agris | How can I install beowulf from the 2.1 netinst cd? | 03:23 |
---|---|---|
agris | can I edit /etc/apt/sources.list on the 2.1 ascii netinst cd and do a manual bootstrap? | 03:23 |
agris | debootstrap | 03:24 |
agris | I'm attemping to do a OpenZFS .8 rootfs | 03:24 |
agris | https://github.com/zfsonlinux/zfs/wiki/Debian-Buster-Root-on-ZFS | 03:24 |
agris | is it possible to install beowulf from a ascii minimal-live ot netsint/ | 05:37 |
agris | huh | 05:38 |
agris | that's trange | 05:38 |
agris | Devuan Beowulf zfs-dkms is 0.7 while Debian buster is zfs-dkms 0.8 | 05:39 |
yeti | buster has 0.7.12-2+deb10u1 | 05:44 |
yeti | 0.8.2-2~bpo10+1 is in buster's backports | 05:44 |
yeti | beowulf may still not have the backports sorted... | 05:45 |
yeti | or has it now? then I need to update 10 or so [sources.list]s | 05:46 |
golinux | Beowulf won;t have backports until it is released | 05:51 |
golinux | When it becomes "stable" | 05:52 |
yeti | :-D | 05:57 |
agris | huh | 06:23 |
agris | devuan_ascii_2.1_amd64_minimal-live.iso doe not appear to be bootable via BIOS or UEFI | 06:23 |
agris | I was having trouble with my Dell PowerEdge but I've now been trying to get the thing working with QEMU | 06:24 |
agris | devuan_ascii_2.1_amd64_minimal-live.iso does not appear to actually be a bootable medium | 06:24 |
agris | can anyone else confirm this? | 06:24 |
agris | yeti, if that's the case https://pkginfo.devuan.org/cgi-bin/d1pkgweb-query?search=zfs&release=any may not be up to date | 06:25 |
agris | wait nvm on the bootableness | 06:35 |
agris | I think i'm just having a streak of bad luck | 06:36 |
agris | apparently the order things are specified in qemu actually matters | 07:12 |
agris | and idk what the hell's wrong with my server. maybe a bad cd burn | 07:12 |
agris | wow | 07:41 |
agris | I can not seem to get qemu to work reliably | 07:41 |
agris | if I so much as change the order in which cdrom or -m is specified the system refused to boot | 07:41 |
agris | I"m so confused | 07:41 |
agris | holy crap | 08:08 |
agris | am I crazy? I can not seem to get qemu to freakin work! | 08:08 |
agris | machine type does not support if=scsi,bus=0,unit=0 | 08:08 |
agris | how can a X86 machine not support scsi | 08:09 |
agris | wtf | 08:09 |
agris | snf why is it ignoring -smp 4 | 08:09 |
agris | -cdrom doesn't seem to work either | 08:09 |
agris | ascii doesn't recognize virtio disks | 08:09 |
yeti | -drive file=vda.qcow2,if=virtio | 08:10 |
agris | yeti, yeah, disks don't show up in ascii with virtio | 08:11 |
agris | http://dpaste.com/1THWF0S | 08:11 |
yeti | I installed from netinstall and it works | 08:11 |
yeti | I never used the other devuan ISOs | 08:12 |
yeti | so they may havw glitches no yeti has ever seen before | 08:12 |
yeti | or so... | 08:12 |
agris | so | 08:12 |
agris | system works with just qemu-system-x86_64 -cdrom devuan.iso | 08:13 |
agris | but as soon as you try and start specifying other drives qemu seems to just do whatever it wants | 08:13 |
yeti | every time qemu did not do what I expected, it was PEBCAK | 08:13 |
yeti | maybe I'm just lucky | 08:14 |
* yeti should by some lottery tickets | 08:14 | |
yeti | bUy | 08:14 |
yeti | qemu-system-x86_64 -name devuan2-amd64 -enable-kvm -smp 2 -m 2048 -device virtio-net,netdev=vmnic -netdev user,id=vmnic,hostfwd=tcp::20264-:22 -drive file=vda.qcow2,if=virtio -vga virtio | 08:20 |
yeti | intentionally no sound... | 08:20 |
yeti | maybe someone knows more qemu magic | 08:21 |
yeti | then I'm all ears too | 08:21 |
yeti | I add -cdrom ... or other options as needed... the startscript has "$@" to catch them | 08:21 |
yeti | hostfwd=tcp::20264-:22 is for redirecting ssh | 08:22 |
yeti | into the VM | 08:22 |
rrq | sound needs an env var setting plus a parameter.. as in: QEMU_AUDIO_DRV=alsa qemu-system-x86_64 -soundhw ac97 ... | 08:22 |
yeti | or -soundhw all | 08:23 |
yeti | if lazy | 08:23 |
rrq | sure | 08:23 |
rrq | and "-serial mon:stdio -echr 0x1c" is useful esp if you start a getty on ttyS0 (the -echr is for having ^\ rather than ^A as wemu meta character) | 08:28 |
agris | ah yes | 10:19 |
agris | finnaly got things working | 10:19 |
agris | I had to emulate a megaraid sas controller, put it in SATA compatibility mode, and attach my virtual disks to that | 10:19 |
