I think I remember having the exact same problem on an ASCII system and I've been able to solve it.
Nvidia almost always gives you at least SOME trouble when you install the drivers on linux.

Try stopping this service, purging it, reinstalling

service stop nvidia-persistenced
apt purge nvidia-persistenced
[maybe reboot here, if you do it after uninstalling service won't be running when you try to install again]
apt install nvidia-persistenced

I've just tried this fix and it didn't work when I booted from real dvd. Caps lock works, but not alt-f1/f2. The console font didnt change.
After that I rebooted from the same image located on hdd, and it worked. The console font changed.
It SEEMS that the probability of encountering this error is much higher when booting from real physical dvd.

I am using the image that fsmithred posted in this thread named
devuan_beowulf_3.0.0_beta_2020-03-25_amd64_desktop-live.iso

In my opinion, if you're using ZFS (or BTRFS), there's one more reason to separate root and home partitions (besides the ones already mentioned):
If you make a snapshot of your root filesystem, and later want to rollback (maybe after you've broken something), they should be separate -
or else you will rollback your personal data (home) together with the system files on root file system. They should be decoupled.

I can't tell much about EFI, buy you almost never NEED, an extra /boot, but in my opinion it's always a good idea to have one - for example, grub doesn't support all filesystems that linux proper does, different versions of grub can or cannot read different filesystems, also you may boot a livecd for a rescue and have trouble unlocking your luks or importing zfs pool, so /boot should be a separate (smallish) partition, formatted with an fs that everyone can read. I recommend ext3 for /boot.

In early days of windows NT/XP, it was common practice to have a separate C: drive with a simple filesystem (FAT/FAT32), because not everyone could read NTFS back than.

The same logic applies today - it's generally more reliable if you have a separate /boot formatted with an fs everyone can read. But of course this depends on your particular needs and use case.

Nothing is "necessarily" snapshotted with ZFS/BTRFS, you snapshot what you want.
I always snapshot the root filesystem after the first successful boot, and than before large upgrades or before making potentially breaking changes.
Since my /home is separate, I don't have to worry about my data/work being rolled back, and just operate on root.

[EDIT:] also keep in mind that if you do have extra /boot on a simple filesystem such as ext3, it's contents will NOT get snapshotted. You'll have to be careful with your boot files, or back the contents manually.

Debian based systems, including Devuan, do a good job of keeping an older, working copy of both your kernel and your initrd - thus there is seldom a need for backing up the full contents of /boot.
[End-EDIT]

If you're talking about a production server, there may be other specific cases where you must do snapshots.

Good rule of a thumb is to make at least one swap partition, sized for example half of your ram. That will give you some breathing space.

Yes, I know that. I installed wicd just in case I wouldn't be able (for some reason) to write the correct interfaces file, but switching to dhcp worked.

If it works for you then it means that's MY problem, not a Beowulf problem.

Another potentially interesting topic:
When I installed the system initially w/ debootstrap, I chrooted into it and copied the configuration for my eth<X> cards directly from my working jessie /etc/network/interfaces file. I did that through copy/paste, so a spelling error is out the question.
When I rebooted, I got error message, something about "missing attribute: address for card eth<X>", don't remember exactly, but I have checked the "address" attribute, and it was set to the static IP I wanted to use.
I then rebooted into stable jessie, chrooted into new beowulf sys, installed wicd with all kinds of it's clients (cli, gtk, etc.), and changed the content of /etc/network/interfaces file to use dhcp. After rebooting back into the beowulf install, the card configured itself properly through DHCP. I think without wicd, just by starting the networking service
I don't know if any of the packages which wicd brought in fixed the issue, OR whether there's a problem with parsing manually configured eth<X> cards.

If maybe someone tried a manual config (with static router, ip, gateway, etc. specified in /etc/network/interfaces) for eth<X> card, that could clarify if it was my misconfiguration, or a problem with one of the packages being unable to parse a perfectly valid config file.

WRT mate desktop:
I installed mate after the first boot into a working, cli only system, with network already configured as described above, by running tasksel in console, and selecting "devuan desktop env" and "mate" entries. It worked flawlessly, other than apt/dpkg on btrfs being SLOW AS MOLASSES.

EDIT: typo

Just tried a few more times:
Sometimes it boots immediately, and very quickly, without this message. In those cases, system seems to keep working for as long as I wish.

And sometimes the message gets displayed. For the cases when it does, waiting a couple of minutes (2 full circles) helps - the system boots. However, for the one case it did boot after this significant timeout, the system FROZE a couple of minutes into the live session. No caps-lock working, no soft reboot, just power off and reset helped.

One more thing I've noticed: when boot goes ok, the console font is changed from the bios font to the standard debian thin letters font.
When I get these timeout messages, the font on console is unchanged from the standard bios-vga font.

The font change happens(when it does) a couple of steps before the messages start appearing.
When the messages appear, it is always in the "Starting lvm volume groups" phase.

edit: typo

Thanks for your reply, fsmithred.

I've downloaded beowulf's .deb file for debootstrap already. The only dependency seems to be wget.
But I don't know if I will install it. The reason I come so rarely to this forum with problem-posting is because I like to keep
strictly to the official release, not mixing any packages no matter how simple they seem to be.

There seems to be the option to use  DEBOOTSTRAP_DIR env variable when running debootstrap, so it should be possible
to simply unpack the deb file into a separate location and use the aforementioned variable to launch the beowulf's version. I may try it out later.

That has always been my understanding as well, that's why I tried to use the livecd.

However unfortunately I could not boot it, as described in an earlier post, despite being able to boot jessie, ascii, and debian10 live cd's with the same method.

Since I have succeeded booting a netinst beowulf image, maybe I'll use that if I decide to go on.

But it sure would be nice to be able to use a live iso.

Could not boot the desktop-live amd64 iso from hdd using grub.

Stuck on
"activating lvm volume groups... <timeout>"
than something like
"device /dev/sd<x> not available in udev database even after waiting for 1000000 microseconds.<another timeout>"

This second line is repeated for every hdd and partition sda,sda1,... sdb,sdb1,sdb2. ... loop0,loop1,
than goes back to the first device again and probably an infinite circle.

The grub cfg is as follows:

set root="hd<X>,<Y>"
set isofile="/devuan_beowulf_3.0.0_beta_amd64_desktop-live.iso"
loopback loop ($root)$isofile
linux (loop)/live/vmlinuz boot=live fromiso=/dev/sd<X+1><Y>$isofile username=devuan
initrd (loop)/live/initrd.img

x is hdd number(from 0 onward), y is partition number (from 1 onward).

I've used this method to boot jessie, ascii, and debian10 live isos succesfully many times in the past.

EDIT:
Corrected the error message - it's microseconds, not seconds.
Also burned the iso to a dvd, booted from there - same result.
Maybe if I wait for hours something will change, but I doubt it. Even ctrl-alt-del doesn't work.
Debian 10 live iso boots fine on this machine.

EDIT2:
The sha256 sum is correct.

I also tried the beowulf netinst image, and it booted correctly from hdd using grub.

It's alive!

Installing packages for ubuntu 18.04 did it for me.

They have all necessary patches applied to make it work with newer kernels.
No dependencies problems, libraries from ascii are of high enough versions, and it doesn't pull in any unnecessary stuff with it

Just download manually and install.deb files for these packages
oss4-base oss4-dkms oss4-gtk

from bionic beaver repos.

Sysv init scripts are included.