14 Oct 2021 |
jochensp | michael: do you haven an idea here? ^ | 18:50:41 |
Izzy | OK, I'd say give it until tomorrow. When it showed up, but (as expected) does not ping the Google accounts, I repeat the toot with @twitter.com instance added. | 19:29:21 |
FestplattenSchnitzel | jochensp the libvirt or the virtualbox one? Emanual said something like "it might not work due to qemu conversion". I'll take a look. | 19:39:58 |
jochensp | libvirt | 19:40:15 |
jochensp | the conversion works here | 19:40:38 |
jochensp | the the vagrant_box_image is 1T but when doing vagrant up it seems to put a 20G image on top | 19:41:21 |
FestplattenSchnitzel | Sounds weird. Have you tested Virtualbox? | 19:43:08 |
jochensp | trying to set it up now | 19:43:29 |
jochensp | with virtualbox tests run fine and the image is 1T | 19:50:08 |
FestplattenSchnitzel | Great! | 19:50:32 |
jochensp | FestplattenSchnitzel: I think I found the libvirt problem: https://salsa.debian.org/cloud-team/debian-vagrant-images/-/blob/master/utils/vagrant/libvirt/metadata.json#L4 | 19:59:43 |
jochensp | FestplattenSchnitzel: can you change that to 1000 in your PR? (I still need to run resize2fs /dev/vda1 in there, though) | 20:07:25 |
FestplattenSchnitzel | Yep, did that. | 20:23:37 |
jochensp | thx | 20:24:55 |
FestplattenSchnitzel | I also found out, that this parameter is actually needed:
Virtual size has to be set for allocating space in storage pool.
https://github.com/vagrant-libvirt/vagrant-libvirt/blob/master/lib/vagrant-libvirt/action/handle_box_image.rb#L137
| 20:25:13 |
jochensp | hm.. the image I generated on the gcc host actually hangs in grub rescue (not at "Waiting for domain to get an IP address...") | 20:33:05 |
jochensp | I will upload my locally generated image | 20:33:43 |
jochensp | my findings for today: vagrant up --provider=libvirt on gcc hangs in grub (same image works on my PC). qemu parameters looks reasonably similar, manually running qemu with the image works. qemu is the same version (modulo bullseye-backports, libvirt is the same version (installed from testing to match my unstable) | 22:03:20 |
jochensp | I upgraded the gcc box to bullseye without problems and fdroid --server still works | 22:04:52 |
jochensp | also dpkg -V shows nothing and I did a manual search for extra files on the system (found /etc/libvirt/qemu/buildserver_default.xml, /etc/modprobe.d/kvm-intel.conf, /etc/profile.d/vagrant_disable_symlinks_in_shared_folders.sh, /usr/local/bin/enable_kvm_nesting.sh) but nothing seems relevant | 22:12:37 |
jochensp | also nested kvm is enabled | 22:12:45 |
jochensp | michael: if you have any idea or time to discuss ^ | 22:13:11 |
15 Oct 2021 |
| 🏔️hik changed their profile picture. | 02:04:41 |
| mimi89999 left the room. | 04:04:34 |
| mimi89999 joined the room. | 04:07:05 |
uniq | jochensp did you try connecting with virt-manager for debugging? | 06:53:16 |
jochensp | michael: yes, I get a grub with "attempt to read or write outside of disk" | 06:53:34 |
jochensp | running the same xml file with virsh create works, though | 06:53:59 |
jochensp | michael: got it running :) | 07:07:45 |
jochensp | seems like vagrant(?) creates a wrong qcow2 image for the VM. I ran vagrant up --no-destroy-on-error; pkill -f qemu; virsh edit foo_default and changed the disk image from foo_default to the base image | 07:10:41 |