agris | huh | 10:39 |
agris | I tried to install zfsutils-linux/stable-backports | 10:39 |
agris | zfsutils-linux : Conflicts: insserv (< 1.18) but 1.14.0-5.4+b1 is to be installed | 10:40 |
agris | Shouldn't I be using the same userspace tools as the kernel module version | 10:40 |
agris | ? | 10:40 |
agris | or is there a way to bypass the insserv conflict? | 10:41 |
agris | crap | 10:41 |
agris | This is broken | 10:41 |
agris | cannot create 'bpool': invalid feature 'userobj_accounting' | 10:41 |
agris | if somebody could double check this but the zfs userspace utils for ascii-backports can't actually be installed in ascii | 10:44 |
agris | and yeah | 10:49 |
agris | the minimal love is not UEFI | 10:49 |
agris | bios only | 10:49 |
agris | I'm going to try to dist-upgrade the minimal live and install beowulf manually from ram | 10:51 |
agris | damn | 11:05 |
agris | beowulf's backports aren't built until it's stable | 11:06 |
agris | this isn't going to work | 11:10 |
agris | kernel version conflict | 11:10 |
agris | can't boot into a new kernel without wiping my ramdisk | 11:10 |
gnarface | what are you trying to do now? in theory you could load the new kernel with kexec | 11:13 |
agris | gnarface, I'm just trying to install Devuan with a recent version of ZFS | 11:13 |
agris | gnarface, I can't use ASCII because the backported userspace tools can't be installed due to a dpkg conflict | 11:14 |
agris | and I can't build the beowulf kernel modules without booting into the beowulf kernel | 11:14 |
agris | I can't reboot into the beowulf kernel because my entire beowulf install is only inside the ramdisk. cycling the power will wipe the ramdisk | 11:15 |
gnarface | right | 11:15 |
gnarface | maybe kexec could help | 11:15 |
gnarface | not sure | 11:15 |
agris | gnarface, I've never had a successful kexec. It's always just crashed the system in my past experiences using it | 11:16 |
agris | but Ii'm willing to try | 11:16 |
agris | Is there a Devuan-specific way to kexec the new kernel? | 11:16 |
gnarface | eh, i can't help you with it unfortunately, i don't know | 11:16 |
gnarface | why is the beowulf install only in a ramdisk? not enough harddrive space? | 11:17 |
agris | gnarface, because I'm trying to install Devuan to a zfs root | 11:18 |
agris | and I'm booted into the minimal live iso | 11:19 |
agris | which btw only works in BIOS mode | 11:19 |
gnarface | yea, but... do you not have any spare drive space? | 11:19 |
gnarface | using the minimal iso shouldn't preclude using it | 11:19 |
agris | I do | 11:19 |
agris | I think I broke Devuan anyways | 11:20 |
agris | I installed 4.9 headers in beowulf from ascii | 11:20 |
gnarface | well, that's possible... you really shouldn't be having ascii-backports complaints about ascii... | 11:20 |
agris | gnarface, what do you mean? | 11:20 |
gnarface | you said you had a package conflict and something from ascii-backports couldn't be installed in ascii. you said that right? that should never be the case unless you made a mistake *before* that | 11:21 |
agris | if that's the case maybe the backported package had a major bug and Ii'm the first person to install it | 11:21 |
agris | gnarface, no when I was having that issue I didn't screw the system up at that point | 11:21 |
gnarface | you could be right, it's just highly unlikely | 11:22 |
agris | zfsutils-linux from ascii-backports | 11:22 |
agris | conflicts with ascii's insserv | 11:22 |
gnarface | you sure that's actually ascii's insserv? | 11:23 |
agris | yes | 11:23 |
gnarface | you might have accidentally got the beowulf, don't you think? did you check the version? | 11:23 |
agris | yes, look for yourself https://pkginfo.devuan.org/stage/ascii/ascii/insserv_1.14.0-5.4+b1.html | 11:24 |
gnarface | well, linking me that doesn't prove that's the version you have installed, but i'll take your word for it | 11:24 |
gnarface | i assume you're capable of actually checking, i just wanted to make sure you thought of it | 11:24 |
agris | > zfsutils-linux : Conflicts: insserv (< 1.18) but 1.14.0-5.4+b1 is to be installed | 11:25 |
agris | that does | 11:25 |
gnarface | oh hmm | 11:26 |
gnarface | could it want insserv from backports too? | 11:26 |
gnarface | check to see | 11:26 |
agris | there is no insserv from backports | 11:26 |
