Your wish list for Proxmox

[Domino]

Been with Proxmox for years, been great for lab non-commercial use as a platform to simplify vm/container management with mouse clicks.

But I've over the years come to realise that it seems to lack in a few areas, and I know that making any feature requests tends to just dissapear into the void because the devs are more commercial use orientated than not and who can blame them, everyone needs to pay their bills... however I'd still like to put out some very important suggestions:

1) The most important I believe is that the out-dated web-ui needs to go 21st century mode and go all fancy 'material' like as many web-apps are these days. Just still looks like its stuck in the 90s. I know this is 'beauty' before function, but beauty draws in the larger crowd these days. A complete overhaul on appearance (re-write using a new framework, Rust based for example?) is definitely needed.
2) Move away from Perl and go completely Python - the scripting language of the 21st century - although this would annoy countless existing users with prior investment in perl scripts - so this could could be considered possibly in a way where Perl is still supported but everything Proxmox codebase wise goes full on Python.
3) Add a full hardware offloads networking UI layer to control and manage SRIOV VFs. So many advanced users are being forced to manage tonnes of scripts and even writing services to manage this alone, can become cumbersome once you've got a few hundred VFs via multiple smart NICs.
4) Add full RDMA support via the UI interface where when available in hardware it can be utilised with a few clicks, for example with storage pools, shares etc.
5) Add a CPU core pinning UI facility with configuration via UI for hugepages too.
6) Add a much simpler plugins API feature using Python for example so that there can be greater opensource crowd draw-in for potentially unlimited features without core dev staff of Proxmox having to dedicate time on features which they do not deem commercially rewarding (for example CPU pinning in No.5). There is a great resource of open-source python devs on github who can easily jump in and grow Proxmox's feature list exponentially if this one single avenue was catered for by the core-devs in the baseline framework.
7) Look into bringing Incus in as beta sooner rather than later, Proxmox does not want to be left behind, Incus is already 'stealing' potential Proxmox customers even with a non-official UI!

I'd be interested in knowing what other users dream of having in Proxmox, and any comments regarding my wishlist are encouraged.

I would like to finish this post by thanking the Proxmox devs for all their great work thus far, but it indeed needs to be said that the future is going to be very tough, and the much needed groundwork needs to be laid immediately if we want to see Proxmox grow to levels unimaginable rather than fade into obscurity.

 

[Max Carrara]

1) The most important I believe is that the out-dated web-ui needs to go 21st century mode and go all fancy 'material' like as many web-apps are these days. Just still looks like its stuck in the 90s. I know this is 'beauty' before function, but beauty draws in the larger crowd these days. A complete overhaul on appearance (re-write using a new framework, Rust based for example?) is definitely needed.

We're on it chief — as mentioned in our Datacenter Manager Alpha Release post, we're working on a Rust-based front-end that's built on top of yew. You can find the code on git.proxmox.com in the User Interface & Related Tooling section.

2) Move away from Perl and go completely Python - the scripting language of the 21st century - although this would annoy countless existing users with prior investment in perl scripts - so this could could be considered possibly in a way where Perl is still supported but everything Proxmox codebase wise goes full on Python.

We're making steady efforts do actually rewrite a lot of our Perl code in Rust, not Python. Also, if you're familiar with PyO3, we've got our own Perl equivalent of that: perlmod. That library allows us to call Rust code directly from Perl.

6) Add a much simpler plugins API feature using Python for example so that there can be greater opensource crowd draw-in for potentially unlimited features without core dev staff of Proxmox having to dedicate time on features which they do not deem commercially rewarding (for example CPU pinning in No.5). There is a great resource of open-source python devs on github who can easily jump in and grow Proxmox's feature list exponentially if this one single avenue was catered for by the core-devs in the baseline framework.

There's a lot of work being done in this direction as well; I'm currently in the process of brushing up our storage plugin API, so that it's easier for third-party developers to write their own plugin for their storage solution. If you've got any more concrete suggestions, do let me know!

7) Look into bringing Incus in as beta sooner rather than later, Proxmox does not want to be left behind, Incus is already 'stealing' potential Proxmox customers even with a non-official UI!

Just gave it a little look; seems to look quite snazzy, but it doesn't seem that feature rich yet at all (no hate here; I'm just saying). In what way is this project "stealing" customers?

