Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Nicolas George]

ce (12023-06-08):
> What about ads for car insurance?

Yes, what about them? What do you think they have special?

(Hint: an ad for a car insurance is not to convince you to subscribe to
any insurance rather than none, it is to convince you to subscribe to
this insurance rather than any other.)

-- 
  Nicolas George


signature.asc
Description: PGP signature

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[ce]

On 6/8/23 01:34, to...@tuxteam.de wrote:
Ad industry /is/ about convincing people to do things which 
potentially damage them. So it is deceptive by design. Read up on Big 
Tobacco for a good example.

What about ads for car insurance?

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[tomas]

On Thu, Jun 08, 2023 at 12:45:38AM +0200, Oliver Schoede wrote:
> 
> On Tue, 6 Jun 2023 06:05:18 +0200
>  wrote:
> 
> >On Mon, Jun 05, 2023 at 05:59:11PM -0400, Celejar wrote:
> >
> >[...]
> >
> >> The only case I can see in which such offloading would
> >> be unethical is where the website operator is somehow engaging in
> >> deceptive behavior, but assuming it is not [...]
> >
> >A pretty strong assumption given that the crushing maturity of
> >the internet is fuelled by the ad industry [...]

> So somehow there still is no such thing as free lunch. You could just
> as well "blame" a cable TV network for running all those ads, your TV
> set after all won't eat less power. No profit means no fancy
> shows,  sports, nor fancy websites. On the web things get quickly fuzzy
> of course, but in general neither is exactly deceptive.

Ad industry /is/ about convincing people to do things which potentially
damage them. So it is deceptive by design. Read up on Big Tobacco for
a good example.

>  We know what
> we're doing and what we're doing is voluntary and the catches, if not
> obvious, are obviously well known. Know a workaround or work without
> it. I'm still a (somewhat) regular terminal links user, a text browser
> that is, no javascript not to mention anything more demanding [...]

We are some kind of elite, don't forget that. Think of all those
folks pushed to standard browsers (banking) and smartphones (again,
banking, in some countries even basic public services).

They haven't the means to fight that; things are set up so they
don't even realise it, so most of the time they haven't even motive.
So it's on us.

> [...] After all those years uBlock
> Origin probably saved me tangible money too, especially with German
> electricity costs (who's to blame?), but then what's cheating?

Ah. uBlock. A free lunch, after all?

;-)

Cheers
-- 
t


signature.asc
Description: PGP signature

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Oliver Schoede]

On Tue, 6 Jun 2023 06:05:18 +0200
 wrote:

>On Mon, Jun 05, 2023 at 05:59:11PM -0400, Celejar wrote:
>
>[...]
>
>> The only case I can see in which such offloading would
>> be unethical is where the website operator is somehow engaging in
>> deceptive behavior, but assuming it is not [...]
>
>A pretty strong assumption given that the crushing maturity of
>the internet is fuelled by the ad industry, which, barred some
>exceptions, can be characterised as "deception for hire".
>
>Cheers

So somehow there still is no such thing as free lunch. You could just
as well "blame" a cable TV network for running all those ads, your TV
set after all won't eat less power. No profit means no fancy
shows,  sports, nor fancy websites. On the web things get quickly fuzzy
of course, but in general neither is exactly deceptive. We know what
we're doing and what we're doing is voluntary and the catches, if not
obvious, are obviously well known. Know a workaround or work without
it. I'm still a (somewhat) regular terminal links user, a text browser
that is, no javascript not to mention anything more demanding, find it
quite comfortable for text-dominated sites, like docs or Wikepedia,
doesn't go well with physics/math content though. Also ok for a quick
brush-up on news sites, where there's still a need, most don't work
anymore but some do, not the ads. After all those years uBlock
Origin probably saved me tangible money too, especially with German
electricity costs (who's to blame?), but then what's cheating?

Greetings,
Oliver

Re: A hypervisor for a headless server?

[rhkramer]

On Friday, June 02, 2023 11:34:58 AM Mario Marietto wrote:
> Excuse me,but there is something within your argumentation that I don't
> like and I want to express what it is. Let's take Linux as an example of
> what I want to say. Linux is well known to be an OS that can be installed
> on the old machines,helping the people that can't buy a new computer to
> surf the net and to do the basic things that they couldn't do using a more
> complete and modern PC built with new hardware components. And this is a
> linux quality that everyone loves and one of the reasons why Linux is
> growing faster on the market.  I think that everyone likes this,right ?
> So,why the same logic can't be applied to those software tools that go in
> the same direction,to those tools that help the users to have those
> functions that those old computers cannot give to them anymore ? Here it
> seems there is a contradiction. You may argue that developing for a small
> number of old computers isn't worth trying. But,first of all,I think that
> there are a LOT of old PCs in the world,since poor people aren't only a
> niche. We could discuss,anyway, how to reach the right amount of money to
> pay the developers. I'm thinking of opening a crowdfunding campaign for
> example. Or any other method to have the money that I can't imagine now. To
> do a project like this is socially accepted and helps to work on the
> perception of the users that computer science is something that they can
> use to develop their life in a good way. Maybe by helping one of those poor
> children,we are contributing to educate someone that in the future will
> make great things for humanity. I think that using the old relation that
> there is between costs and benefits is not applicable in every kind of
> situation. There are already a lot of people who work on projects that they
> like,but that they have a low social impact. Why not to work on a project
> that aims to extend the functions of an old PC. And what's better than
> using two operating systems on a single old pc? My old pc has 2 measly
> cpu's, I used one for the host and the other for linux emulated with bhyve
> and the performance was decent. Is there something that's more useful and
> generous than this kind of project ?

Creating a new acronym / proof reading mark:

nws;dr (no white space, didn't read)

-- 
rhk 

| No entity has permission to use this email to train an AI. 

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[tomas]

On Mon, Jun 05, 2023 at 05:59:11PM -0400, Celejar wrote:

[...]

> The only case I can see in which such offloading would
> be unethical is where the website operator is somehow engaging in
> deceptive behavior, but assuming it is not [...]

A pretty strong assumption given that the crushing maturity of
the internet is fuelled by the ad industry, which, barred some
exceptions, can be characterised as "deception for hire".

Cheers
-- 
t


signature.asc
Description: PGP signature

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Celejar]

On Sun, 4 Jun 2023 16:17:47 +0800
Bret Busby  wrote:

> On 4/6/23 14:32, Max Nikulin wrote:
> 
> > 
> > I believe, web site creators should be blamed more aggressively than 
> > browser developers for RAM requirements of contemporary web applications.
> > 
> 
> That was the point that I was making - I had not, as a twisted response 
> indicated, criticised Firefox regarding the misuse of resources - I 
> explicitly referred to malignant web application developers (for those 
> that do not understand the term, a web application is the application, 
> on the web application hosting server, that the user accesses, using a 
> web browser, not the web browser itself) that steal users' resources 
> using client-side processing (by using malware such as javascript using 
> client side processing), rather than properly and ethically using 
> server-side processing, such as .jsp or Perl .cgi applications.
> 
> The problem is that some web developers (and, especially, their 
> employers) offload the processing that should be done on the business 
> web application hosting server, to the victim users' personal computers. 
> It is a malignant exploitation, like the "gig economy".

I am quite puzzled by your perspective: you repeatedly express moral
indignation at the offloading of processing to users' machines, calling
this "malignant exploitation" and "steal[ing]" and implying that it is
unethical. Why? What duty does the website owe you to do any processing
at all for you? The only case I can see in which such offloading would
be unethical is where the website operator is somehow engaging in
deceptive behavior, but assuming it is not, why do you feel that there
is an ethical problem here? What right does a user have to demand
that someone else perform some processing for him?

-- 
Celejar

Re: A hypervisor for a headless server?

[tomas]

On Sun, Jun 04, 2023 at 11:44:11AM -0400, Michael Stone wrote:

[...]

> > But,first of all,I think that there are a LOT of old PCs in the world,since
> > poor people aren't only a niche.
> 
> I think you underestimate the scale of the e-waste problem. Simply giving
> people better, less obsolete, hardware is (IMO) a much better use of
> resources than trying to continue to use older hardware for no real reason
> other than a desire to use really old hardware. Even just considering
> environmental consciousness, saving a 5 year old PC from the landfill and
> throwing out a 10 year old PC is a net positive in terms of energy
> efficiency if nothing else.

There is some truth to this. In this sweeping generality, though, this
is more false than true.

Cheers
-- 
t


signature.asc
Description: PGP signature

Re: 10 year old machines are slow (was: A hypervisor for a headless server?)

[Curt]

On 2023-06-02, Bret Busby  wrote:
>
> Whoever posted the message to which the above message is a reply, is 

An enduring mystery to know why Monnier refuses the convention of
attributions. Then again, one of the smaller mysteries. 

Re: A hypervisor for a headless server?

[Michael Stone]

On Fri, Jun 02, 2023 at 05:18:38PM +0200, zithro wrote:

On 02 Jun 2023 14:31, Michael Stone wrote:
I don't recommend xen for new projects. It has more pieces and tends 
to be more fragile than qemu+kvm, for no real benefits these days. 
(IMO)


Define "more pieces" and "more fragile" ?


You need to juggle kernel version, qemu version, and xen version. You 
need a bootable dom0 *as well as* a bootable xen hypervisor. If any of 
these things mismatch or stop working, things break. The xen-specific 
pieces are generally less well known and less operationally tested 
because there are fewer users. The xen developers have gone through 
several vm models and various deprecations in the past few years, and 
there have been actual breakages for users of the debian packages due to 
the many combinations of features which can break in the presence of 
changes (such as changes needed for security issues) and the difficulty 
(infeasibility?) of testing all the possible combinations. That would be 
less of an issue if rolling your own and tracking xen upstream directly, 
but this is a debian list, and the debian packages face a different set 
of constraints.



It has a really low TCB and still used by amazon for their cloud.


As a legacy service. New VMs are deployed using different technologies. 
They were the only major cloud service to go with xen, and their 
continued use seems more a matter of leaving it running for legacy 
instances being less work than migrating everything. (Which is basically 
where I still have deployed.) Amazon is also not using a xen package 
from a general purpose OS, and has quite a large team devoted to the 
care and feeding of that infrastructure. It's basically an apples to 
boxcars comparision unless the person trying to decide which hypervisor 
to go with happens to be running one of the largest clouds in the world. 
(Which begs the question of why on earth they'd be looking for answers 
on debian-user.)



You don't even need qemu if running fully virtualized guests (PV/PVH).


xen's continuing search for the next great thing 
(pv/hvm/pvhvm/pvh/pvhv2) has itself been a source of operational pain. 
From the perspective of taking the best advantage of the technology 
available at the time it's great, but from the perspective of wanting to 
set something up and just have it keep running, it's a pain. (And, to 
the point, kvm has been less of a pain because for better or worse its 
model has remained more stable.)


None of this is to say that xen is a bad project or that some people may 
find it the best option, but I'll continue to not recommended it as a 
general solution for people looking to deploy a new vm environment. It's 
just easier to go with kvm.

Re: A hypervisor for a headless server?

[Michael Stone]

On Fri, Jun 02, 2023 at 05:34:58PM +0200, Mario Marietto wrote:

Excuse me,but there is something within your argumentation that I don't like
and I want to express what it is. Let's take Linux as an example of what I want
to say. Linux is well known to be an OS that can be installed on the old
machines,helping the people that can't buy a new computer to surf the net and
to do the basic things that they couldn't do using a more complete and modern
PC built with new hardware components. And this is a linux quality that
everyone loves and one of the reasons why Linux is growing faster on the
market.


This is a misunderstanding of a general purpose OS. netbsd is the 
project that supports old hardware for the sake of old hardware. Linux 
has never been that, and when the choice comes to supporting something 
old or supporting something new, the decision is generally to abandon 
the old. There's no other way to make progress. If both can coexist, 
fine--but if you want something that specifically caters to 
(functionally obsolete) hardware you need a project with that as a goal 
rather than a project that has general utility as a goal.



But,first of all,I think that there are a LOT of old PCs in the world,since
poor people aren't only a niche.


I think you underestimate the scale of the e-waste problem. Simply 
giving people better, less obsolete, hardware is (IMO) a much better use 
of resources than trying to continue to use older hardware for no real 
reason other than a desire to use really old hardware. Even just 
considering environmental consciousness, saving a 5 year old PC from the 
landfill and throwing out a 10 year old PC is a net positive in terms of 
energy efficiency if nothing else.

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Stefan Monnier]

> With no client-side javascript, it's not possible to change just a part of
> a web page[0]. The server must send the whole web page to be rendered by the
> client. So while it decreases CPU usage in the client, it increases network
> usage. Isn't it unethical to also "steal" more bandwidth than necessary?

Indeed, early uses of client side (Javascript) processing really helped
make web sites more efficient: for the server, the client, and the
network in between.

And then web developers realized that a browser-with-Javascript is just
a sort of VM.  So now we have "web applications" running in that VM,
where the backward/forward buttons make you leave/reenter the
application rather than move through past states of it, and you can't
use bookmarks to refer to the current state any more :-(


Stefan

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[tomas]

On Sun, Jun 04, 2023 at 10:34:04AM -0400, Greg Wooledge wrote:
> On Sun, Jun 04, 2023 at 04:30:46PM +0200, to...@tuxteam.de wrote:
> > So the practice is that the whole internet dumps the whole framework
> > schtack [2] on you.
> 
> https://en.wikipedia.org/wiki/WebAssembly

We need better craftspeople, not better tools.

And no, I'm not actually blaming the people themselves, but an
environment which doesn't encourage that.

Cheers
-- 
t


signature.asc
Description: PGP signature

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Greg Wooledge]

On Sun, Jun 04, 2023 at 04:30:46PM +0200, to...@tuxteam.de wrote:
> So the practice is that the whole internet dumps the whole framework
> schtack [2] on you.

https://en.wikipedia.org/wiki/WebAssembly

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[tomas]

On Sun, Jun 04, 2023 at 08:17:43AM -0300, Eduardo M KALINOWSKI wrote:

[...]

> With no client-side javascript, it's not possible to change just a part of a
> web page[0]. The server must send the whole web page to be rendered by the
> client. So while it decreases CPU usage in the client, it increases network
> usage. Isn't it unethical to also "steal" more bandwidth than necessary?
> 
> [0] There are frames (now deprecated) and iframes, but they only get you so
> far. And each (i)frame must be a complete html page.

This is the theory, yes. In practice, here's one example: my browser takes
roughly 12sec to "boot" our company chat app (a stripe.js monster, AFAICS).

All that to ask me whether I want to download their "native" [1] app or
"view" the thing in the browser. When I opt for the browser it continues
"booting" for a few secs.

So the practice is that the whole internet dumps the whole framework
schtack [2] on you.

Cheers

[1] An electron app. Yeah, right.
[2] A pun, not a typo.
-- 
t


signature.asc
Description: PGP signature

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[songbird]

Max Nikulin wrote:
...
> I believe, web site creators should be blamed more aggressively than 
> browser developers for RAM requirements of contemporary web applications.

  no kidding, rather poor design in many web sites these 
days, loading and reloading images, large images for 
little purpose, videos which don't really show or say
much, etc.

  my biggest peeves in recent times is login pages which
are full of stuff (when all i want to do is login.  don't
make it a mess which takes too long to load up.  just 
let me login, ok?  grrr!) and pages which want me to 
accept their cookies but are so full of stuff if i click
too soon i get an error, so i'm having to wait a few 
moments before i can click.


  songbird

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Eduardo M KALINOWSKI]

On 04/06/2023 05:17, Bret Busby wrote:

On 4/6/23 14:32, Max Nikulin wrote:



I believe, web site creators should be blamed more aggressively than 
browser developers for RAM requirements of contemporary web applications.




That was the point that I was making - I had not, as a twisted response 
indicated, criticised Firefox regarding the misuse of resources - I 
explicitly referred to malignant web application developers (for those 
that do not understand the term, a web application is the application, 
on the web application hosting server, that the user accesses, using a 
web browser, not the web browser itself) that steal users' resources 
using client-side processing (by using malware such as javascript using 
client side processing), rather than properly and ethically using 
server-side processing, such as .jsp or Perl .cgi applications.


The problem is that some web developers (and, especially, their 
employers) offload the processing that should be done on the business 
web application hosting server, to the victim users' personal computers. 
It is a malignant exploitation, like the "gig economy".


With no client-side javascript, it's not possible to change just a part 
of a web page[0]. The server must send the whole web page to be rendered 
by the client. So while it decreases CPU usage in the client, it 
increases network usage. Isn't it unethical to also "steal" more 
bandwidth than necessary?


[0] There are frames (now deprecated) and iframes, but they only get you 
so far. And each (i)frame must be a complete html page.


And even with regards to CPU usage your model might not be so great. 
Instead of re-rendering just the part of the page that needs to be 
changed (say, the message pane in a webmail application), with no 
client-side scripting the whole interface must be re-rendered, which can 
be resource intensive. So while I'd agree that with client-side 
scripting resource usage in the client is higher, it might not be as 
higher as you think.




--
Eduardo M KALINOWSKI
edua...@kalinowski.com.br

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Bret Busby]

On 4/6/23 14:32, Max Nikulin wrote:



I believe, web site creators should be blamed more aggressively than 
browser developers for RAM requirements of contemporary web applications.




That was the point that I was making - I had not, as a twisted response 
indicated, criticised Firefox regarding the misuse of resources - I 
explicitly referred to malignant web application developers (for those 
that do not understand the term, a web application is the application, 
on the web application hosting server, that the user accesses, using a 
web browser, not the web browser itself) that steal users' resources 
using client-side processing (by using malware such as javascript using 
client side processing), rather than properly and ethically using 
server-side processing, such as .jsp or Perl .cgi applications.


The problem is that some web developers (and, especially, their 
employers) offload the processing that should be done on the business 
web application hosting server, to the victim users' personal computers. 
It is a malignant exploitation, like the "gig economy".


..
Bret Busby
Armadale
West Australia
(UTC+0800)
..

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Max Nikulin]

On 03/06/2023 18:37, The Wanderer wrote:

On 2023-06-03 at 07:18, Max Nikulin wrote:

On 03/06/2023 17:40, The Wanderer wrote:


Hey, now. I once had a Firefox session (with "restore tabs from
previous session" enabled, and about six-to-eight windows) with
5,190 open tabs, and that computer only had 24GB of RAM.


Modern browsers supports "unloaded" tabs, so most of your tabs likely
were similar to bookmarks with page resources not loaded to RAM.


That feature was, AFAIK, first introduced in the BarTab addon which I
mentioned. So, yes, and although in hindsight I didn't state it
explicitly, I intended to convey that by mentioning that addon.


Sorry, I never used BarTab, so I was unaware that tab unloading appeared 
in this add-on earlier than in Firefox. For me an "open tab" is the one 
that is rendered, has DOM tree in memory and perhaps running JS, webasm, 
animated images and styles, so some pages may be really hungry for RAM.


Most of your tabs are just some records and will load resources from net 
when you really open them.


I appreciate that browsers limit consumed resources by unloading page 
content when a tab is not accessed for some period of time. It is great 
that users may have hundreds of tabs despite I mostly have no more than 
a couple of dozens.


I just would not call a tab "open" because I consider it as a synonym to 
"loaded". Anyway add-ons for advanced tab management hides most of them.


I believe, web site creators should be blamed more aggressively than 
browser developers for RAM requirements of contemporary web applications.



P.S. Perhaps in future tabs as UI element in browsers will be merged
with bookmarks and browsing history. The only prerequisite to better
save state of scroll position and partially filled forms.


I'm not sure quite what you're envisioning, but one reason why I keep so
many open tabs rather than using e.g. bookmarks instead is because I
want to be able to preserve forward/back history within each tab; I
don't know of any other feature that enables doing that.


Thanks, I have never considered such use case, but it is not against of 
fusing of tabs, bookmarks, and history. Your tabs are a kind of advanced 
bookmarks, a favorite nodes in browsing history graph. Current bookmark 
UI is just too limited in browsers, so tabs are more flexible and more 
convenient for you.


I mostly open new tabs to follow links (actually it is more close to 
enqueue a page for reading). That is why usefulness of forward-backward 
history is quite limited for me. Unfortunately opener is not saved for 
tabs. (I consider annotating of visited pages is more important, but it 
is another story.)


"Pure" tabs are hot cache of rendered pages where current DOM state is 
important for following interaction. Everything else are just records in 
some database. For me, tabs UI is a kind of L1 cache, a subset of pages 
closely related to the current or planned soon activity.


Of course, I do not insist that everybody should think of browser UI in 
my terms, it is just a point of view.

Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?

[Bret Busby]

On 3/6/23 20:26, Dan Ritter wrote:

Bret Busby wrote:





I don't see how they can be both cheap and cost far too much.

-dsr-



Cheap and nasty construction, selling for excessive prices.