gnarface | huh | 11:27 |
gnarface | that does seem like a mistake then | 11:27 |
agris | my guess is that since insserv seems to be related to init scripts | 11:28 |
agris | and Debian....well you know | 11:28 |
rrq | agris: is that ascii-backports/contrib or ascii-backports/main ? | 11:32 |
agris | rrq, ZFS stuff is from contrib | 11:32 |
agris | insserv is main | 11:33 |
rrq | right. yes I can repeat the problem. | 11:34 |
gnarface | rrq: are you also not seeing a newer insserv in ascii-backports? | 11:35 |
rrq | correct. next one is beowulf (and buster) 1.18.0-2 | 11:36 |
rrq | there is however a zfs-util in ascii-bacports/main as well | 11:36 |
rrq | zfsutils-linux | 11:36 |
rrq | the contrib one is 0.7.12-2+deb10u1~bpo9+1 and the main one is (merely) 0.7.3-3+devuan1 | 11:37 |
agris | if this is an actual issue and one related to init would it not make sense to make the ascii backport based on 0.8 rather than 0.7 | 11:42 |
agris | ? | 11:42 |
agris | I'm not really sure why the userspace tools even depend on insserv in the first place | 11:46 |
agris | seems like fixing it may be as simple as removing that as a dependency in dpkg | 11:46 |
agris | should just provide /sbin/zpool and /sbin/zfs | 11:47 |
gnarface | it does seem like maybe either the package needs to simply have the version requirement lowered, or there is something deeper wrong... | 11:52 |
GyrosGeier | 10:19 < agris> I had to emulate a megaraid sas controller, put it in SATA compatibility mode, and attach my virtual disks to that | 12:06 |
GyrosGeier | any particular reason you're not simply using virtio-disk? | 12:06 |
GyrosGeier | these just show up as a single PCI device AFAIK | 12:07 |
agris | it doesn't show up in the OS installer | 12:07 |
GyrosGeier | ah | 12:07 |
GyrosGeier | wait | 12:08 |
GyrosGeier | the Devuan installer ISO should certainly recognize virtio disks | 12:08 |
GyrosGeier | that driver has been in the kernel for ages | 12:08 |
GyrosGeier | "-drive file=/dev/darine/jenkins,format=raw,if=none,id=drive-virtio-disk0,cache=none,aio=native -device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x6,drive=drive-virtio-disk0,id=virtio-disk0,bootindex=1" | 12:10 |
GyrosGeier | that is what libvirt generated for a virtual disk in my Jenkins box | 12:11 |
agris | that's ok it's not something i want to troubleshoot right now | 12:14 |
GyrosGeier | virtio is likely to be faster tho | 12:14 |
GyrosGeier | especially if you also use memballoon | 12:15 |
rrq | agris: fwiw, zfsutils-linux from ascii-backports main (0.7.3-3+devuan1) installs without issue. the other one is a "plain" buster version, and presumably/obviously it needs some hands-on for devuan ascii | 13:07 |
rrq | or, since you might be on the ledge already: you could look at installing 0.8.2-3 from unstable (contrib & main) | 13:16 |
fsmithred | agris, have you talked with mason in #devuan-dev about zfs? He was working on an encrypted zfs-root install back in Sep/Oct. | 13:27 |
agris | no | 13:27 |
agris | isn't encrypted zfs datasets only supported in .8? | 13:27 |
fsmithred | #devuan-dev.log:Oct 01 12:05:51 <mason>Alright, the zfs-initramfs internal docs are wrong, and walking through the code has made everything fairly straightforward. | 13:29 |
fsmithred | I know nothing of what is in which version | 13:29 |
fsmithred | I think maybe he was working with a newer version | 13:30 |
devuanusing9 | Devuan keyring, some keys are expired, what to do? | 14:01 |
fsmithred | what keys are expired? | 14:05 |
devuanusing9 | fsmithred: your key is expired, in the gpg file at files.devuan.org | 14:14 |
devuanusing9 | In the mirror https://mirror.leaseweb.com/devuan/ the same problem happens | 14:15 |
fsmithred | devuanusing9, I just updated it recently. | 14:18 |
devuanusing9 | fsmithred: the expiration key is october 2018, is that correct? | 14:19 |
devuanusing9 | the expiration date sorry | 14:19 |
fsmithred | 4096R/094C5620 2017-10-07 [expires: 2021-09-20] | 14:19 |
devuanusing9 | where can I find the updated key? | 14:19 |
fsmithred | maybe you need to refresh | 14:19 |
fsmithred | you're running gpg on your system to check this? | 14:20 |
devuanusing9 | yes, but on a clean installation | 14:20 |
devuanusing9 | so I just downloaded the file in files.devuan.org | 14:20 |
