After replacing id_rsa.pub w/ id_ecdsa.pub, I ran "make build/vm/ubuntu" which downloaded a bunch of stuff and eventually failed w/:

(initramfs) [6ndd: invalid number: '1m'

I followed the README on a MacBook Air M1 in order to start an ubuntu VM. Nothing was printed after executing vmctl start ubuntu and vmctl attach ubuntu was not able to find a screen. A new crash report was created with the information:

Generic Platform
FIXME: "Unhandled system register S3_0_C0_C6_2"

I am running macOS 12.2.1

When trying to boot a Debian 10.8 installation, boot process got stuck (picture below). MacBook Pro 2020, M1 Debian 10.8 Linux-Kernel 4

vmcli --kernel=linux --initrd=initrd --cpu-count=4 --memory-size=3072 --disk=disk.img --cmdline='console=tty0 console=hvc0 root=/dev/mapper/vm--vg-root ro apparmor=0' --network=nat

I continue to get the following error:

❯ sudo make build/vm/ubuntu mkdir -p build mkdir -p build/vm mkdir -p build/vm/ubuntu cd build/vm/ubuntu && ../../../vmbuilders/ubuntu.sh % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 11.1M 100 11.1M 0 0 3169k 0 0:00:03 0:00:03 --:--:-- 3168k gunzip: vmlinux.gz: not in gzip format make: *** [build/vm/ubuntu] Error 1

Hi there,

I followed the instructions in the readme and tried to increase the vm volume to 16G by: dd if=/dev/null of="${VMCTLDIR}/ubuntu/disk.img" bs=1g count=0 seek=16

I can see that in my vm folder the disk.img size increase to 16G

total 3171224 -rw-r--r-- 1 ctsai staff 17179869184 Sep 13 10:51 disk.img drwx------ 3 ctsai staff 96 Sep 12 15:00 screen -rw-r--r-- 1 ctsai staff 13 Sep 7 11:17 0.ipaddr -rw-r--r-- 1 ctsai staff 128 Sep 7 10:25 vm.conf drwxr-xr-x 7 ctsai staff 224 Sep 7 10:20 ubuntu -rw-r--r-- 1 ctsai staff 13 Jun 14 08:54 0.macaddr -rw-r--r-- 1 ctsai staff 851 Jun 14 08:53 user.yaml -rw-r--r-- 1 ctsai staff 27621838 Jun 14 08:52 initrd -rw-r--r-- 1 ctsai staff 13672704 Jun 14 08:52 vmlinux

but when I start the vm I only see 1.4G storage:

Welcome to Ubuntu 20.04.4 LTS (GNU/Linux 5.4.0-117-generic x86_64)

• Documentation: https://help.ubuntu.com
• Management: https://landscape.canonical.com
• Support: https://ubuntu.com/advantage

System information as of Tue Sep 13 17:53:47 UTC 2022

System load: 0.05 Usage of /: 98.9% of 1.40GB Memory usage: 9% Swap usage: 0% Processes: 109 Users logged in: 0 IPv4 address for enp0s1: 192.168.65.3 IPv6 address for enp0s1: fd73:e176:203:8e05:4c8c:42ff:fe46:31b8

=> / is using 98.9% of 1.40GB

0 updates can be applied immediately.

The list of available updates is more than a week old. To check for new updates run: sudo apt update

Last login: Mon Sep 12 22:00:34 UTC 2022 on hvc0

Can anyone point me in the right direction here? Thanks

./build/vmctl list ● stopped Applications ● stopped Library ● stopped System ● stopped Users ● stopped Volumes ● stopped bin ● stopped cores ● stopped dev ● stopped etc ● stopped home ● stopped opt ● stopped private ● stopped sbin ● stopped tmp ● stopped usr ● stopped var

I have create two guest VMs(w1, w2)and can ssh into both of them from the host Mac. But I can't ping or anything from w1 to w2 or from w2 to w1. Please someone help me to resolve this issue.

Network configuration

• Host Mac: 192.168.64.1
• w1: 192.168.64.8
• w2: 192.168.64.9

Ping test result

• Success: host Mac -> w1, w2
• Success: w1, w2 -> host Mac
• Fail: w1 -> w2 (Destination Host Unreachable)
• Fail: w2 -> w1 (Destination Host Unreachable)

HTTP test result (via curl)

• Success: host Mac -> w1, w2
• Success: w1, w2 -> host Mac
• Fail: w1 -> w2, w2 -> w1
• Success: w1 -> w1, w2 -> w2

I have once encountered cloud init error after restarting Ubuntu, probably due to bug or wrongly remove essential packages would cause major problematics and not able to SSH with vmctl.

Linux partition could be customize with separate home image or allow user definite separate partition would be essential to avoid data loss in case we screw up?

Or a shared folder we can work from Linux and macOS is a neat feature.

Hi,

Firstly, thanks for this project. Extremely useful. 😄

I have noticed an issue, and found the root cause, but I'd like to discuss the direction of any PR.

The issue:

When running vmcli on main with the networking portion left in, the VM never starts and silently fails. If I remove the networking code, then vmcli starts the VM with no issues (except no networking, obviously).

Debugging this, found the following log in console: default 08:55:10.482714+0000 InternetSharing VMNet in NAT mode can't be started now due to an active InternetSharing session on en0.

