Trying for the first time to do a VM on TrueNAS SCALE community edition (currently running 25.10.3.1 - Goldeye). (I used jails some in CORE, was hoping to bring some things I did that way back to life.)
The only way I can find to do anything is to connect via VNC. Doing that, I can interact with the install disk and watch much of the install go through – but at the end of that, it leaves me with a nearly-blank screen (still viewed via VNC) that says Ubuntu 24.04.4 LTS ubuntu-server tty1at the top. (Nothing I type in through VNC echos or causes any action; disconnecting VNC and reconnecting does show the same screen again, though.)
Note that this is an Ubuntu server install – no desktop.
Then I thought about a serial terminal connect, and noted the “serial shell” option in the VM. If I click that, what I get is
WARNING: Your user does not have sudo privileges so /usr/bin/virsh command will run
on your behalf. This might cause permission issues.
error: failed to connect to the hypervisor
error: authentication unavailable: no polkit agent available to authenticate action 'org.libvirt.unix.manage'
Repeating over and over again with a pause. (Not scrolling down the screen, blanking and starting at the top again.)
So,…what’s the deal? Seems like the VNC thing should be getting me in touch with the console on the VM, and it is for a while, I can see the install steps flowing by, but then it freezes up. And “serial shell” really sounds like it should get my a tty connection to the VM, but it fails with errors I don’t understand. I don’t think I did anything unusual – but who knows, this whole VM procedure in the TrueNAS GUI is new to me. Help?
Try restarting the Ubuntu VM, and then launching the Serial Shell immediately afterwards. When you get to grub, edit the kernel line for Ubuntu server to include a nomodeset option before –.
I do get to grub. Looks like the install never completed. So messing around with that, and will try your nomodeset option suggestion. Probably after dinner.
It might be changing the serial speed. Try console=tty0 console=ttyS0,115200. This should give you console on both the VNC window and the Serial Shell.
Sure, for several years and currently running Ubuntu server 24.04 with several apps on Truenas CE 25.10.3 but work from the VM’s CLI after logging in through SSH.
Maybe the hang is caused by something entered in the VM’s setup screens.
Well, that’s hopeful, anyway. Didn’t seem likely the whole VM scheme didn’t work, but having multiple people actually confirm they’ve got it working is encouraging.
There are enough selections in VM creation that I don’t fully understand that it’s completely possible I’ve got something wrong. (I’ve run vmware and virtualbox VMs quite a lot in various contexts, and even when it was completely new to me I’ve never had anything like this amount of trouble getting one set up.)
Do people run the install through the serial shell, or through VNC mostly?
Let’s see. My config params are:
Boot method: UEFI
Hyper-V: off (that’s for windows, according to the docs)
TPM: off (needed for windows, but this is for an Ubuntu guest)
1 cpu, 2 cores, 1 thread
CPU mode: host model (also tried passthrough for some).
Custom CPU mode I was avoiding since it seems to ask for something much harder than presenting the actual host CPU. And also because I tried to use it once and discovered that the first 3 CPU models I tried from the list couldn’t be supported on my actual CPU (which also confirmed my feeling that it was a complication).
Not clear on “Host MOdel” vs. “Host passthrough”, either. The docs read to me like “host model” is the simplest case, though?
Not clear from the docs what “display device” even is; but since what I’m looking for is a headless server install, basically?
Okay, I’ll try a few more different things, see if they go anywhere.
Hey, something here has caused a breakthrough – the installer actually asked me what language and keyboard I’m using! Never got that before.
Will take a while to run this through, and maybe I’ll be able to figure out exactly what’s different that matters.
(This run is host passthrough, 2 threads per CPU. Oh, and I removed the minimum memory setting. The threads may be the only thing I never tried before, but the combination isn’t one I’ve used I don’t think.)
Anyway, breakthrough; dunno if this will lead directly to a win yet, but further along than ever before. Thank you for that!
After confirming it can restart successfully, I changed cpu mode from host passthrough to host model. That still boots. (Of course, the install might possibly not work in a configuration that will, after installation, boot?)
I’ll probably poke at other things, see what will and won’t run.
One immediate question though; where is the disk storage for this virtual? I picked a pool, I remember, but nothing has appeared any of the places I think of.
Huh; never saw /dev/zvol before, but it exists. So the path given there exists. It traces back through symlinks to a block special device /dev/zd0. So there isn’t a filesystem containing that data accessible from the host? (Used jails for years, got kind of used to the way things worked there.) Can I get it to do something else?