fsmithred | try gpg --refresh-keys | 14:20 |
devuanusing9 | pub rsa4096 2017-10-07 [SC] [expired: 2018-10-07] 67F5013216271E85C251E480A73823D3094C5620 | 14:20 |
devuanusing9 | uid [ expired] fsmithred (aka fsr) <fsmithred@gmail.com> | 14:20 |
fsmithred | you get that after refreshing? | 14:21 |
devuanusing9 | no, after refreshing, it revoked one of the keys | 14:21 |
devuanusing9 | but the other continues with the old expiration | 14:22 |
fsmithred | oh, maybe we need to refresh on the server | 14:22 |
devuanusing9 | ok | 14:22 |
devuanusing9 | how long will it take more or less? | 14:23 |
fsmithred | rrq, are you still awake? | 14:23 |
fsmithred | depends on how brave and daring I am | 14:23 |
devuanusing9 | ok, ok | 14:23 |
devuanusing9 | do you know any other server where I can download the updated key? pgp.mit.edu seems unresponsive when I search your nick | 14:23 |
fsmithred | damn, that's the keyserver I use | 14:24 |
fsmithred | I pushed the update more than a month ago, so it should have propagated to other keyservers | 14:25 |
devuanusing9 | fsmithred: it's odd, because in the devuan-devs.pgp file , updated 13 Oct 2019, the key it contains is the expired one | 14:26 |
devuanusing9 | devuan-devs.gpg | 14:27 |
devuanusing9 | sha256sum: 5ae0c9c14b0e906a75546c4f4dd9885af07a97d6dc370a55c735f069f89d8c97 | 14:27 |
fsmithred | devuanusing9, it may take a day to get this sorted out because of times zones and sleep cycles. | 14:34 |
fsmithred | mit keyserver not working here, either. | 14:36 |
devuanusing9 | fsmithred: ok, thanks for the info | 14:40 |
devuanusing9 | fsmithred: I hope this does not reduce security of the downloads if verified with old key | 14:41 |
fsmithred | it's the same key as before. Problem is that I updated it and then forgot to push it to the server until after it showed as expired. | 14:43 |
devuanusing9 | ok | 14:45 |
tuxd3v | hello all, dev1ner's spread out in the Galaxy1 :) | 17:27 |
tuxd3v | hello all, I made a v0.1 beta release of devuan for linux sunxi arm64.. | 18:28 |
tuxd3v | its a minimal image | 18:29 |
tuxd3v | but it has the basics working including ethernet.. | 18:29 |
tuxd3v | if someone wants it: | 18:33 |
tuxd3v | https://meocloud.pt/link/af27f314-c7fc-4a5f-a028-037c30e76f85/DEVUAN/ | 18:33 |
tuxd3v | file: devuan-ascii-arm64-0.1-beta.img.tar.gz | 18:34 |
fsmithred | tuxd3v, I'm downloading it, but I probably can't boot it. No arm hardware here and no previous luck trying to boot arm images with qemu. | 18:43 |
tuxd3v | fsmithred: the image inside has: | 18:45 |
tuxd3v | in sector 8, the uboot - bootloader, | 18:46 |
tuxd3v | then at 1024 has 1 /boot partition with kernel, and device tree for sunxi, also boot.src(script for uboot to execute) | 18:47 |
tuxd3v | then at 101MB the rootfs | 18:47 |
tuxd3v | with 1 GB size for rootfs | 18:47 |
tuxd3v | the simple way would be to mount lets say in /mnt | 18:47 |
tuxd3v | the second partition | 18:47 |
fsmithred | yeah, I can do that | 18:47 |
fsmithred | But I can't chroot it. I tried that with rpi image and it complained. | 18:48 |
* fsmithred wants arms | 18:49 | |
tuxd3v | then copy to it the file '/usr/bin/qemu-aarch64-static' , present in package 'qemu-user-static' | 18:50 |
tuxd3v | and only then | 18:50 |
tuxd3v | mount --bind /proc, sys, dev,dev/pts and chroot to it :) | 18:50 |
fsmithred | yeah, did the binds | 18:50 |
tuxd3v | then you can use the binaries inside | 18:50 |
fsmithred | copy what? | 18:51 |
tuxd3v | because ''/usr/bin/qemu-aarch64-static' is in the image and it will translate arm64 to lets say x86 :) | 18:51 |
fsmithred | or I should ask, copy it to where? | 18:51 |
tuxd3v | let me see, but its straightfoward | 18:51 |
tuxd3v | you need to copy /usr/bin/qemu-aarch64-static to the target of chroot | 18:55 |
tuxd3v | in this case would be /mnt | 18:55 |
fsmithred | ah, ok, thanks | 18:55 |
tuxd3v | I was searching in my notes | 18:55 |
tuxd3v | no problem | 18:55 |
tuxd3v | as fas as I know everything will work, even apt-get and so on.. | 18:56 |
fsmithred | still downloading (~350KB/s) | 18:56 |
fsmithred | cool | 18:56 |
tuxd3v | but 'qemu-aarch64-static' needs to be inside, becaue it needs to translate from target binaries aarch64(arm64) to host x86 | 18:56 |
