Neil_Young
Contributor
Contributor

Unable to launch a VM after fresh install with latest technology preview on M1

Jump to solution

Since the update to the latest tech preview a couple of days ago I can open my older VMs only in recovery mode. Trying to install a 20.04.3 as a new VM works, but the thing is stuck at first boot as all the old ones are and recovery doesn't work anymore

The screen is attached.

 

What am I doing wrong?

 

26 Replies
rhaleblianbeta
Enthusiast
Enthusiast

Yep and yep.  And please note, in case you haven't seen the amendment to my separate post about the Ubuntu Server link above:  as of yesterday that Live CD already updates to an unusable kernel at install time, removing the ISO's kernel in the process.  So it's now a no-go.

0 Kudos
rhaleblianbeta
Enthusiast
Enthusiast

Regarding new installs, I tried something dastardly.

1. from
https://ubuntu.com/download/server/arm

download 20.04.4 LTS.

2. Make the new VM as usual, EXCEPT say that you want to change the VM settings.
The settings dialog will appear.  Disconnect the networking adapter.

3. Continue with the installation.  The `cloud-init` step will take a while then continue. Some installation screens will be missing.  The install itself will be deceptively quick.

4. Reboot when instructed.  The boot sequence will get stuck consulting the network.  Go into the settings and turn the networking adapter back on.

5. The startup will continue and you should land on the login prompt.

6. You can do still do `apt update` and `apt upgrade` as usual, if that's how you roll, which will eventually install a newer kernel.  But now you'll have the original one too.  Use that original one when you next boot, via the boot loader menu.  That kernel is

```
$ uname -a
Linux confusion 5.4.0-100-generic #113-Ubuntu SMP Thu Feb 3 18:44:51 UTC 2022 aarch64 aarch64 aarch64 GNU/Linux
```

I have only been running the VM for 10 minutes now so I could be crying wolf again...

0 Kudos
Technogeezer
Virtuoso
Virtuoso

That sounds like a reasonable way to get around the issue. Good job!

I think we’re all going to have to get a bit creative until this issue gets resolved. 

0 Kudos
rhaleblianbeta
Enthusiast
Enthusiast

I'll put additional notes here:

https://gist.github.com/rhaleblian/44f2cb4418ee978603dc0b15dadf0e02

There's a note on there about how to install a desktop environment in the case that you have Server, since I'm sure most people ultimately wanted to install Desktop.

If this works for anybody, please report.  I'll update that other "NOW BROKEN!" thread.

0 Kudos
fabienmagagnosc
Enthusiast
Enthusiast

We are clearly now reaching a wall, I'm using autoprotect/snapshot to avoid the worth situation with my kernel 5.10 to be "autoremoved" to unstall another non working one (5.17 ... I'm looking at you !)

 

Anyway, until VMware team come back to us, we are stuck now for almost any distribution.

note : people speaking about staiblity with Fedora, Centos Stream (now) or ubuntu, clearly doesn't understand the minimum vital to survide in this "OS jungle" we are now.

0 Kudos
rhaleblianbeta
Enthusiast
Enthusiast

Added notes to the GitHub gist on how to install a desktop environment after installing Ubuntu server.

No tricks here, just a well-cited approach.

After doing that, things feel a lot like Desktop would have.

0 Kudos
Forge128
Contributor
Contributor

If you'd like a longer-term band aid, you could try

sudo apt-mark hold linux-image-5.4.0-100-generic linux-headers-5.4.0-100-generic

 

That will cause those packages to remain installed and available, so you can apt update/upgrade to your heart's content, with at worst a hung boot and the need to select the alternate kernel version in grub. Do make sure to remove the pin when the package is eventually no longer needed, else it'll hang around forever.