"Here is this thing that cost me a dollar to make. I will sell it to you 
for a hundred dollars, with no worthwhile warranty."


..
Bret Busby
Armadale
West Australia
(UTC+0800)
..

Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?

[Dan Ritter]

Bret Busby wrote: 
> Last year, I bought the computer described below, as a refurbished machine,
> and, it is far superior to the new computers that do not come with enough
> RAM to be worthwhile.
> 
> This computer, with 128GB RAM, I regard as far superior to an i9 computer
> with 8GB RAM.

OK.

> The new computers are rubbish.

No evidence for that, and the statement is so broad as to be
useless.

>   Info: 14-core model: Intel Xeon E5-2660 v4 bits: 64 type: MT MCP arch:
> Broadwell rev: 1 cache:

First sale date: Q1 2016.

The discussion about relative speed of old computers is specific
to a feature which became ubiquitous in 2012, and the
speculation about not being worse than 3x was specifically about
2010 vs current models.

> Some computers, like this one, perform far better, than the cheap and nasty
> new computers (which cost far more, and, far too much)

I don't see how they can be both cheap and cost far too much.

-dsr-

Re: 10 year old machines are slow (was: A hypervisor for a headless server?)

[Dan Ritter]

Bret Busby wrote: 
> 
> Whoever posted the message to which the above message is a reply, is showing
> a lack of knowledge of computers; the "speed" of a computer, involves more
> components than simply the CPU - an i9 with 2GB of RAM, will probably not be
> as "fast" as in i3 with 32GB of RAM.


The context of the discussion is that there is a specific
feature -- virtualization support -- which means that for people
running VMs, an old computer produced with that feature is much
faster than its twin produced six months earlier.

Thanks for telling me that I lack knowledge of computers. I
shall keep that in mind.

-dsr-

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[The Wanderer]

On 2023-06-03 at 07:18, Max Nikulin wrote:

> On 03/06/2023 17:40, The Wanderer wrote:
> 
>> Hey, now. I once had a Firefox session (with "restore tabs from
>> previous session" enabled, and about six-to-eight windows) with
>> 5,190 open tabs, and that computer only had 24GB of RAM.
> 
> Modern browsers supports "unloaded" tabs, so most of your tabs likely
> were similar to bookmarks with page resources not loaded to RAM.

That feature was, AFAIK, first introduced in the BarTab addon which I
mentioned. So, yes, and although in hindsight I didn't state it
explicitly, I intended to convey that by mentioning that addon.

> P.S. Perhaps in future tabs as UI element in browsers will be merged
> with bookmarks and browsing history. The only prerequisite to better
> save state of scroll position and partially filled forms.

I'm not sure quite what you're envisioning, but one reason why I keep so
many open tabs rather than using e.g. bookmarks instead is because I
want to be able to preserve forward/back history within each tab; I
don't know of any other feature that enables doing that.

I also can't think of another UI paradigm for interfacing with such a
setup that would work any better than tabs do.

-- 
   The Wanderer

The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man. -- George Bernard Shaw



signature.asc
Description: OpenPGP digital signature

Re: Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[Max Nikulin]

On 03/06/2023 17:40, The Wanderer wrote:


Hey, now. I once had a Firefox session (with "restore tabs from previous
session" enabled, and about six-to-eight windows) with 5,190 open tabs,
and that computer only had 24GB of RAM.


Modern browsers supports "unloaded" tabs, so most of your tabs likely 
were similar to bookmarks with page resources not loaded to RAM.


P.S. Perhaps in future tabs as UI element in browsers will be merged 
with bookmarks and browsing history. The only prerequisite to better 
save state of scroll position and partially filled forms.

Firefox resource utilization (was Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?)

[The Wanderer]

On 2023-06-03 at 01:41, Nicholas Geovanis wrote:

> On Fri, Jun 2, 2023, 6:10 PM Bret Busby  wrote:

>> This computer, with 128GB RAM, I regard as far superior to an i9 
>> computer with 8GB RAM.

>> Refurbished computer profile (with 128GB RAM (that runs about 200 
>> windows of Firefox (I have one saved session, with 229 windows,
>> and about 3200 tabs), while viewing movies (I also have about 10
>> movies open at present, in Celluloid and SMPlayer), although, at
>> present, I have only about 127 Firefox windows open, with 1689
>> tabs):
> 
> Holy cow! :-) No wonder you have 128GB RAM. You will need that much
> for that much Firefox. It's a peeve of mine how resource intensive it
> is for a browser compared to the competition.

Hey, now. I once had a Firefox session (with "restore tabs from previous
session" enabled, and about six-to-eight windows) with 5,190 open tabs,
and that computer only had 24GB of RAM.

...And it used BarTab, or rather a variant that I maintained myself
after the original maintainer decided getting a small subset of that
addon's functionality integrated into the upstream browser was sufficient.

I'm doing much better now; I've got only 1,491 open tabs, split across
eight browser windows, on a machine with 64GB of RAM. (I think I had it
down under 1,000 total tabs at one point. I still hope to get it down to
a range of 200-500 total tabs; I don't think I'll realistically be
comfortable running with fewer than that, given what I use them for.)

And all of this is with older Firefox versions, which are considerably
more resource-intensive in my experience than newer Firefox is.

I may dislike a lot of things about where Mozilla has taken Firefox, but
I do prefer to see it criticized for the faults it actually has.
Resource-intensive it may be, but unless extra windows increase resource
usage far more than extra tabs do, you don't need anything *close* to
128GB for ~1700 Firefox tabs in ~128 Firefox windows.

-- 
   The Wanderer

The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man. -- George Bernard Shaw



signature.asc
Description: OpenPGP digital signature

Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?

[Bret Busby]

On 3/6/23 13:41, Nicholas Geovanis wrote:
On Fri, Jun 2, 2023, 6:10 PM Bret Busby > wrote:


On 3/6/23 06:33, Nicholas Geovanis wrote:
 >
 >
 > On Fri, Jun 2, 2023, 4:49 PM Bret Busby mailto:b...@busby.net>
 > >> wrote:
 >
 >     On 2/6/23 23:55, James H. H. Lampert wrote:
 >
 >     
 >
 >      > Luddites of the World Unite! You have nothing to lose but your
 >     upgrade
 >      > treadmills
 >
 >     If, by upgrade treadmills, you mean the flatbed treadmills,
that have a
 >     belt that is turned by the human walking on it, rather than the
 >     electric
 >     ones with electric motors for lazy humans, the ones that have
the belt
 >
 >
 > I'm afraid he meant the treadmill that used to be called "planned
 > obsolescence". The thought that a perfectly satisfactory machine no
 > longer suffices for you because it is "yesterday's model".
Thereafter it
 > will stop working with newer machines (or software) which are
intended
 > to be incompatible with it.
 > And what is the end in view?
 > Sell you a new machine.
 >
 >

Interesting.

Last year, I bought the computer described below, as a refurbished
machine, and, it is far superior to the new computers that do not come
with enough RAM to be worthwhile.

This computer, with 128GB RAM, I regard as far superior to an i9
computer with 8GB RAM.
.

Refurbished computer profile (with 128GB RAM (that runs about 200
windows of Firefox (I have one saved session, with 229 windows, and
about 3200 tabs), while viewing movies (I also have about 10 movies
open
at present, in Celluloid and SMPlayer), although, at present, I have
only about 127 Firefox windows open, with 1689 tabs):


Holy cow! :-)
No wonder you have 128GB RAM. You will need that much for that much 
Firefox. It's a peeve of mine how resource intensive it is for a browser 
compared to the competition.




The problem in the demand for resources via web browsers, is the 
gratuitous malicious use of javascript; client side processing, that 
steals a user's resources, rather than server side processing, which is 
what ethical web application developers use (server side processing, 
that is, that is used by ethical web application developers).


Running SeaMonkey, with javascript disabled, uses hardly any resources; 
on my i7, 16GB RAM, All-In-One (also, a refurbished Dell), I have 
currently 80 Firefox Windows, 20 Pale Moon Windows, and, (free of 
javascript) 16 SeaMonkey Windows running.


Also, having the primary HD, with an appropriate (I use 32GB as 
standard) swap partition, as an NVVME (?) SSD, causes the systems to run 
better.


My point was, and, is, that the "speed" of a system is not solely 
reliant on, and, should not be assessed solely on, the age of the CPU; 
many other factors, including the amount of RAM, the speed and capacity 
of the primary HD, and, the responsible use of a swap partition, should, 
I believe, be taken into account.


..
Bret Busby
Armadale
West Australia
(UTC+0800)
..

Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?

[Nicholas Geovanis]

On Fri, Jun 2, 2023, 6:10 PM Bret Busby  wrote:

> On 3/6/23 06:33, Nicholas Geovanis wrote:
> >
> >
> > On Fri, Jun 2, 2023, 4:49 PM Bret Busby  > > wrote:
> >
> > On 2/6/23 23:55, James H. H. Lampert wrote:
> >
> > 
> >
> >  > Luddites of the World Unite! You have nothing to lose but your
> > upgrade
> >  > treadmills
> >
> > If, by upgrade treadmills, you mean the flatbed treadmills, that
> have a
> > belt that is turned by the human walking on it, rather than the
> > electric
> > ones with electric motors for lazy humans, the ones that have the
> belt
> >
> >
> > I'm afraid he meant the treadmill that used to be called "planned
> > obsolescence". The thought that a perfectly satisfactory machine no
> > longer suffices for you because it is "yesterday's model". Thereafter it
> > will stop working with newer machines (or software) which are intended
> > to be incompatible with it.
> > And what is the end in view?
> > Sell you a new machine.
> >
> >
>
> Interesting.
>
> Last year, I bought the computer described below, as a refurbished
> machine, and, it is far superior to the new computers that do not come
> with enough RAM to be worthwhile.
>
> This computer, with 128GB RAM, I regard as far superior to an i9
> computer with 8GB RAM.
> .
>
> Refurbished computer profile (with 128GB RAM (that runs about 200
> windows of Firefox (I have one saved session, with 229 windows, and
> about 3200 tabs), while viewing movies (I also have about 10 movies open
> at present, in Celluloid and SMPlayer), although, at present, I have
> only about 127 Firefox windows open, with 1689 tabs):


Holy cow! :-)
No wonder you have 128GB RAM. You will need that much for that much
Firefox. It's a peeve of mine how resource intensive it is for a browser
compared to the competition.