fsmithred | I have to go out soon, but I'll play with this later today. Thanks for making it. | 18:56 |
tuxd3v | no problem | 18:57 |
tuxd3v | :) | 18:57 |
tuxd3v | its always nice to see that something we do is useful :) | 18:57 |
tuxd3v | I am planning to extend it to next stable kernel 5.4 in due time | 18:57 |
tuxd3v | I am testing 5.4-rc7 now | 18:58 |
tuxd3v | but that image ships with stable 5.3.9, which works well | 18:58 |
tuxd3v | everything else it has is standard to devuan | 18:58 |
tuxd3v | ho, it has 1 user more than only root | 18:59 |
tuxd3v | it has 'devuan' user | 18:59 |
tuxd3v | passwd 'devuan' | 18:59 |
tuxd3v | :) | 18:59 |
fsmithred | great, same as the live isos | 19:00 |
tuxd3v | yeah :) | 19:00 |
fsmithred | root pass is toor? | 19:01 |
tuxd3v | no I forgot that... its also devuan :( | 19:01 |
fsmithred | ok | 19:01 |
fsmithred | that would be my second or third guess | 19:01 |
tuxd3v | should I change it to toor? | 19:01 |
tuxd3v | I believe | 19:01 |
fsmithred | yeah, if it's going to end up being a hosted image | 19:01 |
golinux | That has been the convention at devuan | 19:02 |
fsmithred | better to all be the same | 19:02 |
tuxd3v | ok | 19:02 |
golinux | Thanks for your contribution tuxd3v | 19:02 |
yeti | the nonroot user is not a good idea | 19:02 |
tuxd3v | you right, I will update it to toor | 19:02 |
tuxd3v | golinux: no problem | 19:03 |
tuxd3v | it was something I always wanted to help devuan with | 19:03 |
tuxd3v | :) | 19:03 |
tuxd3v | and I have plenty of boards around :) | 19:03 |
tuxd3v | after sunxi boards, I will shift to rockship ones :) | 19:04 |
tuxd3v | already have some work done for it, but that later :) | 19:04 |
tuxd3v | yeti: why you think nonroot is not a good Idea? | 19:05 |
yeti | the user should be free how to name his user 1000 | 19:05 |
tuxd3v | my toughs were that you should login as a normal user, and only then switchs to root | 19:05 |
tuxd3v | maybe you right | 19:06 |
yeti | we networms have user 1000 named differently than devuan on hundreds of other systems... | 19:06 |
yeti | right? | 19:06 |
tuxd3v | maybe I could find a way to create a 1001 devuan user, and when he logs in will be asked by the new user to create as 1000 ? | 19:07 |
golinux | We need to set up that space on the d1g forum. | 19:07 |
yeti | that would have been my one of my next lines | 19:07 |
yeti | :-) | 19:07 |
fsmithred | with the live isos, you can change the user during the installation | 19:09 |
fsmithred | since this image is just going to be dd'd, there's no chance to do that | 19:09 |
yeti | debian has an installer for sunxi images | 19:09 |
yeti | but i dont know where its sources are | 19:10 |
fsmithred | yeah, we don't have that yet, but there's been some talk about doing it | 19:10 |
fsmithred | ...eventually | 19:10 |
yeti | it still may need a usb-uart-cable | 19:10 |
fsmithred | what's that for? | 19:11 |
yeti | it did not recognise the usb-keyboard | 19:11 |
fsmithred | oh | 19:11 |
tuxd3v | we could adapt it to devuan, I don't know if its feasible, but if it his could be a option | 19:11 |
yeti | I tried to install a banana pi r1 with it | 19:11 |
yeti | but without keyboard... :-/ | 19:11 |
yeti | arm stuff @debian is an adventure too | 19:11 |
tuxd3v | hehe, life is an adventure :) | 19:12 |
yeti | next time i open the BPI-R1, I'll connect rx/tx/gnd | 19:12 |
tuxd3v | yeti: yes to that, it could be that it works as serial for stdout and for stderr | 19:14 |
tuxd3v | maybe it could be that way you set a new user.. | 19:14 |
fsmithred | I can't chroot the mounted image. Getting "failed to run command '/bin/bash': Exec format error | 19:15 |
fsmithred | which is what I got before adding qemu-aarch64-static | 19:15 |
yeti | 2.HHGTTG 18,91: ''Will you shut up and listen!'' hissed Zaphod, ''this time there's going to be excitement and adventure and really wild things.'' | 19:15 |
yeti | found it! | 19:15 |
tuxd3v | yes | 19:15 |
tuxd3v | its the qemu user static for aarch64 that needs to be inside :) | 19:16 |
fsmithred | yeah, I copied that to the mounted image | 19:16 |
tuxd3v | it tries to interpret arm64 binaries in your host target arch and gives you that error becasue they are diferent archs | 19:16 |
fsmithred | do I need to repack the image and boot it with qemu? | 19:16 |
tuxd3v | no only chroot is enough | 19:17 |
tuxd3v | let me check again the location of the quemu binary | 19:17 |
fsmithred | ok | 19:17 |
fsmithred | got it! | 19:19 |
tuxd3v | maybe you need to copy it to '/usr/bin/' | 19:19 |
fsmithred | chroot mnt /qemu-aarch64-static /bin/bash | 19:19 |
tuxd3v | like in the origin host | 19:19 |
tuxd3v | yes | 19:19 |
tuxd3v | that is a way to run a command inside | 19:19 |
fsmithred | oy | 19:20 |
fsmithred | every command | 19:20 |
tuxd3v | try copy 'qemu-aarch64-static' to chroot '/usr/bin' | 19:20 |
tuxd3v | I believe that after that you should be able to work with any command | 19:21 |
fsmithred | ok | 19:21 |
tuxd3v | whithout having to type the qemu inary everytime | 19:21 |
yeti | binfmt_misc support is installed? | 19:22 |
tuxd3v | I confirm qemu binary needs to be to chroot '/usr/bin' | 19:23 |
fsmithred | yeah, I saw some binfmt package get installed along with the aarch64-static | 19:23 |
tuxd3v | I will try from my side too :) | 19:24 |
tuxd3v | so that I could iron out any problem.. | 19:24 |
fsmithred | weird. 'echo' works, 'ls' and 'cp' do not work. | 19:25 |
yeti | ls /proc/sys/fs/binfmt_misc/ | 19:26 |
yeti | is aarch64 mentioned there? | 19:26 |
fsmithred | nope | 19:26 |
fsmithred | I'm looking on the host | 19:26 |
yeti | should be the same | 19:27 |
yeti | proc would be bindmounted | 19:27 |
fsmithred | oh yeah | 19:27 |
yeti | list the contents of the qemu-aarch-thingy | 19:27 |
yeti | look for a config file for binfmt_misc | 19:28 |
yeti | e.g. mono-runtime installs some such stuff to transparently execute DOnet binaries | 19:28 |
yeti | doTnet | 19:28 |
fsmithred | no config files in that package | 19:29 |
yeti | /usr/share/binfmts | 19:29 |
fsmithred | python2.7 and python3.5 | 19:30 |
yeti | if there were aarch64, it probably would show in /proc/... too | 19:30 |
MinceR | botNET | 19:30 |
yeti | time to ask marjorei | 19:30 |
tuxd3v | maybe some dependency, I entered directly and executed a apt-get update, it works nicely | 19:30 |
tuxd3v | let me see dependencies needed.. | 19:31 |
yeti | I only vaguely remember that fro executing an i386 binary on PI and Cubie3 | 19:31 |
tuxd3v | I found this dependencies: 'binfmt-support' 'qemu-user-static' | 19:32 |
tuxd3v | I also mounted everything with '--bind' | 19:32 |
tuxd3v | proc, sys, dev, dev/pts | 19:33 |
yeti | and you can transparently start aarch64 binaries without prefixing it with qemu-...? | 19:33 |
tuxd3v | yeti: yes | 19:33 |
yeti | fsmithred may not have all deps | 19:34 |
fsmithred | yeah, I'm adding them now | 19:34 |
yeti | ok | 19:34 |
fsmithred | binfmt-support added. Do I also need qemu-user-binfmt? | 19:34 |
yeti | yip | 19:34 |
fsmithred | um | 19:35 |
fsmithred | that wants to remove qemu-user-static | 19:35 |
tuxd3v | I think not, only thos 2 | 19:35 |
tuxd3v | those | 19:35 |
tuxd3v | root@desktop0:/# file /bin/ls | 19:35 |
tuxd3v | /bin/ls: ELF 64-bit LSB shared object, ARM aarch64, version 1 (SYSV), dynamically linked, interpreter /lib/ld-linux-aarch64.so.1, for GNU/Linux 3.7.0, BuildID[sha1]=a242161239cb1f4ea9b8a7455013295e5473e3ec, stripped | 19:35 |
tuxd3v | root@desktop0:/# ls -l / | 19:35 |
tuxd3v | total 121500 | 19:35 |
tuxd3v | drwxr-xr-x 2 root root 4096 Oct 30 04:05 bin | 19:35 |
tuxd3v | drwxr-xr-x 2 root root 4096 Jun 10 2018 boot | 19:35 |
fsmithred | ok | 19:35 |
tuxd3v | drwxr-xr-x 15 root root 3240 Nov 11 16:55 dev | 19:35 |
tuxd3v | -rw-r--r-- 1 root root 124346447 Nov 11 18:26 devuan-ascii-base-arm64.tar.gz | 19:35 |
tuxd3v | drwxr-xr-x 44 root root 4096 Nov 11 18:34 etc | 19:35 |
tuxd3v | drwxr-xr-x 2 root root 4096 Jun 10 2018 home | 19:35 |
tuxd3v | drwxr-xr-x 10 root root 4096 Oct 30 04:05 lib | 19:35 |
tuxd3v | drwxr-xr-x 2 root root 4096 Oct 30 04:03 media | 19:35 |
tuxd3v | drwxr-xr-x 2 root root 4096 Oct 30 04:03 mnt | 19:35 |
tuxd3v | drwxr-xr-x 2 root root 4096 Oct 30 04:03 opt | 19:35 |
tuxd3v | dr-xr-xr-x 218 root root 0 Oct 27 21:27 proc | 19:35 |
tuxd3v | drwx------ 2 root root 4096 Oct 30 04:03 root | 19:35 |
tuxd3v | drwxr-xr-x 4 root root 4096 Oct 30 04:03 run | 19:35 |
tuxd3v | drwxr-xr-x 2 root root 4096 Oct 30 04:07 sbin | 19:35 |
tuxd3v | drwxr-xr-x 2 root root 4096 Oct 30 04:03 srv | 19:35 |
tuxd3v | dr-xr-xr-x 13 root root 0 Oct 28 17:45 sys | 19:35 |
