Building ProxMox on Devuan

rhY

Active Member
Dec 6, 2019
10
2
43
48
Your Mom's House
openairships.com
Hey all! Long time ProxMox fan, have used it to great success in many scenarios.

I'm very interested in migrating all of my backend servers and services to Devuan as it is far more stable and easy to manage than Debian. I know I'm going off the reservation here, but is it possible to build ProxMox (even an older version) on Devuan? Would I just follow the Wheezy instructions? Which version would I be eventually be installing? Are there still security updates to the older versions?

Thanks in advance,
Rhy
 
You have not done much side by side comparison between Devuan and Debian, clearly.

The stability is so much greater in Devuan it's not even funny. Not to mention efficiency, easier to manage code, faster boot times, cron support, and several other industry standard solutions that are totally broken from Debian 8 forward. I strongly suggest you run some tests of your own between Devuan and Debian and see for yourself. You can't argue with the 5-10% CPU gains, 15-20% memory gains, and several seconds of boot time speed that Devuan has over Debian at this point.

Debian has gotten more and more bloated, as you'd expect using giant, poorly-understood monolithic corporate garbage such as systemD. Run the tests. Load a giant mail server on Debian, and then Devuan, and run some scripts that abuse the heck out of it. Debian will die way sooner than Devuan.

On most of the servers I manage i was staying with Debian 7 as long as possible, and only much later started migrating them to Devuan. So far I've had near 100% success rate, which I certainly did not on any of the services and containers I was trying on Debian 8. I haven't even bothered with Debian 9 or 10, because the Debian 8 experience was so pathetic. Maybe it's gotten better?

At any rate, you should definitely look at Devuan very closely. Your statements above were not only false, but almost exactly the opposite of reality. I would love to see the evidence for what you said, if you have any.
 
At any rate, you should definitely look at Devuan very closely. Your statements above were not only false, but almost exactly the opposite of reality. I would love to see the evidence for what you said, if you have any.

Seems you really believe that. But as said above, we need many features from systemd, so Devuan is no option.
 
You have not done much side by side comparison between Devuan and Debian, clearly.

You talked about building an old Proxmox Wheezy version on Devuan - this is old software with known bugs and you will put users to risk. This is a really bad idea. Current Proxmox version heavily depends on systemd, so you cannot build this with Devuan.

I do not want a basic discussion about Devuan here, this is the Proxmox VE support forum only.
 
OK, well I'll give it a go and see how ProxMox 3 runs on it. Thanks for the civil conversation. It's appreciated. Is there a list of known vulnerabilities for ProxMox 3 so I can patch them, or fix them myself?
 
Hi, I wonder if there is any progress around this? (Last post for 4.5 years ago.)

I'm also interested in running ProxMox on a Devuan server. (No chance for moving back to Debian.)

Up to now I have been running VirtualBox, but I am considering migrating to ProxMox because of stability issues.

So, is ProxMox still closely tied to systemd services?
 
Hi, I wonder if there is any progress around this? (Last post for 4.5 years ago.)

Well, they made their preferences clear back then. Systemd got only more mainstream since.

I'm also interested in running ProxMox on a Devuan server. (No chance for moving back to Debian.)

Up to now I have been running VirtualBox, but I am considering migrating to ProxMox because of stability issues.

If you are looking to replace VB, why not start with virt-manager / libvirt instead?

So, is ProxMox still closely tied to systemd services?

If you do not like systemd, you will hate PVE in terms of monolithic complexity. :)

(NB I do run some systems without systemd, but they are special use cases - so I am not in either camp.)
 
Current Proxmox version heavily depends on systems

What systemd only features does it actually use, out of curiosity? This was someone asking long about about building it on Wheezy, but I do not see any features that a regular init cannot handle if done right (it's usually easier going backwards). Note I am not asking you to do it, but is it just "too much work and we/Debian had picked a horse already" or do I miss something that only systemd can provide?
 
What systemd only features does it actually use, out of curiosity? This was someone asking long about about building it on Wheezy, but I do not see any features that a regular init cannot handle if done right (it's usually easier going backwards). Note I am not asking you to do it, but is it just "too much work and we/Debian had picked a horse already" or do I miss something that only systemd can provide?
Yes I'm also curious about this. At some point I want to fork ProxMox and do something really radical with it:

https://supermox.weebly.com/
 
I just suspect a publicity stunt.

But I wonder why a hypervisor needs anything "tightly coupled" with any sort of init.

That said, when it already jumped over to the ESP travesty for EFI install, why it is not going UKIs and making an exception for SecureBoot.
 
I just suspect a publicity stunt.

But I wonder why a hypervisor needs anything "tightly coupled" with any sort of init.

That said, when it already jumped over to the ESP travesty for EFI install, why it is not going UKIs and making an exception for SecureBoot.
Definitely not a publicity stunt. Every IT professional I work with is conservative, hates Lennard Poetering, and most of them, if they use debian at all hate systemd. I'm also curious what exactly is required from systemd for proxmox to work. You would think it would be fairly agnostic regarding init systems. I've gotten almost everything I've tried to work fine in devuan. Although for desktop purposes, getting nvidia drivers working can be a bit arduous. But that's typical of almost every distro, frankly. nvidia sucks.
 
Definitely not a publicity stunt. Every IT professional I work with is conservative, hates Lennard Poetering, and most of them, if they use debian at all hate systemd. I

Well we seem to know different professionals, most I know doesn't even have an opinion on systemd. Those I know who use Debian ( myself included on my Private Hardware) don't hate systemd, my guess is thst they would use Devuan if they do. The one thing I dislike is using Ubuntu as server but this has nothing to do with systemd but Canonicals way of doing things.

I also know nobody who "hates" Poettering, my own opinion: He is a gifted engineer with strong opinions ( not all I agree with, hard-coding Google-DNS so "Users always have working DNS" is just stupid, allthough I get where he comes from, he design systems for endusers not for people like us ) who explains them well enough ( see https://0pointer.net/blog/ ) that it's easy to develop your own opinion on it. This doesn't mean that I agree with them, imho it's pointless to design Linux as Desktop System ;)
My own opinion is: systemd has it's downsides but I prefer them to the downsides of classical init Systems.
At work all our systems ( sadly Ubuntu not Debian) have systemd and I wouldn't want it any other way. The sandboxing features of systemd alone it are worth it: https://wiki.archlinux.org/title/Systemd/Sandboxing
And declarative units are way easier to write and debugging than init bash scripts.
I suspect most people who think that systemd-based systems are "more complex" and "harder to manage" are just used to the old way of doing things, of course it's easier if you done it that way for fourty years ;) But that's not a technical Argument.

Now this doesn't mean that systemd-less Systems doesn't have their usecases. Alpine is a great lightweight base for docker or System containers, I would never use Ubuntu or Debian as Base for a docker Image. But migrating Proxmox products to Devuan due to the unwillingless of old horses to learn new tricks would be a massive waste of ressources.

One thing I agree with you that your post wasn't for publicity, otherwise you wouldn't have waited over a year with the answer
 
Last edited: