Cloudmin Install Succeeds, but failure at boot drops to initramfs shell

SYSTEM INFORMATION
OS type and version Ubuntu 24.04
Webmin version Latest (?)
Cloudmin version Latest Xen Install (?)

Please note the (?) above regarding the versions are because I assume the Cloudmin install script (Xen version was used) pulls the latest, but I cannot verify as the system is now unbootable.

I ran the Cloudmin Xen install script on a fresh Ubuntu 24 installation on a Vultr VPS, and had no issues or errors show up during the install. Once the installation completed, I rebooted as per Cloudmin’s note about the Xen installs requiring a restart to enable the new Xen related kernel that is installed with Cloudmin via that script. Upon reboot, the system fails to boot, and the only way I can see anything pre OpenSSH server start is via Vultr’s “view console” option, which I believe is NoVNC based. Because of the console viewer not having the capability to copy and/or paste console text, I am unable to provide anything other than screenshots or screen recordings and to type out what I believe are pertinent bits from the screen’s output. Please keep that in mind as I attempt to share what I saw and was able to remember (memory issues due to TBI) and write down…

I performed some other tasks while the server rebooted and was unable to scroll back in NoVNC. but the error presented when I noticed the server hadn’t finished restarting was:

Scanning For Btrfs filesystems
done.
Begin Waiting for root filesystem... Begin: Running /scripts/local-block ... 
mdadm: No arrays found in config file or automatically
done.
mdadm: No arrays found in config file or automatically [editor note: repeats several times]
done.
Gave up waiting for root filesystem device. Common problems:
 - Boot args (cat /proc/cmdline )
   - Check rootdelay= (did the system wait long enough?)
 - Missing modules (cat /proc/modules; ls /dev)
ALERT! UUID=[TOOLONGTOOREMEMBERLOL] does not exist. Dropping to a shell!

and then it drops to busybox, with what looks like a initramfs context? been a while since I used busybox for anything. The prompt is

(initramfs)

Which reminds me of shell contexts or virtual environments…I digress…

So, as you might guess, I’m utterly confused here. This was a fresh install, that booted with no issues the one time I rebooted to finish the package update and upgrade that is standard before things like this Cloudmin install script.

Once it rebooted from my pre-cloudmin package upgrades, I pulled the script for cloudmin-xen via wget, and ran it by calling the script with the sh command, answered any queries that showed up (basically just the “are you sure” and “this is a supported OS right?” type questions. No errors popped up, and the script ran it’s course, and upon being notified that it was complete and what URL I could reach the interface at, I was told as expected that I needed to reboot in order to be able to manage the system properly with the Xen kernel extensions or something along that line.

Again, no errors, all seemed just like expected per the Cloudmin information I had read before starting this all up.

My questions are:

  1. Has this issue come up with the Cloudmin Xen installer before, and maybe has been archived off the forums? (I performed a search but didn’t come up with anything on the forums, and searching the internet at large didn’t return anything that was close enough to this situation for me to consider)
  2. Even if this issue has not reared it’s head before, does anyone have any ideas on where I go from here to resolve the issue? I will say, I’m posting this and troubleshooting at the same time so as to be sure to get this report out to the forum, and if I resolve it on my own or via answers/suggestions here in conjunction or alone, I plan on posting the discovered solution in a general writeup on my site that I will also put an abstract of here in order to help future users (always been a peeve of mine, people who don’t come back after getting a solution, since by not stating they found the solution and relating it to the forum they kill a lot of the value in searching for answers online).

Any help that can be offered, I would appreciate greatly, and I offer my thanks in advance (don’t worry, they’ll be offered again too. Gratitude is baseline necessary in my eyes).

Cheers, and I wish you all a good evening/night/day whatever the case may be when you see this post.

Cheers,
Sherwood McGowan
massive:Ronin

Not sure you can use a standard VPS, it may need to be bare metal.