tuxd3v | drwxrwxrwt 2 root root 4096 Nov 11 18:34 tmp | 19:35 |
tuxd3v | drwxr-xr-x 10 root root 4096 Oct 30 04:03 usr | 19:35 |
tuxd3v | drwxr-xr-x 11 root root 4096 Oct 30 04:03 var | 19:36 |
tuxd3v | sorry for the spamm | 19:36 |
tuxd3v | I should tried a smaller command .. | 19:36 |
tuxd3v | I am sorry | 19:36 |
fsmithred | oh! | 19:37 |
tuxd3v | then you needto have '/usr/bin/qemu-aarch64-static' in chroot same place '/usr/bin/qemu-aarch64-static' | 19:37 |
fsmithred | I know why it keeps telling me 'no such file or directory' for ls | 19:37 |
yeti | https://forums.parallax.com/discussion/166691/run-bstc-on-cubietruck-raspberry-pi | 19:38 |
yeti | maybe something changed since wheezy days... | 19:38 |
yeti | so my memories may be outdated | 19:38 |
fsmithred | ok, I have to precede the command with qemu-aarch64-static and I have to use the path to the command | 19:39 |
tuxd3v | yeti: no need to prefix qemu binary, or even to use it at all, after it been in place ;) | 19:39 |
tuxd3v | fsmithred, that is the first aproach , and it works, but its a bit time consuming | 19:39 |
tuxd3v | the next step is without prefixes | 19:40 |
fsmithred | how do I get there? | 19:40 |
tuxd3v | you doesn't even notice that you are in a chroot environment | 19:40 |
tuxd3v | root@desktop0:/# find / -name qemu-aarch64-static | 19:41 |
tuxd3v | /usr/bin/qemu-aarch64-static | 19:41 |
tuxd3v | I only have 1 binary | 19:41 |
tuxd3v | there | 19:41 |
tuxd3v | and I haven't set any env var.. at least that I now, I simply uncompressed the image, and chrooted into it, copy qemu to inside of it, and its ok | 19:42 |
fsmithred | alright, I'll play with this later | 19:42 |
fsmithred | gotta go | 19:42 |
fsmithred | Thanks for help. | 19:42 |
tuxd3v | fsmithred, no problen | 19:43 |
tuxd3v | :) | 19:43 |
tuxd3v | fsmithred: I thik I got it: | 19:50 |
tuxd3v | sudo update-binfmts --display|grep -EA 7 "aarch64.*enabled" --color | 19:50 |
tuxd3v | check if arch aarch64 is enabled with update-binfmts.. | 19:51 |
tuxd3v | in the host | 19:51 |
cgdae | could anyone here explain how to install devuan unstable (aka beowulf)? i haven't been able to find info on devuan.org. i did find http://pkgmaster.devuan.org/devuan/dists/unstable/main/installer-amd64/current/images/netboot/mini.iso, but when installing, it said 'no kernel modules found' when doing 'download installer components'. | 20:39 |
emdete | cgdae: i use the life-iso, install this minimal system, reboot, change the apt sources and do an upgrade... | 20:48 |
cgdae | emdete: ah, thanks. does that mean that there isn't a way to install beowulf directly (without going via stable or similar)? | 20:51 |
emdete | cgdae: i'm not sure but because i like to start with a minimal system that was the best way i found. i even looked for how that iso can be build to strip it even more ;) | 21:04 |
golinux | cgdae: Yes, you'll want to upgrade from ascii by changes the sources list to beowulf which hasn't been officially released yet. | 21:29 |
golinux | Depending on your setup. it could go more or less smoothly | 21:30 |
emdete | golinux: can you tell how to build the life iso? | 22:12 |
golinux | Depends how you want to build it | 22:13 |
golinux | From a running desktop you can use refractasnapshot. I should be in the Devuan repos. Readme is here: https://git.devuan.org/devuan-doc/documentation/tree/master/dev1fanboy | 22:15 |
golinux | I > it | 22:15 |
golinux | There is also the live-sdk: https://devuan.org/os/distro-kit | 22:16 |
golinux | emdete: ^^^ | 22:17 |
golinux | Oops clipboard misfire | 22:26 |
golinux | Try https://refracta.org/docs/readme.refractasnapshot.txt | 22:26 |
fsmithred | emdete, it's possible to do a debootstrap install from the live isos. That would be the leanest possible install. | 22:50 |
emdete | golinux: do you suggest to use refracta or distro-kit? | 22:54 |
fsmithred | emdete, it depends on what you want to do | 22:56 |
mason | cgdae: I've been doing straight Beowulf installs recently. Vaguely similar to the Refracta method, I have a generic install with the tools I want on a portable USB hard drive (and again on a VM I can attach to virtual disks, for virtual installs) and I go from there, debootstrap driven by scripting. | 22:57 |
mason | I should do a pxe version and then I could just have one. | 22:58 |