For the remaining points, I can't remember if those things are on our bug tracker or in progress already; feel free to open feature requests on our Bugzilla if you're really eager to see these things get integrated into PVE. Chances are much higher there that somebody will pick it up and start working on it if it's on our tracker. Some of the things you mentioned might already be on it, FYI.

I would like to finish this post by thanking the Proxmox devs for all their great work thus far [...]

Thank you! We try our best.

[...] but it indeed needs to be said that the future is going to be very tough, and the much needed groundwork needs to be laid immediately if we want to see Proxmox grow to levels unimaginable rather than fade into obscurity.

Don't worry, we're getting there.

 

[fabian]

7) Look into bringing Incus in as beta sooner rather than later, Proxmox does not want to be left behind, Incus is already 'stealing' potential Proxmox customers even with a non-official UI!

this one in particular doesn't really make sense - incus is the same kind of software as PVE (a management tool for containers and Qemu) but with a lot less features..

 

[Kingneutron]

> 1) The most important I believe is that the out-dated web-ui needs to go 21st century mode and go all fancy 'material' like as many web-apps are these days. Just still looks like its stuck in the 90s. I know this is 'beauty' before function, but beauty draws in the larger crowd these days. A complete overhaul on appearance (re-write using a new framework, Rust based for example?) is definitely needed.

The interface is functional. We don't need a "fancy webapp" for a hypervisor, did you learn nothing from windows 8?

> 2) Move away from Perl and go completely Python - the scripting language of the 21st century - although this would annoy countless existing users with prior investment in perl scripts - so this could could be considered possibly in a way where Perl is still supported but everything Proxmox codebase wise goes full on Python.

No. Speaking from an outside perspective, Python is a mess of breaking changes. Rust is probably more sustainable.

> 5) Add a CPU core pinning UI facility with configuration via UI for hugepages too

You can already do CPU Core pinning in the PVE GUI, as well as commandline scripting.
Hardware / Processors / CPU Affinity

https://github.com/kneutron/ansitest/blob/master/proxmox/proxmox-vm-assign-cpu-cores.sh

 

[bobmarley2021]

@Domino coincidentally, I opened a FR on Proxmox Bugzilla a couple of weeks back that sort of aligns with your first point here, and your post was referenced.

Looking at the Proxmox Data Center Manager thread here, you can immediately see some UI improvements and it's great to hear both @fabian and @Max Carrara confirm that improvements to this are well underway.

@Kingneutron

> The interface is functional. We don't need a "fancy webapp" for a hypervisor, did you learn nothing from windows 8?

It's not entirely though is it? For example the mobile UI for PVE has no functionality except to provide information. You cannot perform any actions and interact with the nodes etc. If I use PBS now as an example (which does not have a mobile UI), or force desktop mode view for PVE, the UI fails to render properly on devices with smaller screens (Firefox for iOS for example - you cannot reach the right hand side of the page). I have also noticed several other UI issues that make it unpleasant to use. On a large iPad pro there are lots of issues with the keyboard showing/hiding, it's incredibly annoying to use the noVNC interface, for example. I agree that function shouldn't be trumped by pure eye candy (or a fancy webapp as both you and Domino put it) - but [in my case] it's not eye candy that's being requested . It's a modernised interface that is compatible with a wider array of devices/browsers. This is nothing like Windows 8/Vista/whatever.. I don't want glowing / shiny UI's elements, just one that works with other browsers... it's a totally different scenario. IT teams and homelab users alike use an array of devices to manage their infrastructure nowadays - it's not always a desktop/laptop. It's a lot of times quicker and easier to carry around a tablet/pad/phone.

 

[Kingneutron]

As both a homelabber and a professional sysadmin, I've never seen a phone interface be particularly useful for anything much beyond quick emails, texting and reading the news. Tablet maybe to some extent, but nothing so far beats a PC, keyboard/mouse and browser (plus commandline / ssh.)

Sorry, but tablet/app interfaces are (probably) always going to be a distant back-burner priority. The real work for PVE is getting done on the primary web interface. I mean, it's open source and there's an API, feel free to put some work into it if you think it'll make a difference. But it's never been a priority - or even a real option for me.

 

[gfngfn256]

While I totally agree with the notion that functionality & operation should be the primary focus of any HV system, I do believe that sometimes "you have to keep up with the Joneses". My point being; any OS is only as good as its clients/customers who actually use it, so if Proxmox is not going to keep up in attracting the wider client/customer base, it is eventually going to lag behind in its general usage, and so yes, if some potential users out there do want a snazzy GUI etc., put it out there. Imagine a basic scenario of a new user giving a different OS a spin "to see what its like", & he finds the general GUI cumbersome & not to his liking, he is going to move on to some other OS that actually maybe more snazzy but potentially overall less functional.

As has already been pointed out, Proxmox due to recent shifts in the HV scene, has been given a great & golden opportunity to attract a much wider user-base to this awesome open-source product, so the team must make every effort to use this chance to its fullest.

Let me just finish my rant, by yet again commending the whole Proxmox team on an absolutely great product, support & forum.


BTW: I also would appreciate if the desktop-view on a Phone (Android) could actually access the GUI somewhat functionally (snazziness not required).

 

[Domino]

The 'windows 8' argument didnt turn Microsoft away, we're now at Windows 11 and it is evolving to the day (much to the hate of old-techies), the argument is not relevant... this is about the future, the gen-z and beyond...

Regarding Incus, it was not about features, it is about attraction to what doesn't look like it squeezed out of an engineers toolkit...

The aim of modern app design is 'pretty', 'simplicity', but VERY powerful 'under the hood'.... in essence, suck those businesses in that see it as an attractive 'easy to use' tool, something modern users can easily relate to... but the power then becomes apparent as the said modern user gets more involved as their needs 'grow'... we all remember when all we simply wanted was a small 'container' farm, but a year later our needs went huge... it is where you begin, that is where your roots began - crowd attraction is a social skill, Microsoft didnt' shoot themselves in the foot with Windows 8, they knew they needed to change to stay relevant, even a business that size with its monopoly of the desktop industry had to move with the times... business is business.

...never frighten the modern user away to other solutions that are far more behind on features but so much more 'welcoming'... beauty is the way of today... 'complexity' must never become the label, the standard, the marketting - because it is at that point that you begin to realise you kept your door open just a few inches when it should have been open the size of a grand stadium - and now you only have a bunch of bearded old farts hanging around your garden picnic, meanwhile next doors has the entire world partying away like its 1999 'que Prince'!

https://www.youtube.com/watch?v=rblt2EtFfC4

 

[proxuser77]

The 'windows 8' argument didnt turn Microsoft away, we're now at Windows 11 and it is evolving to the day (much to the hate of old-techies), the argument is not relevant...

Windows is one of the worst examples when it comes to UI and UX design. It may look nice on the surface, but as soon as you dig a little deeper you are confronted with all kinds of UI and UX inconsistencies, some of which date back to Windows 2000 or even NT4.0.

Regarding Incus, it was not about features, it is about attraction to what doesn't look like it squeezed out of an engineers toolkit...

An even worse example because Incus itself does not even have an official web interface.

this is about the future, the gen-z and beyond...

and now you only have a bunch of bearded old farts hanging around your garden party, meanwhile next doors has the entire world partying away like its 1998!

Then maybe GenZ and beyond should start taking their future into their own hands by getting involved and actually doing something instead of just complaining. Just like those bearded old farts did in 1998 when they were your age, because without their work, the things you complain about would not even exist.

They have earned the right to hang out at garden parties or wherever the heck they want. Have you?

 

[Domino]

I would like to add, that in my professional consultancy career, I have over the past decade on numerous instances attempted to offer Proxmox as a platform solution, not once have I been successful. The issues always revolved around presumptions driven by how it looked...

There is a natural inclination these days to 'looks overly complex', businesses are not technical, they have their products to sell, their services to provide, they run away from 'complexity'. When we look at cloud solutions, they really are on the face of it very simple to navigate, their UI design is modern, it is about putting the advertisement out there 'use me with very little headaches'... this really is a big driver in customer focused cloud solutions, Amazon and Microsoft with their huge cloud services know this all too well, but numerous smaller 'boys' are also now gaining traction. Indeed many of them lag well behind Proxmox in 'features', but the principle of the story here is that once something becomes labelled as 'too complex' purely based on appearance alone, then that something begins to find itself falling into obscurity without notable commercial growth, this can lead to the demise of said something if not acted upon sooner rather than later.

Proxmox needs to survive in growth, with financial rewards for the development team and company, not in obscurity with declining/stagnating monetary income.

Hopefully in the future I'd like to be able to offer this platform as a solution, and not have to deal with presumptions from non-technical people but indeed they are the gatekeepers of financial transactions so their presumptions matter greatly.

 

[Domino]

Bad example: Windows is one of the worst examples when it comes to consistent UI and UX design. It may look nice on the surface, but as soon as you dig a little deeper you are confronted with all kinds of UI and UX inconsistencies, some of which date back to Windows 2000 or even NT4.0.

An even worse example: Incus itself does not even have an official web interface.



Then maybe GenZ and beyond should start taking their future into their own hands by getting involved and actually doing something instead of just complaining. Just like those bearded old farts did in 1998 when they were your age, because without their work, the things you complain about would not even exist.

They have earned the right to hang out at garden parties or wherever they want. Have you?

Totally missing the 'point'. Ah well, thank you for your time. Have a good day sir.

 

[proxuser77]

There is a natural inclination these days to 'looks overly complex', businesses are not technical, they have their products to sell, their services to provide, they run away from 'complexity'.

Sorry sir, but I think *you* are missing the point here, because the non-technical people in a company will hopefully never get to see the Proxmox interface, and the IT administrators who have to work with it are hopefully at least somewhat "technical" ;-)

So I'd say you have to compare apples to apples, which in this case would be VMware or Xen Orchestra. Neither of those have a particularly "sexy" UI, nor are they designed for non-technical people.

And don't get me wrong, I'm not saying Proxmox is perfect or that there is no room for improvement, there is always room for improvement, but to demand that a management UI for a professional hypervisor should be aimed at non-technical users, I don't know. Should the receptionist be able to manage a company's IT infrastructure, or what is your goal here?

If it is indeed something along the lines the latter, then such a company would probably be better off with something like Synology rather than a full-blown hypervisor like Proxmox VE.

 

[Domino]

While I totally agree with the notion that functionality & operation should be the primary focus of any HV system, I do believe that sometimes "you have to keep up with the Joneses". My point being; any OS is only as good as its clients/customers who actually use it, so if Proxmox is not going to keep up in attracting the wider client/customer base, it is eventually going to lag behind in its general usage, and so yes, if some potential users out there do want a snazzy GUI etc., put it out there. Imagine a basic scenario of a new user giving a different OS a spin "to see what its like", & he finds the general GUI cumbersome & not to his liking, he is going to move on to some other OS that actually maybe more snazzy but potentially overall less functional.

As has already been pointed out, Proxmox due to recent shifts in the HV scene, has been given a great & golden opportunity to attract a much wider user-base to this awesome open-source product, so the team must make every effort to use this chance to its fullest.

Let me just finish my rant, by yet again commending the whole Proxmox team on an absolutely great product, support & forum.


BTW: I also would appreciate if the desktop-view on a Phone (Android) could actually access the GUI somewhat functionally (snazziness not required).

You sir, get my point 100%.

I love Proxmox, it is a core platform in my lab which I rely upon to keep my technical skills up-to-date, where else are you going to recompile the kernel on the fly to try out the latest and greatest technologies.... Hyper-V and VMware don't provide that level of customization, far from it. This has been my primary use-case, and it pains me to notice that it is being left out of conversations when it comes to containerization/VM management.

People used to say "beauty doesnt matter, it's what is on the inside that counts" - honestly I wouldn't choose my lover or even car or smartphone based purely on that addage alone, doubt most others would...

In today's increasingly superficial world of 10 second attention spans and severely trigger-happy judgemental societies right down to inferior-critical-thinking trends... it would be a self-inflicted disaster if one were to not adapt, reimagine itself.

Proxmox needs to take this very seriously, prophetic indeed.

 

[bobmarley2021]

As both a homelabber and a professional sysadmin, I've never seen a phone interface be particularly useful for anything much beyond quick emails, texting and reading the news. Tablet maybe to some extent, but nothing so far beats a PC, keyboard/mouse and browser (plus commandline / ssh.)

Sorry, but tablet/app interfaces are (probably) always going to be a distant back-burner priority. The real work for PVE is getting done on the primary web interface. I mean, it's open source and there's an API, feel free to put some work into it if you think it'll make a difference. But it's never been a priority - or even a real option for me.

I totally get and respect your point of view - but we'll have to agree to disagree with each other on this one. Yes I get that things like command line are easier on a proper keyboard. But I happen to know of a couple of people who work in the industry that make use of a tablet or phone for quick tasks in Web UI's for whatever application. But then that's the joy of choice. Yes it may not be quite as popular as sitting at a desktop or laptop and of course there will be some tasks that can be performed infinitely quicker that require a mouse/kb, but it doesn't mean mobile doesn't have a place. (With that said, I also acknowledge that mobile is not a forefront focus for this project). I mean, for example, do you use the command line every time you login to Proxmox? There's going to be a percentage of people who just want to view some stats, take a couple of snaps and logout - granted not every time - but that's what the GUI is for, else we'd only have CLI. I am only a home labber but I often login to various applications on my phone or pad just to tab one or two things, maybe type a few words in whatever text field then logout, I use the console maybe 10-15% of the time, mainly on initial vm/container setup. As my comment above said, I am not necessarily after a completely mobile optimised site (although it would be nice, and I did mention this in my FR on Bugzilla) but just for it to render properly on mobile and be able to use it without glitches would be a great start

While I totally agree with the notion that functionality & operation should be the primary focus of any HV system, I do believe that sometimes "you have to keep up with the Joneses". My point being; any OS is only as good as its clients/customers who actually use it, so if Proxmox is not going to keep up in attracting the wider client/customer base, it is eventually going to lag behind in its general usage, and so yes, if some potential users out there do want a snazzy GUI etc., put it out there. Imagine a basic scenario of a new user giving a different OS a spin "to see what its like", & he finds the general GUI cumbersome & not to his liking, he is going to move on to some other OS that actually maybe more snazzy but potentially overall less functional.

As has already been pointed out, Proxmox due to recent shifts in the HV scene, has been given a great & golden opportunity to attract a much wider user-base to this awesome open-source product, so the team must make every effort to use this chance to its fullest.

Let me just finish my rant, by yet again commending the whole Proxmox team on an absolutely great product, support & forum.


BTW: I also would appreciate if the desktop-view on a Phone (Android) could actually access the GUI somewhat functionally (snazziness not required).

Talking of people potentially moving on - there is some guy on the FR I opened that pretty much said that exact thing. They are looking at Hypervisors now and he was deterred by the UI!


Genuine question: what shifts in the Hypervisor scene?

I too am incredibly grateful to the Proxmox team, and appreciate their truly great products. I opened my feature request respectfully as a polite suggestions - I hope not one is taking offence to something that came purely from a good place!

As you mentioned that about the android app, this is the same with the mobile web view in PVE. It reminds me of an old iPhone web app, but just gives information. No interactions with the Node or VMs are possible.

 

[gfngfn256]

Genuine question: what shifts in the Hypervisor scene?

Just search this forum & the general web. Here is a hint, that also indicates the progress in Proxmox development to harness this HV evolvement.

 

[proxuser77]

In today's increasingly superficial world of 10 second attention spans and severely trigger-happy judgemental societies right down to inferior-critical-thinking trends... it would be a self-inflicted disaster if one were to not adapt, reimagine itself.

OK, I think I get your garden picnic analogy now.

But to be honest, if this is really the kind of audience that software for managing professional IT infrastructure should be aimed at, then I'd rather stick to my boring garden picnic, or maybe I don't go out at all anymore. After all, it might get dangerous out there when people with 10-second attention spans are in charge of maintaining critical infrastructure.

Seriously, I can of course understand the wish for a more modern, and mobile-friendly UI, at least to a certain extent. And yes, this can help attract new customers. But UI changes must be handled carefully, and functionality should always take precedence over flashy design elements. Otherwise, you risk alienating your existing customers. And new customers won’t stick around for long either if the product doesn’t live up to its shiny surface.

That said, I think Proxmox is already on the right track with the new frontend, which already made its way into the Datacenter Manager. Evolution, not revolution, is what professional users want. At least, that's what I think.

 

[bobmarley2021]

Just search this forum & the general web. Here is a hint, that also indicates the progress in Proxmox development to harness this HV evolvement.

Thanks, I did google and I was aware of that for quite a while but because that train wreck in slow motion has been happening for quite some time now, I wrongly assumed something else fortuitous [for Proxmox] had happened.