While I appreciate the reply, I really must ask, is there any information that might help me or others understand why you think it MAY need to be bare metal? I’m not trying to pick a fight, I’m genuinely trying to work out just how you come to your semi-conclusion (you’ve got to admit, your reply is not exactly exuded confidence in what you say, and without citations to allow the reader to see the information you read and felt confident enough in that you linked it for myself and others to read, and within reason, to consider to be some level of information pertinent to the question asked…

I’m not getting strange errors other than the UUID based root filesystem error, and that is the only thing causing it to not boot (based on errors shared, and considering those are the only errors thus far). Bare metal vs compute instance (not a VPS setup, as they are typically a chroot/jail subset of a server’s OS, vs a Vultr instance being an entire OS plus installed applications that have some restrictions placed on them but are leaps and bounds above the VPS type offering of webhosts inthe past.

From what I understand so far, Cloudmin is primarily a Docker or LXC container orchestration engine (I know this isn’t a complete explanation, simplifying to save keystrokes and time consumed) essentially, and if I am able to run a server I’ve spun up an instance of straight off the Vultr “Marketplace” that is a Debian/Ubuntu based Linux OS that pre-installs Docker, and presents the image the new instance is based off of as a starting point for mid-to-large Docker implementations, wouldn’t it make sense that a simple Ubuntu 24.04 Server instance with the Xen-Kernel-based-Cloudmin install script downloaded and ran without issue should result in a bootable system, should it not?

I want to add that Vultr is not an everyday webhosting VPS setup. They are more akin to an AWS/GCS type of setup, a large worldwide cloud compute/vGPU/nGPU provider that spins up instances that are capable of Docker and/or Kubernetes implementations of their own. As such, those are the reasons why I personally felt that it was possible to install Cloudmin on a Vultr compute instance.

I hope that a forum member, such as anyone on the @staff group, will see this and reply before I deallocate a mini-pc from my Lab’s current setup so I can then wipe and install with Ubuntu 24.04 and then run the Xen-Cloudmin install script to see what issues it may have (likely to be the same considering I am not encountering anything that could be considered to be signifying a need for a bare metal setup.

I’m going to wait unti I can get a bit more declarative of a response, but I do appreciate the reply. I got up this morning expecting a lot of views and far, far less replies, so far that’s what’s happened, but I will continue on my side of things,and hopefully I determine the error’s true source and the solution myself. If not, well, one can only hope that the next replier is someone who provide just slightly more information than thus far.

Thank you all, and good day/afternoon

The install script makes some assumptions that probably aren’t safe on a virtual machine where network cannot be managed by Cloudmin.

Cloudmin, the module (server-manager) can be used on a virtual machine, though you won’t get much use out of it in such a case.

It is not this at all.

It is for managing VMs. It happens to support Docker and LXC, but Docker is a terrible fit for the Cloudmin model, and we’re removing it (and moving it to Virtualmin). LXC is also being retired. Cloudmin going forward will be for managing KVM instances. Xen support will stick around for a while, but the installer will setup KVM only.

We’re trying to simplify Cloudmin because it’s trying to do too much and it ends up being a confusing user experience.

But, it isn’t and will never be a container orchestration tool.

1 Like

Joe,
Thanks for the response, that resolves my question since I was intending it for use with Docker and/or LXC. But, since you say Docker is being migrated into the Virtualmin product, that is my current paid product so I will just keep that going for now.

I have only a single follow up question. Do you have a rough timeline for Docker implementation into Virtualmin? I’m not looking for a definitive answer, just a rough idea so I have something to look forward to :slight_smile:
Thanks again

Soonish. Weeks, probably. The code is mostly done, the UI needs polishing and we don’t have docs.

@Jamie, I’m still looking forward to you setting up new cloudmin-gpl and cloudmin-pro repos with Cloudmin having only KVM-related code. Without this, we won’t make much progress with Cloudmin until we have a lightweight and simple version of Cloudmin with KVM only for Cloudmin 10, as we discussed earlier on Google Meet.

This will happen after the release of Virtualmin 8, which is coming soon.

Do you mean weeks for Virtualmin 7.40.0 and WP Workbench 2.0.0 release, right? I doubt you meant Docker support in Virtualmin, as that would take much more time.