emdete | creating a minimalistic boot usb stick which is able to be (unattended) installed to hd if desired. | 22:59 |
emdete | golinux: refracta seems to provide that, i already use that with a patched script in a second partition.but i would like to put that script into the iso and remove some packages... | 23:00 |
fsmithred | ok, refractainstaller is not an unattended activity, but you could make your own script to do that. | 23:00 |
mason | emdete: Things to look at that could be adapted: https://wiki.debian.org/DebianInstaller/Preseed https://wiki.debian.org/FAI | 23:01 |
fsmithred | refractasnapshot is a WYSIWYG remastering tool. It copies the running system. | 23:01 |
emdete | fsmithred: which i did based on an older version (someone here provided that) | 23:01 |
emdete | mason: fai is a beast 😉 | 23:01 |
fsmithred | emdete, what exactly do you mean by "unattended" | 23:02 |
fsmithred | ? | 23:02 |
emdete | does live-sdk include any kind of installer to copy to hd? | 23:03 |
fsmithred | no | 23:03 |
fsmithred | unless you tell it to include refractainstaller | 23:03 |
fsmithred | a few people have gotten it to include debian-installer, but I'm not one of those people | 23:03 |
emdete | fsmithred: i start a script and it installs with no further interaction | 23:03 |
mason | emdete: You can script up debootstrap for something similar. Half a sec and I'll pastebin my most recent cut - note that it prompts for root password, dialog for locale, etc. | 23:03 |
fsmithred | ok, that's pretty easy to do if you know what partition you're going to use | 23:04 |
fsmithred | and a few other things | 23:04 |
fsmithred | you could lift a few lines out of refractainstaller and make your own | 23:04 |
emdete | fsmithred: its for notebooks in a school. the kids boot from stick, run the installer, target is more or less same, same | 23:04 |
mason | emdete: https://bpaste.net/show/ZJFEY | 23:05 |
emdete | yes, i already did so. i just wasnt sure how to build the iso | 23:05 |
fsmithred | for a computer class? | 23:05 |
mason | emdete: You could install via netboot too, and skip the media distribution. | 23:05 |
fsmithred | can you let the kids hack on the script? | 23:05 |
fsmithred | if you have an installation already the way you want it, easiest way to make iso is refractasnapshot | 23:06 |
emdete | sure, but why? its only the installation phase. after that they boot from HD into a console-only linux and start from there to put things together | 23:07 |
emdete | ok. this is good to know. and it seems they have just deb packages. nice. | 23:08 |
emdete | thank you all ! :) | 23:08 |
fsmithred | I say make them all do a debootstrap install and they can learn a few things | 23:08 |
mason | Oh, need to update the script's sources.list. Whoops: https://bpaste.net/show/5TIM2 | 23:08 |
emdete | but your script looks interesting too - you use efi, zfs and crypto... will check what i can steal from that 🙂 | 23:08 |
mason | emdete: I like the idea of not having as much automation and guiding the students through a hand install. | 23:09 |
mason | emdete: Good way for them to learn. | 23:09 |
fsmithred | agris, take a look at mason's script linked above ^^^ | 23:10 |
mason | emdete: Maybe look at a private/local mirror if you're going to have more than a handful of students downloading packages all at once. | 23:10 |
emdete | yes, it's wonderfull if they are stuck in front of a console | 23:10 |
emdete | and the eyes when you show them youtube-dl and mpv with -vo caca is just great :D | 23:11 |
fsmithred | lo | 23:11 |
fsmithred | l | 23:11 |
mason | fsmithred: Once I'm properly happy with the script I'll put it on github or something. Also, it'd be worth having a legacy mode for folks who don't want ZFS. | 23:11 |
fsmithred | cool. agris had questions about zfs yesterday | 23:12 |
emdete | mason: ah, they install step by step, the current class is 7 people and they install python or such - mostly a single package. but that's a good hint to speed up things... | 23:12 |
mason | fsmithred: Ah, I somehow missed that I hadn't rejoined #devuan after the last time I parted, although I remembered #debianfork. | 23:12 |
mason | agris: I'd be happy to help. Just ping me. | 23:13 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!