell we made a 358M initrd | 18:18:00 |
| ElvishJerricco | god dammit nvidia firmware | 18:41:01 |
| ElvishJerricco | K900: do you expect that putting your nvidia modules / firmware in initrd would make it 358M? Because that's what happened | 18:41:35 |
| ElvishJerricco | I don't know why it only happened for them on unstable, but I suspect they left some info out | 18:42:19 |
 K900 | Uhh | 18:48:39 |
| K900 | I think the Nvidia firmware is chonky | 18:48:45 |
| K900 | But I don't know if it's that chonky | 18:48:50 |
 antifuchs | Did it pull in a ton of x11 libs? | 19:03:12 |
| ElvishJerricco | the firmware files wouldn't | 19:30:15 |
|  Adam Neverwas removed their display name adam_neverwas. | 23:24:10 |
| 12 Mar 2025 |
| Adam Neverwas set their display name to Adam Neverwas. | 14:27:30 |
| 13 Mar 2025 |
 mornix | Hi, I saw this is the place to talk about boot counting (via #330017). Is https://github.com/NixOS/nixpkgs/issues/337089 still seen as the way forward? And if so how can I help with boot counting? | 01:16:26 |
| 17 Mar 2025 |
| antifuchs | well this is unpleasant: it seems that systemd-networkd is now starting after zfs filesystems get unlocked, which makes it difficult for my ssh server to get online so I can remotely enter the encryption password | 13:11:46 |
| antifuchs | also, huh, did boot.initrd.systemd start setting networking.usePredictableInterfaceNames to off? I had to set up a rule for eno1 but that device isn't present, only eno1np0 and eno1np1 | 13:32:23 |
| antifuchs | * also, huh, did boot.initrd.systemd change interface naming schemes? I had to set up a rule for eno1 but that device isn't present, only eno1np0 and eno1np1 | 13:34:53 |
| antifuchs | * also, huh, did boot.initrd.systemd change interface naming schemes in nixos-unstable? I had to set up a rule for eno1 but that device isn't present, only eno1np0 and eno1np1 | 13:35:02 |
 gdamjan | that seems to be kernel dependent | 13:41:16 |
| antifuchs | yeah, seems so. I should make up my own names for netdevs | 13:52:11 |
| antifuchs | still, what a damn trap to run into /: | 13:52:20 |
| antifuchs | ok, here's a really annoying thing I don't know how to handle: when critical bootup things fail (or when prompting for my zfs unlock password), systemd kills and re-starts the password prompt every ~2-3 seconds, which makes it impossible to enter the password | 14:54:06 |
| antifuchs | (can baaarely type a 60-char password fast enough, plus the IPMI web console doesn't allow me to type that fast at all) | 14:55:41 |
| gdamjan | also sometimes a BIOS update can mangle the names (happened to me once) | 16:36:43 |
| antifuchs | yeah - I ended up using matchConfig.MACAddress in addition to a name glob, that seems to have fixed the issue | 16:37:11 |
| antifuchs | (the "in addition to" was the key, otherwise child interfaces got the parent interface config which made things look right but behave extremely wrong) (: | 16:37:44 |
| ElvishJerricco | those two shouldn't have any sort of ordering constraints against each other I don't think | 17:58:43 |