The fix:

Disable InternetSharing. Simple as that. Then everything works as expected (with a NAT device, haven't tested bridge mode).

How should we "fix" this? A note in documentation? Or a check in the code for this scenario (if possible?)?

This extends the --folder argument to allow optionally specifying the virtofs tag and a read-only option. This makes it easier to share host directories whose paths might exceed the 36-byte limit for tags, by using a shorter tag instead of the path. While extending the folder argument parsing, add support for read-only shares as well.

This is supported only on Monterey (12.0) and later.

In my limited tests with 12.1, the underlying macOS virtiofs implmentation seems pretty flaky. For example, some basic git operations fail in the guest, but work fine in the host. But offering up the PR now anyway, in the hope that this becomes more robust in later releases.

Closes #20.

In order to enable support for multiple different OS's and first boot setups, this commit rewrites VMCTL in swift from shell script.

This rewrite enables support for other operating systems and first boot setups via the Provider protocol. For example, the default provider used now is the UbuntuProvider which mimics the setup from the vmbuilders/ubuntu.sh script. The other provider implemented is the UbuntuDockerProvider. This one sets up an Ubuntu instance with a few packages and other niceties ready to go for using it as a docker host.

In addition to the Provider protocol, there are now templates along with each provider. These templates describe the various cloud-init yaml files needed to set up a virtual machine on first boot. By using templates we can support quite a bit of variety in the setup and provide specific information to them via the implementing Provider.

Lastly, this rewrite also adds the ability to create new virtual machines via the vmctl create command. This command will download and create all of the files necessary for a new virtual machine.

If we get an escape-q sequence, SIGINT or SIGTERM, request the VM to stop (i.e. graceful shutdown, simulates pressing the power button) and wait for a configurable timeout before forcing a shutdown. This should let most Linux VMs shut down properly when quitting. Sending SIGPIPE or SIGHUP still offers a way to exit the VM immediately/uncleanly.

This also fixes issues with the signal handling where some signals weren't handled as intended (signals can't be bitwise ORed together).

I also had to remove the @available(macOS 12, *) on line 136 as it was causing a compilation error: stored properties cannot be marked potentially unavailable with '@available' when compiling with Xcode 14 on macOS 12.6. I'm not normally writing Swift, so please double-check this :)

Apple has an example for virtualizing macos under macos:

https://developer.apple.com/documentation/virtualization/virtualize_macos_on_a_mac

Can i use vmcli for this purpose too and when it is possible how can i do this?

vmcli and vmctl are awesome tools by the way 👍 .

I am unable to find a way to cleanly stop the VM without it corrupting the ext4 filesystem in disk.img. I am using the default ubuntu setup as built with the makefile. I have tried:

• In the VM guest: sudo shutdown -h now
• In the VM guest: sudo init 0
• From macOS host: vmctl stop ubuntu

All of these leave me with a disk.img that is corrupted. Eventually the corruption builds to the point that the system doesn't start properly. I have the Paragon extfs drivers for macOS, and I can perform the following in macOS:

hdiutil attach -nomount ubuntu/disk.img

After that if I verify the filesystem using the paragon extfs driver it says it needs to be repaired. I can repair it using the paragon extfs driver.

This happens after every shutdown, and from what I understand what I'm doing should exit the VM cleanly.

Can anyone shed light on this or recommend an approach that will prevent the VM from corrupting the ext4 filesystem?

Thanks!

The provided "vmbuilders/ubuntu.sh" is a good start but it didn't work for me to configure the ubuntu cloud image: The openssh-server failed to initialize, my user was not created, no network connection. Instead of using the initramfs we can just provide an iso image with the necessary 'user-data' and 'meta-data' files to initialize the image at the first boot.

# create cloudinit config iso
rm -rf iso_folder seed.iso
mkdir iso_folder

cat << EOF > iso_folder/user-data
#cloud-config
users:
  - default
  - name: $USER
    lock_passwd: False
    gecos: $USER
    groups: [adm, audio, cdrom, dialout, dip, floppy, lxd, netdev, plugdev, sudo, video]
    sudo: ["ALL=(ALL) NOPASSWD:ALL"]
    shell: /bin/bash
    ssh-authorized-keys:
      - $(cat ~/.ssh/id_rsa.pub | head -n 1)
bootcmd:
  - apt remove -y irqbalance
network:
    version: 2
    renderer: networkd
    ethernets:
        enp0s1:
            dhcp4: true
EOF

Plenty of examples here. Then to create the iso file:

touch iso_folder/meta-data

hdiutil makehybrid -iso -joliet -iso-volume-name cidata -joliet-volume-name cidata -o seed.iso iso_folder

Then add the seed.iso file to the vm.conf

cat << EOF > vm.conf
kernel=vmlinux
initrd=initrd
cmdline=console=hvc0 irqfixup root=/dev/vda
cpu-count=1
memory-size=1024
disk=disk.img
cdrom=seed.iso
network=nat
EOF

There's no need to launch the initramfs anymore to edit the image. The openssh initialization, user creation and networking do work for me now. Full example: https://gist.github.com/alexsarmiento/c52874738c3bdb96424c57b297b2678d