Ned Ludd had his head screwed on straight. And was apparently a legendary
lover :-)
I have read that 3 Luddite sledgehammers have survived. There's your
solution for obsolescent  machinery :-)

Some computers, like this one, perform far better, than the cheap and
> nasty new computers (which cost far more, and, far too much), with the
> new computers being best described as rubbish, produced by increasingly
> malicious manufacturers (that make freedom of choice of operating
> systems, and, performance, impossible)
>



..
> Bret Busby
> Armadale
> West Australia
> (UTC+0800)
> ..
>
>

Re: 10 year old machines are slow (was: A hypervisor for a headless server?)

[David Christensen]

On 6/2/23 12:19, Stefan Monnier wrote:

The most recent general-purpose Intel CPU without VT-X is from 2012.

[...]

*everything* on processors that old is slow.


Actually, for many (most?) single-threaded applications, I wouldn't be
surprised if some 2010 CPUs end up within a factor 3 of the most badass
desktop you can find today.



I do Perl development and like to run my module test suites in parallel 
to shorten the develop-test cycle.  Tuned correctly, I can keep all 8 
threads busy on my quad-core Hyper-Threading processors for the majority 
of the run time.



These are the test platforms (all have SATA 6 Gbps SSD's and enough 
memory to avoid swapping):


* Dell Latitude E6520 laptop with Core i7-2720QM (Q1'11)

* Homebrew Antec tower with Intel DQ67SW desktop board and Core i7-2600S 
processor (Q1'11)


* Dell Precision 3630 with Xeon E-2174G (Q3'18)


I will look up the PassMark CPU Mark scores for the various processors, 
gather Perl module test suite run time data, and do a Power Regression 
analysis with LibreOffice Calc:


https://help.libreoffice.org/6.1/en-US/text/scalc/01/statistics_regression.html?DbPAR=CALC

https://www.cpubenchmark.net/cpu_list.php


Here is the data table for the regression analysis:

x   y
CPU MarkTest Time
(dim)   (seconds)
Core i7-2720QM  406824.033
Core i7-2600S   459422.985
Xeon E-2174G971216.544


Here is the resulting equation:

y = exp(6.779011065)*(x^-0.43266897)


Here are the predicted Test Time values for a Core i7-2600 (Q1'11) and a 
Core i7-13700 (Q1'23):


x   y
CPU MarkTest Time
(dim)   (seconds)
i7-2600 533021.461
i7-13700389959.072


So, the Core i7-13700 is predicted to be faster than the Core i7-2600 by 
a factor of 21.461 / 9.072 = 2.366.



David

Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?

[Bret Busby]

On 3/6/23 06:33, Nicholas Geovanis wrote:



On Fri, Jun 2, 2023, 4:49 PM Bret Busby > wrote:


On 2/6/23 23:55, James H. H. Lampert wrote:



 > Luddites of the World Unite! You have nothing to lose but your
upgrade
 > treadmills

If, by upgrade treadmills, you mean the flatbed treadmills, that have a
belt that is turned by the human walking on it, rather than the
electric
ones with electric motors for lazy humans, the ones that have the belt


I'm afraid he meant the treadmill that used to be called "planned 
obsolescence". The thought that a perfectly satisfactory machine no 
longer suffices for you because it is "yesterday's model". Thereafter it 
will stop working with newer machines (or software) which are intended 
to be incompatible with it.

And what is the end in view?
Sell you a new machine.




Interesting.

Last year, I bought the computer described below, as a refurbished 
machine, and, it is far superior to the new computers that do not come 
with enough RAM to be worthwhile.


This computer, with 128GB RAM, I regard as far superior to an i9 
computer with 8GB RAM.


And, on this old computer, apart from (as part of its refurbishment), is 
the 500GB NVVME (?) SSD primary hard drive, the 6TB internal second HD, 
and, using some of the (about)6 USB ports, I have an external USB HDD 
(about 2TB), and a T5 and a T7 external USB SSD drive, with room for 
more; the T5 and T7 drives using the exFAT file system, with 
extraordinarily fast data transfer rates.


So, old computers like this one, are superior to new computers.

And, for years, when dialup computing was used, I had used for a 
mailserver, a used, low spec, HP server that I bought for 100AUD, that 
had an MMX CPU, and, was quite adequate to be a mailserver, running 
postfix and procmail, and, whatever version of Debian was on it, until 
dialup was superseded by "broadband", for which, the modems imposed 
DHCP, rather than static IP addresses, and I had to give up running my 
own mailserver, because it became too complicated, when I could no 
longer use static IP addresses.


And, this computer (not the ex-mailserver) cost about as much as a 
bottom of the range new computer.


The new computers are rubbish.

Refurbished computer profile (with 128GB RAM (that runs about 200 
windows of Firefox (I have one saved session, with 229 windows, and 
about 3200 tabs), while viewing movies (I also have about 10 movies open 
at present, in Celluloid and SMPlayer), although, at present, I have 
only about 127 Firefox windows open, with 1689 tabs):


"
Machine:
  Type: Desktop System: Dell product: Precision Tower 5810 v: N/A 
serial: 

Chassis: type: 7 serial: 
  Mobo: Dell model: 0K240Y v: A02 serial:  UEFI: 
Dell v: A34

date: 10/19/2020
CPU:
  Info: 14-core model: Intel Xeon E5-2660 v4 bits: 64 type: MT MCP 
arch: Broadwell rev: 1 cache:

L1: 896 KiB L2: 3.5 MiB L3: 35 MiB
  Speed (MHz): avg: 1491 high: 2880 min/max: 1200/3200 cores: 1: 1198 
2: 2539 3: 1199 4: 1197
5: 2827 6: 1197 7: 1198 8: 1197 9: 1197 10: 1197 11: 1202 12: 1198 
13: 1357 14: 1201 15: 1199
16: 2880 17: 1197 18: 1197 19: 2727 20: 1197 21: 1198 22: 1304 23: 
1197 24: 1197 25: 2828

26: 1198 27: 1353 28: 1197 bogomips: 111740
  Flags: avx avx2 ht lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx
Graphics:
  Device-1: NVIDIA TU116 [GeForce GTX 1660 SUPER] vendor: ASUSTeK 
driver: nvidia v: 525.105.17
pcie: speed: 8 GT/s lanes: 16 ports: active: none off: HDMI-A-1 
empty: DP-1,DVI-D-1

bus-ID: 03:00.0 chip-ID: 10de:21c4
  Device-2: Sunplus Innovation AAPDQT-0622-W type: USB driver: 
snd-usb-audio,uvcvideo

bus-ID: 3-13:6 chip-ID: 1bcf:2cb4
  Display: x11 server: X.Org v: 1.21.1.4 compositor: marco v: 1.26.0 
driver: X: loaded: nvidia
unloaded: fbdev,modesetting,nouveau,vesa gpu: nvidia display-ID: :0 
screens: 1

  Screen-1: 0 s-res: 1920x1080 s-dpi: 93
  Monitor-1: HDMI-0 res: 1920x1080 dpi: 94 diag: 598mm (23.5")
  OpenGL: renderer: NVIDIA GeForce GTX 1660 SUPER/PCIe/SSE2 v: 4.6.0 
NVIDIA 525.105.17

direct render: Yes
...
Info:
  Processes: 556 Uptime: 9d 12h 26m Memory: 125.72 GiB used: 99.99 GiB 
(79.5%) Init: systemd
  v: 249 runlevel: 5 Compilers: gcc: 11.3.0 alt: 11/12 Client: Unknown 
python3.10 client

  inxi: 3.3.13
"

Some computers, like this one, perform far better, than the cheap and 
nasty new computers (which cost far more, and, far too much), with the 
new computers being best described as rubbish, produced by increasingly 
malicious manufacturers (that make freedom of choice of operating 
systems, and, performance, impossible).


..
Bret Busby
Armadale
West Australia
(UTC+0800)
..

Re: 10 year old machines are slow (was: A hypervisor for a headless server?)

[Bret Busby]

On 3/6/23 03:19, Stefan Monnier wrote:

The most recent general-purpose Intel CPU without VT-X is from 2012.

[...]

*everything* on processors that old is slow.


Actually, for many (most?) single-threaded applications, I wouldn't be
surprised if some 2010 CPUs end up within a factor 3 of the most badass
desktop you can find today.


 Stefan




Since the above message refers to "10 year old" computers, if a person 
searches the list archives, I had posted to this, and, other operating 
system lists, regarding a computer that I bought in 2013 (which is ten 
years ago, this year), which was so advanced, that only two non-MS 
operating systems had drivers for the CPU; it had an Intel i7 CPU, with 
32GB RAM, and, until it stopped working last year or this year, due to a 
grid electricity failure, which, I think, wrecked the power supply for 
the computer (an Acer Aspire "laptop"; - a V3-772G), that computer never 
gave me cause to think it slow.


Whoever posted the message to which the above message is a reply, is 
showing a lack of knowledge of computers; the "speed" of a computer, 
involves more components than simply the CPU - an i9 with 2GB of RAM, 
will probably not be as "fast" as in i3 with 32GB of RAM.


..
Bret Busby
Armadale
West Australia
(UTC+0800)
..

Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?

[Nicholas Geovanis]

On Fri, Jun 2, 2023, 4:49 PM Bret Busby  wrote:

> On 2/6/23 23:55, James H. H. Lampert wrote:
>
> 
>
> > Luddites of the World Unite! You have nothing to lose but your upgrade
> > treadmills
>
> If, by upgrade treadmills, you mean the flatbed treadmills, that have a
> belt that is turned by the human walking on it, rather than the electric
> ones with electric motors for lazy humans, the ones that have the belt


I'm afraid he meant the treadmill that used to be called "planned
obsolescence". The thought that a perfectly satisfactory machine no longer
suffices for you because it is "yesterday's model". Thereafter it will stop
working with newer machines (or software) which are intended to be
incompatible with it.
And what is the end in view?
Sell you a new machine.


> that is turned by the human walking on it, having a slight, and,
> adjustable upward grade, then, such treadmills should definitely not be
> abandoned.
>
> The human powered (rather than electric powered) treadmills are far more
> healthy, both directly for the human powering the treadmill, and, for
> the environment, especially, given that most electricity is generated
> either by burning things, and therefore, creating atmospheric pollution,
> and, poisoning most lifeforms, or, by nuclear meltdowns, causing
> radioactive poisoning, and, even worse toxic waste, than from burning
> things.
>
> So, human powered treadmills, that involve an upward grade, should not
> be abandoned, the abandonment of which treadmills, threatens life, for
> the sake of ever-increasing laziness.
>
> ..
> Bret Busby
> Armadale
> West Australia
> (UTC+0800)
> ..
>
>

Re: A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?

