| mason | fsmithred: We should think about zapping today's images from https://files.devuan.org/devuan_excalibur/installer-iso/ I'd be grateful. While they do, indeed, boot on EFI systems, but after the kernel loads, we're dumped into a busybox shell on my NVMe-only system, and I need to figure out just why: https://0x0.st/XIL-.jpg It continues on to the installer on a non-NVMe system, hence my being torn about | 21:37 |
|---|---|---|
| mason | whether to keep the images or not. | 21:37 |
| mason | (There were a couple errors building the i386 image, so there isn't one at present.) | 21:40 |
| fsmithred | I wasn't even aware that they were there. | 21:43 |
| fsmithred | wasn't there some problem installing to nvme? maybe in the past. | 21:45 |
| mason | There can be. There's a build environment for it. A funny trick is that there seems not to be a linux-image-686 package for Testing at present. | 21:45 |
| fsmithred | do they unexpectedly wipe your whole hard drive or otherwise disturb any existing installations? | 21:45 |
| mason | fsmithred: I seem to remember something about it. I'll grep through logs and through the source. | 21:45 |
| mason | fsmithred: I don't expect so. On the NVMe system it simply hangs on the noted screen without going further. I didn't go through a full install on the other system, but it seemed normal enough. | 21:46 |
| fsmithred | I'm downloading it now | 21:48 |
| mason | kk | 21:48 |
| mason | Didn't find any mention of NVMe woes in IRC logs, so I'm going to grovel through the sores a bit. | 21:50 |
| fsmithred | looking for old wounds to open? | 21:51 |
| mason | Maybe? Or new wounds, depending. The breakage appears to be in efiunpack.sh so far. | 21:52 |
| fsmithred | mason, what's your hardware? | 21:55 |
| fsmithred | oh, I was gonna link a forum thread, but it turned out not to be nvme | 21:56 |
| mason | fsmithred: I'm seeing the error on a ThinkPad T480. | 21:56 |
| fsmithred | uefi or legacy? | 21:57 |
| mason | There's not much useful in dmesg so I'm poking around in busybox to see what we *do* have available. The script has a couple options it tries, so perhaps we can find another if I figure out how it can see the ISO. | 21:57 |
| mason | UEFI. Not sure if this box has legacy. Hm. Will check that soon as I reboot. | 21:57 |
| fsmithred | fake legacy | 21:58 |
| mason | Will check. And I forget the name for that now. Getting old. | 21:59 |
| mason | CSM | 21:59 |
| mason | (Had to look it up.) | 21:59 |
| mason | This is interesting. I'm not sure how it's seeing the USB stick at all at present. Need to look up what an ng device is, because the only obvious storage I see is the actual nvme drive. | 22:00 |
| mason | Guessing ng0n1 is referring to the NVMe namespace, but I'm not sure precisely what it is beyond that. But I'm not seeing any indication that the USB stick is seen at all for some reason. I might try this on another NVMe-only system later to see if it shows up there. | 22:04 |
| mason | (Although the fact that the system has NVMe shouldn't matter to the ISO showing up.) | 22:04 |
| mason | If I pull and then re-insert the USB stick I see it probe, but there's no device daemon to run mknod for me. I'm going to see if I can do that manually and have the mount from efiunpack.sh complete. | 22:06 |
| fsmithred | mason, I tried a legacy install in qemu and it fails when installing base system due to lack of merged usr. I saw this in earlier isos. | 22:08 |
| mason | Hrm, hrm. | 22:08 |
| mason | Alright, I'll see about addressing that. Probably worth zapping the images so folks don't burn time on them. | 22:09 |
| fsmithred | is there a place in the build to edit the debootstrap options? I had to do that in live-sdk to make excalibur live-isos. | 22:09 |
| fsmithred | ok | 22:09 |
| fsmithred | will do that now | 22:09 |
| mason | Thanks. And the various debootstrap invocations are in build.sh and build-sudo.sh. In this case, build-and-publish.sh invokes build-sudo.sh, so it'd be in there. | 22:11 |
| fsmithred | done | 22:11 |
| fsmithred | add --with-merged-usr | 22:12 |
| mason | kk | 22:12 |
| mason | will try that now and it'll pop up a new image | 22:12 |
| fsmithred | cool | 22:12 |
| fsmithred | mason, man page says it's just --merged-usr | 22:17 |
| mason | We need a newer or differnt deboostrap. From the build logs: E: unrecognized or invalid option --with-merged-usr | 22:17 |
| mason | ah, kk | 22:17 |
| mason | That seems to make it less angry. | 22:18 |
| mason | going to have to be out of the office at meeting time, so I'll see if I can get a laptop into suitable shape | 22:19 |
| mason | (kid has a teacher videochat meeting) | 22:19 |
| fsmithred | ok. btw, I'm not seeing that option in my build script, so I'm not sure where I used it. | 22:20 |
| mason | Hopefully I've hit the right place for this purpose. The image is up now if you want to try it. | 22:21 |
| fsmithred | downloading | 22:22 |
| fsmithred | to the wrong computer. :( | 22:23 |
| fsmithred | no, it's right. | 22:23 |
| mason | good good | 22:23 |
| mason | seeing if I can get a browser onto this laptop in time to join | 22:24 |
| mason | Be back presently from elsewhere. | 22:26 |
| fsmithred | failed again with same error, and in syslog it shows --no-merged-usr in the debootstrap command | 22:29 |
| mason | Ah, that's a different version embedded elsewhere then. I'll have to find where it is. | 22:30 |
| fsmithred | menu item bootstrap base failed | 22:30 |
| fsmithred | debootstrap --components=* --debian-installer --resolve-deps --no-merged-usr --no-check-gpg excalibur /target ///file:cdrom | 22:31 |
| fsmithred | uh, that should be file:///cdrom/ | 22:32 |
| mason | will look | 22:32 |
| mason | fsmithred: Out of band, I'll do local builds and push nothing until I see something working, and I'll give you ISOs out of band until we'll happy with one. And then we can do a formal build and have it on files. | 22:49 |
| fsmithred | https://get.refracta.org/files/experimental/devuan-live-speech-amd64-20241012_1314.iso | 22:58 |
| onefang | Just waking up. "<mason> We need a newer or differnt deboostrap." mmdebstrap might be an option. I switched to it in my system build scripts. | 23:55 |
| mason | onefang: Decent option. In this case I passed the wrong argument, and later realized it was to the wrong invocation of debootstrap. Going to poke at that. | 23:59 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
