How to use kernel parameters with autoinstall? #cloudinit #autoinstall #subiquity
How to use kernel parameters with autoinstall? #cloudinit #autoinstall #subiquity
I seem to have hit some kind of weird bug with Ubuntu's autoinstall.yaml
and Ubuntu Server 24.04. It refuses to create the user for identity:
. I have to then define a legacy cloud-init user-data:
hash under autoinstall:
, except it refuses the set the password and instead leaves the user locked with a !
in /etc/shadow
. I've tried lock-passwd: false
, lock_passwd: false
, and lock_passwd: False
to no avail.
Lazy FreeBSD: where is the documentation or example cloud-init user-data
for creating users, setting passwords, adding SSH pubkeys, and installing packages specifically for a FreeBSD system?
I created a cloud-init .iso
file containing my meta-data
and user-data
files, and dd
ed the BASIC-CLOUDINIT
FreeBSD image onto the device, and booted it, and it did set the hostname and created my users, but without any passwords or SSH keys...
I sort of want to convert my DIY lab VM setup shell script to cloud-init
, but it doesn't seem to have any support for populating /etc/crypttab
and the fact that you have to embed files into YAML as base64 is beyond gross; YAML has a |
operator for indented text blocks, but keeping files separate means that editing and syntax highlighting work normally.
@dalias @eniko but to answer the original post: Yes, and most #server #distros should support that using i.e. #CloudInit or you just setup the server in a #VM to your liking and then just dd
that disk image to a USB drive.
@SitaDulip AFAIK @ubuntu / #UbuntuLTS Server should launch an #SSH Server to access it's installer at boot.
Basically you modify the #CloudInit configuration file.
Same can be done with Debian.
Either way, you'd setup a minimalist system as a bootable image you can #SSH into and yoink that onto the target system to boot.
dd
the virtual HDD into an image you can then dd
onto the target machine.If time isn't critical, any halfway decent flashdrive will do, otherwise a 2,5" SSD (in a USB Caddy) is an option as well...
Sur mon blog, je parle de #proxmox, #opentofu #debian et #cloudinit :
https://xieme-art.org/post/instancier-une-machine-virtuelle-debian-dans-proxmox-avec-opentofu-et-cloud-init/
RaspberryPi OS, Rant (swearing, well deserved)
Seeking advice!
I’m exploring (OSS) cloud management solutions like #OpenStack, #CloudStack, #Eucalyptus, #oVirt, #ProxmoxVE, #DigitalOcean, #OpenNebula, and #CloudInit, specifically looking for billing/accounting features.
@OpenStack provides #Ceilometer and #CloudKitty as starting point, what about (the) others?
Any additional recommendations? #CloudComputing #Billing #Accounting #OSS
Boosts appreciated!
Hint: use adduser, which is perl wrapper around useradd when working interactively and useradd from scripts, as it is more portable.
(And if you’re setting up a server in an automated fashion, let cloud-init, etc., handle it for you.)
I first thought #cloudinit is wonderful. But… https://gist.github.com/Informatic/0b6b24374b54d09c77b9d25595cdbd47 Yes the documentation is cancer. #cloud #linux
If you're running #FreeBSD, #NetBSD, #OpenBSD or any other #BSD in a #cloud environment (VM or hardware), I'd love to see pastes from your ifconfig -a
and netstat -r
(with --libxo=json
on FreeBSD) — especially if they're unusual, or "messy"
I'm trying to write a definite ifconfig
parser for #CloudInit
Ansible role: virt_install_vm 1.0.0 released
https://stafwag.github.io/blog/blog/2021/09/19/ansible-role-virt_install_vm_1.0.0_released/
I wrote a few articles how to use cloud images with cloud-init on a “non-cloud” environment.
I finally took the time to create an Ansible role for it. You’ll find the READE.md below.
Virt_install_vm 1.0.0 is available at: https://github.com/stafwag/ansible-role-virt_install_vm
Got the following error on multipass (beta channel) version 0.5:
cannot connect to the multipass socket
If you encounter the same thing, refreshing the snap to version 0.5-27-g49baef0 fixed it for me:
sudo snap refresh multipass --edge --classic
Likely related to this issue: https://github.com/CanonicalLtd/multipass/issues/555
If you want to know what multipass is or how it will be used in #hypha, see https://ar.al/2019/01/05/hypha-spike-deployment-1/
Starting off my Friday morning off on the right foot apparently. Just wrapped up the translations from Terraform RemoteExec provisioning to ignition for the Kube v1.11 upgrade on the linuxlab cluster.
That's one controller down, now to do the workers and prep for HA.