[Bret Busby]

On 2/6/23 23:55, James H. H. Lampert wrote:



Luddites of the World Unite! You have nothing to lose but your upgrade 
treadmills


If, by upgrade treadmills, you mean the flatbed treadmills, that have a 
belt that is turned by the human walking on it, rather than the electric 
ones with electric motors for lazy humans, the ones that have the belt 
that is turned by the human walking on it, having a slight, and, 
adjustable upward grade, then, such treadmills should definitely not be 
abandoned.


The human powered (rather than electric powered) treadmills are far more 
healthy, both directly for the human powering the treadmill, and, for 
the environment, especially, given that most electricity is generated 
either by burning things, and therefore, creating atmospheric pollution, 
and, poisoning most lifeforms, or, by nuclear meltdowns, causing 
radioactive poisoning, and, even worse toxic waste, than from burning 
things.


So, human powered treadmills, that involve an upward grade, should not 
be abandoned, the abandonment of which treadmills, threatens life, for 
the sake of ever-increasing laziness.


..
Bret Busby
Armadale
West Australia
(UTC+0800)
..

Re: 10 year old machines are slow (was: A hypervisor for a headless server?)

[Dan Ritter]

Stefan Monnier wrote: 
> > The most recent general-purpose Intel CPU without VT-X is from 2012.
> [...]
> > *everything* on processors that old is slow.
> 
> Actually, for many (most?) single-threaded applications, I wouldn't be
> surprised if some 2010 CPUs end up within a factor 3 of the most badass
> desktop you can find today.

In 2010, Intel released the i7-930. The i7-975 was released in June
of 2009. In March of 2010, the flagship for speed was the Xeon X7542,
but that's a server CPU.

In 2010, AMD was not competitive with Intel per-core, so we can
ignore them.

PassMark's single thread benchmark is currently won by the Intel
i9-13900ks - score 4796.

The i7-930 gets a 1271

The i7-975 gets a 1489

The Xeon X5698 -- not releases until 2011, but I can't find a
benchmark for the X7542 -- gets a 1922. It's also a server CPU,
not a desktop.

 4796 / 1489 = 3.22

I will admit that this is a synthetic benchmark. It is not my
synthetic benchmark, and a mistake I made earlier led me to
write up an admission that you were right -- until I realized
that I had been looking at the multicore benchmarks of all the
older CPUs, not the single thread.

3.22 is pretty close to 3x, though. It seems likely that if you
had a single-threaded task that didn't rely on RAM bandwidth or
disk latency or bandwidth, you would actually see just a 3x
difference by 2 years later in fairly mainstream CPUs - an
i5-2550, for example.

-dsr-

Re: A hypervisor for a headless server?

[Mario Marietto]

I remembered it badly. I thought I'd never been able to use kvm on
that old PC, but I've just been able to create a FreeBSD vm
on top of Ubuntu 14.04. Qemu and kvm work wonderfully. Probably some
virtualization features miss on the cpu,but...it works anyway.


On Fri, Jun 2, 2023 at 9:07 PM Stefan Monnier 
wrote:

> > Intel I5 cpu,
>
> Side note: this doesn't say much more than "an amd64 CPU produced by
> Intel in the last 13 years and whose release price was somewhere between
> $200 and $300" :-)
>
> See https://en.wikipedia.org/wiki/List_of_Intel_Core_i5_processors for
> more details.
>
>
> Stefan
>
>

-- 
Mario.

10 year old machines are slow (was: A hypervisor for a headless server?)

[Stefan Monnier]

> The most recent general-purpose Intel CPU without VT-X is from 2012.
[...]
> *everything* on processors that old is slow.

Actually, for many (most?) single-threaded applications, I wouldn't be
surprised if some 2010 CPUs end up within a factor 3 of the most badass
desktop you can find today.


Stefan

Re: A hypervisor for a headless server?

[Stefan Monnier]

> Intel I5 cpu,

Side note: this doesn't say much more than "an amd64 CPU produced by
Intel in the last 13 years and whose release price was somewhere between
$200 and $300" :-)

See https://en.wikipedia.org/wiki/List_of_Intel_Core_i5_processors for
more details.


Stefan

Re: A hypervisor for a headless server?

[Stefan Monnier]

> Using qemu is out of discussion,because it is very slow.
> But as I said, bhyve works better than qemu alone.

Hmm... I'd expect qemu to be able to use KVM on all those machines where
Bhyve can be used.  Are you saying that you have a machine where Bhyve
works well but KVM doesn't work at all, or did I misunderstand?

[ I have no doubt that there are situations where Bhyve works better
  than Qemu (with or without KVM), that is not my question.  ]


Stefan

Re: A hypervisor for a headless server?

[Tim Woodall]

On Fri, 2 Jun 2023, Victor Sudakov wrote:



Now I see that a supported minimal headless configuration probably
does not exist at all.



Define supported.

You can boot a xen dom 0 with almost nothing installed other than xen
and the essential set and some sysvinit stuff.

I'd bet systemd would pull in more. And you'll have problems creating
networks to your guests without some iptools stuff.

If you really want a minimal system like this, start with a bootable xen
dom 0, backup, then remove stuff, reboot, backup until it won't boot.

Just looked, my dom0 is using around 1GB of disk and that's without
trying to keep it small. 300M of that is kernel modules.

My dom0 does very little other than host the guests. It's accessible
over ssh or ipmi (no keyboard or screen)

Re: A hypervisor for a headless server?

[Mario Marietto]

 - if bhyve fits your needs, why not run FreeBSD and bhyve?

I use Linux (+ qemu and kvm) and FreeBSD (with bhyve) depending what OS
between these allows me to perform a task faster and better.


- Look at Xen history, you'll see that it started in the mid 2000s.

I like Xen,I've used it for several months,but then I stopped using it
because on Linux I prefer qemu and kvm and on FreeBSD no one is interested
in maintaining Xen anymore. Everyone says that it is superated.


- And then, why not vmm, openBSD's virtual machines ?

I tried it,but I prefer bhyve. It has more functions and above all,the
passthru of my graphic card works on a Linux vm. vmm does not support it.


- But I think xen will run on some older hardware that kvm doesn't.

I think that bhyve is better than xen. So on FreeBSD I use bhyve and on
linux I use qemu+kvm,just because I have a recent hardware. On the old PC I
have installed FreeBSD and I use bhyve.

On Fri, Jun 2, 2023 at 6:54 PM Tim Woodall  wrote:

> On Fri, 2 Jun 2023, Michael Stone wrote:
>
> > On Fri, Jun 02, 2023 at 11:09:36AM +0200, Paul Leiber wrote:
> >> +1 for Xen, AFAIK the standard apt installation doesn't include any
> >> management GUI.
> >>
> >> This is the howto which helped me getting started:
> >>
> >> https://wiki.xenproject.org/wiki/Xen_Project_Beginners_Guide
> >
> > I don't recommend xen for new projects. It has more pieces and tends to
> be
> > more fragile than qemu+kvm, for no real benefits these days. (IMO)
> >
> >
>
> I'm heavily invested in xen but I'd second this. One of my projects for
> this year is to move to kvm.
>
> But I think xen will run on some older hardware that kvm doesn't
> support.
>
>
>

-- 
Mario.

Re: A hypervisor for a headless server?

[Tim Woodall]

On Fri, 2 Jun 2023, Michael Stone wrote:


On Fri, Jun 02, 2023 at 11:09:36AM +0200, Paul Leiber wrote:
+1 for Xen, AFAIK the standard apt installation doesn't include any 
management GUI.


This is the howto which helped me getting started:

https://wiki.xenproject.org/wiki/Xen_Project_Beginners_Guide


I don't recommend xen for new projects. It has more pieces and tends to be 
more fragile than qemu+kvm, for no real benefits these days. (IMO)





I'm heavily invested in xen but I'd second this. One of my projects for
this year is to move to kvm.

But I think xen will run on some older hardware that kvm doesn't
support.

Re: A hypervisor for a headless server?

[zithro]

On 02 Jun 2023 17:34, Mario Marietto wrote:

Excuse me,but there is something within your argumentation that I don't
like and I want to express what it is. Let's take Linux as an example of
what I want to say. Linux is well known to be an OS that can be installed
on the old machines,helping the people that can't buy a new computer to
surf the net and to do the basic things that they couldn't do using a more
complete and modern PC built with new hardware components. And this is a
linux quality that everyone loves and one of the reasons why Linux is
growing faster on the market.  I think that everyone likes this,right ?
So,why the same logic can't be applied to those software tools that go in
the same direction,to those tools that help the users to have those
functions that those old computers cannot give to them anymore ? Here it
seems there is a contradiction. You may argue that developing for a small
number of old computers isn't worth trying. But,first of all,I think that
there are a LOT of old PCs in the world,since poor people aren't only a
niche. We could discuss,anyway, how to reach the right amount of money to
pay the developers. I'm thinking of opening a crowdfunding campaign for
example. Or any other method to have the money that I can't imagine now. To
do a project like this is socially accepted and helps to work on the
perception of the users that computer science is something that they can
use to develop their life in a good way. Maybe by helping one of those poor
children,we are contributing to educate someone that in the future will
make great things for humanity. I think that using the old relation that
there is between costs and benefits is not applicable in every kind of
situation. There are already a lot of people who work on projects that they
like,but that they have a low social impact. Why not to work on a project
that aims to extend the functions of an old PC. And what's better than
using two operating systems on a single old pc? My old pc has 2 measly
cpu's, I used one for the host and the other for linux emulated with bhyve
and the performance was decent. Is there something that's more useful and
generous than this kind of project ?



?

Look at Xen history, you'll see that it started mid 2000s.
And then, why not vmm, openBSD's virtual machines ?

Re: A hypervisor for a headless server?

[Dan Ritter]

Mario Marietto wrote: 
> I mean. I cant use qemu on that I5 cpu because is slow without kvm. Kvm
> does not work on that cpu because it is needs some extensions from the cpu
> that there arent. Bhyve is the only alternative because it is a mix between
> qemu and kvm in terms of speed. So. My question is : how much old cpu there
> are that cant run kvm ? I dont think mine is the only one. May be a good
> idea is to port bhyve on linux to cover the little needs of the users who
> wants a fast hyp on the old cpus. and not,qemu in these cpus is very slow.
> is not the solution. I really think there isnt any better alternative than
> qemu in these situations. The only one is bhyve
>  if someone wants to try the scenarios that im talking about,they will
> understand for sure. and maybe they want to start the porting of bhyve on
> linux.


