VPNer's questions

I had a bare metal Ubuntu Server installation. I created a full disk image with dd and converted to qcow2.

qemu-img convert -O qcow2 full_disk.dd.img full_disk.qcow2

Then I have installed a fresh Ubuntu Server 22.04 on this machine. Later I wanted to import the disk with virt-install:

virt-install --name guest-host --memory 16384 --vcpus 4 --disk path=/var/lib/libvirt/images/full_disk.qcow2,format=qcow2 --os-variant ubuntu22.04 --network network=default --graphics none --import

After this a "console" starts with no output. qemu-system-x86 runs with 100% CPU (on a single core) almost indefinitely. I can connect with:

virsh --connect qemu:///system console guest-host

but still no output. (I can exit with Ctrl+])

virsh edit guest-host

contains:

    <serial type='pty'>
      <target type='isa-serial' port='0'>
        <model name='isa-serial'/>
      </target>
    </serial>
    <console type='pty'>
      <target type='serial' port='0'/>
    </console>

--extra-args="console=ttyS0"

argument is not accepted for imports.

ERROR Kernel arguments are only supported with location or kernel installs.

I tried to edit the grub.cfg to use the serial output but I do not get any output.

I also tried to import to VirtualBox. There, it is stuck at "Booting from Local Disk...".

It is puzzling that it is using constant 100% of the CPU. At some point the VM should be idle. VirtualBox can't boot. I can't see if the VM is doing anything under KVM (since no output).

In order test on a graphical (virt-manager) environment, I tried the same steps on my notebook. Seems like I can't change the firmware and chipset settings.

Progress: in virt-manager, there is an option to change configuration after selecting OS and the disk image. This allowed me to change the firmware (boot) to UEFI (there are a few options, among which /usr/share/OVMF/OVMF_CODE_4M.fd worked and I was able to boot to the graphical output. However, in virt-install (in bash, no GUI) or virsh I was not able to see or access the output. CPU usage isn't 100% anymore. There seems to be some activity. These are all good signs. There is an error: error: no suitable video mode found.. OS might still be booting. I am looking into the network modes to see if SSHD is starting.

Another option is to configure grub.cfg for serial output.

Final: I accepted @algebra's answer. This really solved the booting issue. I was not able to get any output. I decided to install lightdm and ubuntu-desktop to get a desktop environment, where I was able to access the imported OSes. It also turns out that the network devices might be named differently. To get access, you can either edit /etc/netplan/... (permanent) or manually activate the network device (something something up but not permanent). Then you can discover the VM guests on the virtual network with nmap -sn <subnet>. This way you can ssh to the VMs.

PS: I have deleted the older question because now I think that it isn't booting at all.

I have an Ubuntu server on bare metal that was set up by a third party. It runs a proprietary service. There is no proper documentation about what is running and how.

I had to repurpose this server for another service, an Ubuntu Server with Docker containers running the new service.

I need to run the previous proprietary service that was running on the bare metal server. I also need to run it in a container for easier deployment and management.

I am not sure how to move an Ubuntu Server to containers.

I know that this is not the "Docker way" to do this.

One approach I found is to create an image from scratch and transfer the file system except for the following folders: /proc, /sys, /dev, /tmp, /mnt, /media, /run, /boot, and /swapfile

FROM scratch
ADD folders-to-transfer.tar.gz /
CMD ["/bin/bash"]

Is this the correct way?

Others suggested using a VM, which sounds like overkill.

The application stack has multiple components. e.g. api, public application and internal application. I want to make internal application only accessible from internal network IP ranges. I configured my nginx (reverse proxy) to allow only certain IP ranges (e.g. 192.168.1.0/24). I am using a domain name: mydomain.com

Let's say that the internal application is using the subdomain app.mydomain.com There is also a website (running externally) under (www.)mydomain.com The external DNS (authority) is configured to point to my server's public static IP, which is needed for the public API (e.g. api.mydomain.com).

But on the local network (incl. VPN), I want to direct the traffic internally. For instance, instead of the public IP, PCs should use the local IP, so that only those with local IP will be allowed by the nginx reverse proxy. To do this, I am configuring the Windows AD server also running a DNS Server. I am trying to configure this DNS to direct computers in the network to the server on the local network.

When I try to define a primary zone for mydomain.com with an entry for app.mydomain.com to point to the internal server address, nslookup app.mydomain.com is returning the local IP but mydomain.com or www.mydomain.com do not work anymore.

Is there a way to point for all the other subdomains to the external DNS (authority)?

When I try to configure a secondary zone, the transfer fails. I think that the authority (external DNS) does not let me to do that.

I wonder how can achieve this?