The most recent general-purpose Intel CPU without VT-X is from
2012.

The most recent general-purpose AMD CPU without SVM (AMD-V) are
some very low end Sempron APUs and the first generation of
Opteron. All the Zen chips support it, modulo weird early
motherboard issues.

*everything* on processors that old is slow. It's not going to
be on anyone's list to improve performance of unassisted QEMU or
bhyve on very old CPUs.

That said, it is certainly possible to improve performance of
QEMU on non-virtualized hardware, by passing the appropriate
flags for removing hardware that you are not using and selecting
the best-performing drivers.

In the end, the best choices are:

- don't run virtualized. Use a chroot or a container.

- acquire newer hardware. Used laptops with 8GB of RAM and an i5
CPU are going for $100 near me -- less if you buy in quantity.
It is quite likely that you can find better hardware for your
needs for free by posting an ad in your local Craigslist or
equivalent.

- if bhyve fits your needs, why not run FreeBSD and bhyve?

-dsr-

A case for supporting antiquated hardware, was Re: A hypervisor for a headless server?

[James H. H. Lampert]

On 6/2/23 8:34 AM, Mario Marietto wrote:

You may argue that developing for a small number of old computers
isn't worth trying. But,first of all,I think that there are a LOT of
old PCs in the world,since poor people aren't only a niche.


Nor are they the only ones using antiquated hardware, or expecting new 
hardware to remain in service until it physically deteriorates to the 
point of unreliability.


Some of us are Luddites, and damn proud of it. Earlier this year, I 
finished a months-long project of obtaining a notebook computer old 
enough to be viable as a DOSbook (IBM PC-DOS 2000, with no WinDoze 
whatsoever), and configuring it as such, precisely so that I would once 
again have backup hardware, and mobile capability, for my DOS 
applications. As a replacement for my dying "bionic desk lamp" iMac, I 
eschewed both WinDoze and Mac, in favor of a System76 Meerkat, precisely 
because a state-of-the-art Linux system would presumably have a nice 
long lifespan.


I don't trade in my automobiles for new models; I keep them until it's 
time to have them hauled off to their final rusting places. And I spend 
my Saturdays docenting at the International Printing Museum, where I 
frequently operate presses and linecasting equipment that is nearly as 
old, or older, than I am, some of which was already decades old before I 
was born.


Luddites of the World Unite! You have nothing to lose but your upgrade 
treadmills, and Linux and DOS are your friends!


--
JHHL

Re: A hypervisor for a headless server?

[Mario Marietto]

Excuse me,but there is something within your argumentation that I don't
like and I want to express what it is. Let's take Linux as an example of
what I want to say. Linux is well known to be an OS that can be installed
on the old machines,helping the people that can't buy a new computer to
surf the net and to do the basic things that they couldn't do using a more
complete and modern PC built with new hardware components. And this is a
linux quality that everyone loves and one of the reasons why Linux is
growing faster on the market.  I think that everyone likes this,right ?
So,why the same logic can't be applied to those software tools that go in
the same direction,to those tools that help the users to have those
functions that those old computers cannot give to them anymore ? Here it
seems there is a contradiction. You may argue that developing for a small
number of old computers isn't worth trying. But,first of all,I think that
there are a LOT of old PCs in the world,since poor people aren't only a
niche. We could discuss,anyway, how to reach the right amount of money to
pay the developers. I'm thinking of opening a crowdfunding campaign for
example. Or any other method to have the money that I can't imagine now. To
do a project like this is socially accepted and helps to work on the
perception of the users that computer science is something that they can
use to develop their life in a good way. Maybe by helping one of those poor
children,we are contributing to educate someone that in the future will
make great things for humanity. I think that using the old relation that
there is between costs and benefits is not applicable in every kind of
situation. There are already a lot of people who work on projects that they
like,but that they have a low social impact. Why not to work on a project
that aims to extend the functions of an old PC. And what's better than
using two operating systems on a single old pc? My old pc has 2 measly
cpu's, I used one for the host and the other for linux emulated with bhyve
and the performance was decent. Is there something that's more useful and
generous than this kind of project ?

Re: A hypervisor for a headless server?

[zithro]

On 02 Jun 2023 14:31, Michael Stone wrote:
I don't recommend xen for new projects. It has more pieces and tends to 
be more fragile than qemu+kvm, for no real benefits these days. (IMO)


Define "more pieces" and "more fragile" ?
It has a really low TCB and still used by amazon for their cloud.

I would recommend Xen.
It's better security-wise than KVM, is ultra stable, and easy to use.
I run Linuxes, BSDs and Windows as domUs (PCI passthrough, etc).
Plus it doesn't pull hundreds of dependencies.
You don't even need qemu if running fully virtualized guests (PV/PVH).

Re: A hypervisor for a headless server?

[Michael Stone]

On Fri, Jun 02, 2023 at 03:24:13PM +0200, Mario Marietto wrote:

I mean. I cant use qemu on that I5 cpu because is slow without kvm. Kvm does
not work on that cpu because it is needs some extensions from the cpu that
there arent. Bhyve is the only alternative because it is a mix between qemu and
kvm in terms of speed. So. My question is : how much old cpu there are that
cant run kvm ? I dont think mine is the only one. May be a good idea is to port
bhyve on linux to cover the little needs of the users who wants a fast hyp on
the old cpus. and not,qemu in these cpus is very slow. is not the solution. I
really think there isnt any better alternative than qemu in these situations.
The only one is bhyve
 if someone wants to try the scenarios that im talking about,they will
understand for sure. and maybe they want to start the porting of bhyve on
linux.


Realistic answer: if something can't be supported by kvm it's probably 
old enough that upgrading it makes more sense than investing developer 
resources on that niche case. 10 year old machines that do support kvm 
are basically free these days.


Other than that, I'm not going to argue the basically hypothetical case 
of "machines that can use bhyve hardware virtualization but not kvm 
hardware virtualization" because 1) there probably aren't many and 2) 
without details of what exactly is the problem with your particular 
machine I can't provide a sensible response. (There's not really much 
difference between kvm and bhyve requirements, so I'd guess something 
like a bios bug is causing issues.) Again, this just isn't a general 
problem and certainly not something worth a lot of resource expenditure.

Re: A hypervisor for a headless server?

[Mario Marietto]

ok. Thank you very much for the explanations.

On Fri, Jun 2, 2023 at 3:15 PM Michael Stone  wrote:

> On Fri, Jun 02, 2023 at 03:01:04PM +0200, Mario Marietto wrote:
> >Using qemu is out of discussion,because it is very slow. But as I
> said,bhyve
> >works better than qemu alone.
>
> kvm literally uses qemu as its user space, so it's very much not out of
> the discussion. If you can't use the kvm kernel extensions for
> virtualization, then running under qemu gets you the same userspace
> experience with a performance penalty. bhyve has its own requirements
> for what cpu extensions must be present, and AFAIK can't fall back to a
> non-accelerated mode if they are not, regardless of performance.
> Sometimes you might want VMs to run, even more slowly, than to not run
> at all.
>
> I get that there's some machine that you have that can't run kvm. I have
> no idea why without access to the machine. But that's not a general
> problem, that's a problem with some specific piece of hardware. It's not
> like kvm has exotic requirements--I'm running on some hardware that's
> more than a decade old.
>


-- 
Mario.

Re: A hypervisor for a headless server?

[Dan Ritter]

Victor Sudakov wrote: 
> I'm currently going to migrate some FreeBSD VMs from bhyve to a linux
> host. I hope KVM will have no problem with their raw disk images. 


Most raw images are supported; if there's something odd going
on, `qemu-img convert` supports:

blkdebug blklogwrites blkverify bochs cloop compress copy-on-read dmg
file ftp ftps gluster host_cdrom host_device http https iscsi iser luks
nbd nfs null-aio null-co nvme qcow qcow2 qed quorum raw rbd replication
ssh throttle vdi vhdx vmdk vpc vvfat

-dsr-

Re: A hypervisor for a headless server?

[Mario Marietto]

I mean. I cant use qemu on that I5 cpu because is slow without kvm. Kvm
does not work on that cpu because it is needs some extensions from the cpu
that there arent. Bhyve is the only alternative because it is a mix between
qemu and kvm in terms of speed. So. My question is : how much old cpu there
are that cant run kvm ? I dont think mine is the only one. May be a good
idea is to port bhyve on linux to cover the little needs of the users who
wants a fast hyp on the old cpus. and not,qemu in these cpus is very slow.
is not the solution. I really think there isnt any better alternative than
qemu in these situations. The only one is bhyve
 if someone wants to try the scenarios that im talking about,they will
understand for sure. and maybe they want to start the porting of bhyve on
linux.

Il ven 2 giu 2023, 15:01 Michael Stone  ha scritto:

> On Fri, Jun 02, 2023 at 08:41:44AM +, Victor Sudakov wrote:
> >Interestingly, libvirt claims to support bhyve, I just never felt a
> >need for such sophisticated tools to run just several VMs.
>
> Yes, it sounds like you should just ignore libvirt entirely and just
> install qemu-system-x86 (and not qemu-system-gui). That's a minimal
> system with no gui, you just run qemu from the command line to start
> VMs. If you run with --enable-kvm or --machine accel=kvm, then you're
> using kvm (assuming the kernel module is loaded).
>
> That said, there's a huge convenience factor for libvirt. You may end up
> with libraries you'll never use on the server, but so what? You can
> install virt-manager on a client system and manage with a gui that uses
> ssh in the background, or use virsh on the server. If you find yourself
> needing to do something infrequently, it's much easier to discover it in
> the virt-manager gui than it is to dig through docs on how to do it from
> the qemu command line. (This is, of course, the usual tradeoff between
> text and graphical interfaces.) It's also easier to use
> standard/documented solutions for startup, config, storage, etc, than it
> is to remember what bespoke solution you came up with several years ago
> when something breaks, even if all the abstraction layers of libvirt are
> "less efficient".
>
>

Re: A hypervisor for a headless server?

[Michael Stone]

On Fri, Jun 02, 2023 at 03:01:04PM +0200, Mario Marietto wrote:

Using qemu is out of discussion,because it is very slow. But as I said,bhyve
works better than qemu alone.


kvm literally uses qemu as its user space, so it's very much not out of 
the discussion. If you can't use the kvm kernel extensions for 
virtualization, then running under qemu gets you the same userspace 
experience with a performance penalty. bhyve has its own requirements 
for what cpu extensions must be present, and AFAIK can't fall back to a 
non-accelerated mode if they are not, regardless of performance. 
Sometimes you might want VMs to run, even more slowly, than to not run 
at all.


I get that there's some machine that you have that can't run kvm. I have 
no idea why without access to the machine. But that's not a general 
problem, that's a problem with some specific piece of hardware. It's not 
like kvm has exotic requirements--I'm running on some hardware that's

more than a decade old.

Re: A hypervisor for a headless server?

[Mario Marietto]

Using qemu is out of discussion,because it is very slow. But as I
said,bhyve works better than qemu alone.

On Fri, Jun 2, 2023 at 2:44 PM Michael Stone  wrote:

> On Fri, Jun 02, 2023 at 11:21:45AM +0200, Mario Marietto wrote:
> >wait wait. for sure the option should be enabled on the bios,but bhyve
> works in
> >a different way than kvm,so it works even if my cpu does not have all the
> virt.
> >parameters respected. Infact kvm does not work on that cpu. But how many
> cpus
> >there are like mine ? Does Linux feel to cover the gap of an alternative
> to
> >qemu and kvm ? not sure about xen as an alternative.
>
> kvm is literally the hardware acceleration piece. If your CPU doesn't
> support that, use qemu without kvm and you get exactly the same
> experience (just a bit slower). So "linux" "feels" no need to cover the
> gap, because there isn't one.
>
>

-- 
Mario.

Re: A hypervisor for a headless server?

[Michael Stone]

On Fri, Jun 02, 2023 at 08:41:44AM +, Victor Sudakov wrote:

Interestingly, libvirt claims to support bhyve, I just never felt a
need for such sophisticated tools to run just several VMs.


Yes, it sounds like you should just ignore libvirt entirely and just 
install qemu-system-x86 (and not qemu-system-gui). That's a minimal 
system with no gui, you just run qemu from the command line to start 
VMs. If you run with --enable-kvm or --machine accel=kvm, then you're 
using kvm (assuming the kernel module is loaded).


That said, there's a huge convenience factor for libvirt. You may end up 
with libraries you'll never use on the server, but so what? You can 
install virt-manager on a client system and manage with a gui that uses 
ssh in the background, or use virsh on the server. If you find yourself 
needing to do something infrequently, it's much easier to discover it in 
the virt-manager gui than it is to dig through docs on how to do it from 
the qemu command line. (This is, of course, the usual tradeoff between 
text and graphical interfaces.) It's also easier to use 
standard/documented solutions for startup, config, storage, etc, than it 
is to remember what bespoke solution you came up with several years ago 
when something breaks, even if all the abstraction layers of libvirt are 
"less efficient".

Re: A hypervisor for a headless server?

[Charles Curley]

On Fri, 2 Jun 2023 09:12:02 +
Victor Sudakov  wrote:

> I'm currently going to migrate some FreeBSD VMs from bhyve to a linux
> host. I hope KVM will have no problem with their raw disk images. 

You might look into the Debian vagrant-mutate package.

apt show vagrant-mutate

-- 
Does anybody read signatures any more?

https://charlescurley.com
https://charlescurley.com/blog/

Re: A hypervisor for a headless server?

[Michael Stone]

On Fri, Jun 02, 2023 at 11:21:45AM +0200, Mario Marietto wrote:

wait wait. for sure the option should be enabled on the bios,but bhyve works in
a different way than kvm,so it works even if my cpu does not have all the virt.
parameters respected. Infact kvm does not work on that cpu. But how many cpus
there are like mine ? Does Linux feel to cover the gap of an alternative to
qemu and kvm ? not sure about xen as an alternative. 


kvm is literally the hardware acceleration piece. If your CPU doesn't 
support that, use qemu without kvm and you get exactly the same 
experience (just a bit slower). So "linux" "feels" no need to cover the 
gap, because there isn't one.

Re: A hypervisor for a headless server?

[Michael Stone]

On Fri, Jun 02, 2023 at 11:09:36AM +0200, Paul Leiber wrote:
+1 for Xen, AFAIK the standard apt installation doesn't include any 
management GUI.


This is the howto which helped me getting started:

https://wiki.xenproject.org/wiki/Xen_Project_Beginners_Guide


I don't recommend xen for new projects. It has more pieces and tends to 
be more fragile than qemu+kvm, for no real benefits these days. (IMO)

Re: A hypervisor for a headless server?

[Michael Stone]

On Thu, Jun 01, 2023 at 11:53:26PM -0400, Brian Sammon wrote:

"virt-manager", on the other hand, appears to be fundamentally a GUI tool.


But virsh from libvirt-clients isn't.

Re: A hypervisor for a headless server?

[Mario Marietto]

wait wait. for sure the option should be enabled on the bios,but bhyve
works in a different way than kvm,so it works even if my cpu does not have
all the virt. parameters respected. Infact kvm does not work on that cpu.
But how many cpus there are like mine ? Does Linux feel to cover the gap of
an alternative to qemu and kvm ? not sure about xen as an alternative.

Il ven 2 giu 2023, 11:12 Victor Sudakov  ha scritto:

> Andy Smith wrote:
> > Hi Victor,
>
> Hi Andy!
>
> [dd]
> > > Now I see that a supported minimal headless configuration probably
> > > does not exist at all.
> >
> > I don't think that is correct at all, depending on what you mean by
> > "supported". You absolutely will find a guide out there to do what
> > you want, with relative ease I should think.
>
> Yes, I guess the https://wiki.debian.org/KVM seems a good guide and
> even covers the case of a minimal :-) installation.
>
> [dd]
> >
> > I would say that documentation from Ubuntu is likely to be more
> > "enterprisey". The other thing is, if you're coming from a BSD
> > background (you mentioned Bhyve) you probably are a lot more used to
> > there being one way of doing things and that way being thoroughly
> > documented.
>
> That's correct. Though I must admit the FreeBSD Handbook can be
> outdated in places as the project is clearly lacking resources. It is
> still a very good source of knowledge.
>
> > Whereas on Linux there tends to be multiple ways and
> > even the same one can be slightly different on different Linux
> > distributions.
>
> Some Debian documentation is very good too.
> >
> > I am using Xen more at the moment, but I generally wouldn't
> > recommend that to newcomers. I tend to recommend KVM just because
> > there's so many guides for it out there.
>
> I'm currently going to migrate some FreeBSD VMs from bhyve to a linux
> host. I hope KVM will have no problem with their raw disk images.
>
> [dd]
> >
> > I would probably just install qemu-kvm and accept the bloat of a lot
> > of packages that I would never use, use virsh to manage the VMs from
> > command line, and perhaps over time worm out which packages can be
> > safely removed.
>
> OK, thank you, maybe I'll go this route.
>
> --
> Victor Sudakov VAS4-RIPE
> http://vas.tomsk.ru/
> 2:5005/49@fidonet
>

Re: A hypervisor for a headless server?

[Victor Sudakov]

Andy Smith wrote:
> Hi Victor,

Hi Andy!

[dd]
> > Now I see that a supported minimal headless configuration probably
> > does not exist at all.
> 
> I don't think that is correct at all, depending on what you mean by
> "supported". You absolutely will find a guide out there to do what
> you want, with relative ease I should think.

Yes, I guess the https://wiki.debian.org/KVM seems a good guide and
even covers the case of a minimal :-) installation.

[dd]
> 
> I would say that documentation from Ubuntu is likely to be more
> "enterprisey". The other thing is, if you're coming from a BSD
> background (you mentioned Bhyve) you probably are a lot more used to
> there being one way of doing things and that way being thoroughly
> documented. 

That's correct. Though I must admit the FreeBSD Handbook can be
outdated in places as the project is clearly lacking resources. It is
still a very good source of knowledge.

> Whereas on Linux there tends to be multiple ways and
> even the same one can be slightly different on different Linux
> distributions.

Some Debian documentation is very good too. 
> 
> I am using Xen more at the moment, but I generally wouldn't
> recommend that to newcomers. I tend to recommend KVM just because
> there's so many guides for it out there.

I'm currently going to migrate some FreeBSD VMs from bhyve to a linux
host. I hope KVM will have no problem with their raw disk images. 

[dd]
> 
> I would probably just install qemu-kvm and accept the bloat of a lot
> of packages that I would never use, use virsh to manage the VMs from
> command line, and perhaps over time worm out which packages can be
> safely removed.

OK, thank you, maybe I'll go this route.

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Paul Leiber]

Am 02.06.2023 um 09:28 schrieb Victor Sudakov:

Miles Fidelman wrote:



On Thu, Jun 1, 2023, 9:58 PM Victor Sudakov mailto:v...@sibptus.ru>> wrote:

 Dear Colleagues,

 There is a hypervisor called bhyve for FreeBSD. It's completely
 headless, no graphics, runs as a daemon and provides serial and VNC
 consoles.

 Can you please advise a similar headless and minimal hypervisor for
 Debian or Ubuntu?



The classic would be Xen (which I've been running for years). There's
also Virtual Box, and VMware ESXi.  You might check out the list at
https://www.hitechnectar.com/blogs/open-source-hypervisor/


Thanks for reminding about Xen, it looks minimalist enough, I am
going to have a look at it.



+1 for Xen, AFAIK the standard apt installation doesn't include any 
management GUI.


This is the howto which helped me getting started:

https://wiki.xenproject.org/wiki/Xen_Project_Beginners_Guide

(Disclaimer: I don't have experience with any other virtualization 
package than Xen.)

Re: A hypervisor for a headless server?

[Victor Sudakov]

Nicolas George wrote:
> Victor Sudakov (12023-06-02):
> > Oh, I'm a rare kind of newbie. I have 25 years of FreeBSD
> > experience and about 10 years of Solaris experience.
> 
> Newbies who think they have a lot of experience are, sadly, not a rare
> breed.

I agree. But once you start measuring experience by the actual years a
person has actively worked with this or that system, you can judge
more or less objectively.

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Victor Sudakov]

Mario Marietto wrote:
> Hello to everyone. I follow every day the development of bhyve for FreeBSD
> and I have even collaborated with some of its developers to add the
> functionality of the passing through of one nvidia gpu to a linux guest.
> What to say ? that bhyve is a programming gem. Qemu and kvm have more
> functionalities but they are even old. Bhyve is a fresh product that is
> evolving fast. Qemu + kvm for example don't work on my old PC that has an
> Intel I5 cpu,because it does not have all the virtualization requirements.
> For the sake of my curiosity I tried bhyve and...it worked. I don't know
> why,but I know that it requires less virtualization directives. Some
> developers talked about the idea to rewrite it to make it a standalone tool
> and I think that's a nice idea. As I think that a cool idea could be to
> rewrite its code to port it to Linux. It could be used as a light
> hypervisor,for those old machines like mine,that don't have all the
> hardware prerogatives needed to run qemu and kvm.

Still, from my experience, if you disable hardware virtualization in
BIOS Setup, bhyve does not work.

I agree with you that bhyve is a masterpiece. The way it works with
ZFS is brilliant too. I usually use the vm-bhyve shell with it, but it
seems to be supported by some hypervisor management tools like
libvirt.

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[tomas]

On Fri, Jun 02, 2023 at 08:36:51AM +, Victor Sudakov wrote:

[...]

> Oh, I'm a rare kind of newbie. I have 25 years of FreeBSD
> experience and about 10 years of Solaris experience. However I still
> consider myself a newbie in Linux as I work with it only since 2020
> and in rather limited ways.

:-)

This was my impression, too. But actually, that's what
happens with most of us. I'd be a FreeBSD newbie and
a Solaris newbie too (unless some HP/UX experience way
back then is worth anything).

Glad to see you are finding your path. Mind the bumps
and enjoy the strange landscape :)

Cheers
-- 
t


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[tomas]

On Fri, Jun 02, 2023 at 10:41:42AM +0200, Nicolas George wrote:
> Victor Sudakov (12023-06-02):
> > Oh, I'm a rare kind of newbie. I have 25 years of FreeBSD
> > experience and about 10 years of Solaris experience.
> 
> Newbies who think they have a lot of experience are, sadly, not a rare
> breed.

"Newbie" is a relative term in a highly multidimensional space.
Expect conflicts of perspective!

I wouldn't say "sadly". That's what makes life interesting.

Cheers
-- 
t


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Victor Sudakov]

to...@tuxteam.de wrote:
> On Fri, Jun 02, 2023 at 08:05:18AM +, Andy Smith wrote:
> > Hello,
> 
> [...]
> 
> > Most of the time with most packages it's obvious, but I have seen
> > some weird things from time to time! KVM is such a big package that
> > I shy away from just advising --no-install-recommends to those
> > inexperienced with it.
> 
> 100% agreed. Whoever deviates from the "recommended" way should be
> prepared (and willing) to learn a few things on the way. Which may
> be a good thing or not :)


Still the idea to have a base hypervisor package: fully tested,
functional and not missing any important files, and a set of optional
GUI tools/shells is very appealing to me. Much like you can install,
for example, Git (CLI tested and ready) and a bunch of optional GUI
tools or editors around it.


-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Victor Sudakov]

Andy Smith wrote:

[dd]

> 
> Most of the time with most packages it's obvious, but I have seen
> some weird things from time to time! KVM is such a big package that
> I shy away from just advising --no-install-recommends to those
> inexperienced with it.

Thanks for your opinion. I've made a mental note to myself to treat
KVM differently from bhyve. The latter is really small and has a very
small footprint on the system. 

Interestingly, libvirt claims to support bhyve, I just never felt a
need for such sophisticated tools to run just several VMs.

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Nicolas George]

Victor Sudakov (12023-06-02):
> Oh, I'm a rare kind of newbie. I have 25 years of FreeBSD
> experience and about 10 years of Solaris experience.

Newbies who think they have a lot of experience are, sadly, not a rare
breed.

Goodbye.

-- 
  Nicolas George

Re: A hypervisor for a headless server?

[Mario Marietto]

Hello to everyone. I follow every day the development of bhyve for FreeBSD
and I have even collaborated with some of its developers to add the
functionality of the passing through of one nvidia gpu to a linux guest.
What to say ? that bhyve is a programming gem. Qemu and kvm have more
functionalities but they are even old. Bhyve is a fresh product that is
evolving fast. Qemu + kvm for example don't work on my old PC that has an
Intel I5 cpu,because it does not have all the virtualization requirements.
For the sake of my curiosity I tried bhyve and...it worked. I don't know
why,but I know that it requires less virtualization directives. Some
developers talked about the idea to rewrite it to make it a standalone tool
and I think that's a nice idea. As I think that a cool idea could be to
rewrite its code to port it to Linux. It could be used as a light
hypervisor,for those old machines like mine,that don't have all the
hardware prerogatives needed to run qemu and kvm.

On Fri, Jun 2, 2023 at 10:10 AM  wrote:

> On Fri, Jun 02, 2023 at 08:05:18AM +, Andy Smith wrote:
> > Hello,
>
> [...]
>
> > Most of the time with most packages it's obvious, but I have seen
> > some weird things from time to time! KVM is such a big package that
> > I shy away from just advising --no-install-recommends to those
> > inexperienced with it.
>
> 100% agreed. Whoever deviates from the "recommended" way should be
> prepared (and willing) to learn a few things on the way. Which may
> be a good thing or not :)
>
> Cheers
> --
> t
>


-- 
Mario.

Re: A hypervisor for a headless server?

[Victor Sudakov]

to...@tuxteam.de wrote:
> > 
> > But Andy Smith wrote recently that "installing without recommends is not a 
> > supported use
> > case" and I believe him.
> 
> FWIW, I do install with no-recommends in general:
> 
>   tomas@trotzki:~$ cat /etc/apt/apt.conf.d/95no-recommends 
>   APT::Install-Recommends no;
> 
> "Not supported" seemed a bit strong to me: what does mean "not
> supported" in a non-commercial distro, anyway?
> 
> I'd prefer to say: not recommended for newbies. The problem with
> you is that you want to be a newbie and not a newbie at the same
> time.
> 
> This doesn't work :)

Oh, I'm a rare kind of newbie. I have 25 years of FreeBSD
experience and about 10 years of Solaris experience. However I still
consider myself a newbie in Linux as I work with it only since 2020
and in rather limited ways.

> 
> In the concrete case: some newbie would expect to have the GUI
> tools installed when (s)he installs KVM. Some other not. Since
> APT doesn't (yet) support mind reading (alas, the proprietary
> drivers and that), the GUI tools are recommended.

That's what documentation is for! If a newbie wants an installation
without GUI, they look for a how-to about headless installation.

It's a pity I came across https://wiki.debian.org/KVM only after I had
already posted to the list. I think this page covers my case. I really
with that the first reply to my question had been a link to this page, it
could have saved a lot of electrons and carbon.

> 
> That's what "recommends" is for. You can switch it off, even
> in general (see above), but then you'll have to be prepared to
> look into package descriptions and come up yourself with "oh,
> perhaps I want to install that, too".

Yes, sure.

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[tomas]

On Fri, Jun 02, 2023 at 08:05:18AM +, Andy Smith wrote:
> Hello,

[...]

> Most of the time with most packages it's obvious, but I have seen
> some weird things from time to time! KVM is such a big package that
> I shy away from just advising --no-install-recommends to those
> inexperienced with it.

100% agreed. Whoever deviates from the "recommended" way should be
prepared (and willing) to learn a few things on the way. Which may
be a good thing or not :)

Cheers
-- 
t


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Andy Smith]

Hello,

On Fri, Jun 02, 2023 at 09:49:39AM +0200, to...@tuxteam.de wrote:
> FWIW, I do install with no-recommends in general:
> 
>   tomas@trotzki:~$ cat /etc/apt/apt.conf.d/95no-recommends 
>   APT::Install-Recommends no;
> 
> "Not supported" seemed a bit strong to me: what does mean "not
> supported" in a non-commercial distro, anyway?

It is a philosophical question but I think that with every package
there is a social contract regarding what behaviour can be expected
from the package.

The packages are put together with an expected set of recommends.

If you do not install all of those recommends then you can obviously
expect some features of the package to not be present. If the
upstream project has done things well, error messages will be very
clear when you try to use a feature that is missing a dependency.

However, in the real world you will sometimes experience confusing
behaviour where it's not clear if that is expected or what — if
anything — is missing. At best, improving that once identified is
going to be a minor or wish list bug priority and it's work that the
Debian maintainer can't really be expected to investigate or carry
patches for. They already set the recommends.

So, when I say "not supported", I mean that I think it is beyond
expectations that the package should still work exactly as
documented if you do not install recommends, and should only be done
by experienced users who are prepared to rule out missing recommends
any time they see strange behaviour.

> I'd prefer to say: not recommended for newbies.

Agreed.

> you'll have to be prepared to look into package descriptions and
> come up yourself with "oh, perhaps I want to install that, too".
> 
> It's not hard. 

Most of the time with most packages it's obvious, but I have seen
some weird things from time to time! KVM is such a big package that
I shy away from just advising --no-install-recommends to those
inexperienced with it.

Thanks,
Andy

-- 
https://bitfolk.com/ -- No-nonsense VPS hosting

Re: A hypervisor for a headless server?

[Stanislav Vlasov]

пт, 2 июн. 2023 г. в 12:38, Victor Sudakov :

> > > Running "apt install qemu-kvm" on a Debian 11 AWS EC2 instance which has
> > > never had any X-Window or desktop environment in its entire life,
> > > tries to install qemu-system-gui, adwaita-icon-theme, libgtk-3-common
> > > and a lot of similar stuff.
> >
> > Even with `--no-install-recommends`?
> > qemu-system-x86 package _recommends_ qemu-system-gui, but you may
> > install without.
>
> But Andy Smith wrote recently that "installing without recommends is not a 
> supported use
> case" and I believe him.

It may be unsupported, but it work for me and my servers and I did not
break any system with it.
I don't use libvirt/virtmanager (hosting control panel was written
before virtmanager in debian), but qemu/kvm work fine without gui
wrappers with console on vnc.

> On the other hand, https://wiki.debian.org/KVM#Installation advises it:
>
> "When installing on a server, you can add the --no-install-recommends
> apt option, to prevent the installation of extraneous graphical
> packages"
>
> The truth is definitly out there somewhere.

I think Debian's wiki and Debian's official manuals
(https://www.debian.org/doc/user-manuals) are enough in this case.
In the installation guide you can see how to install without recommends.

-- 
Stanislav

Re: A hypervisor for a headless server?

[Andy Smith]

Hi Victor,

On Fri, Jun 02, 2023 at 07:38:15AM +, Victor Sudakov wrote:
> Stanislav Vlasov wrote:
> > пт, 2 июн. 2023 г. в 12:18, Victor Sudakov :
> > 
> > > Running "apt install qemu-kvm" on a Debian 11 AWS EC2 instance which has
> > > never had any X-Window or desktop environment in its entire life,
> > > tries to install qemu-system-gui, adwaita-icon-theme, libgtk-3-common
> > > and a lot of similar stuff.
> > 
> > Even with `--no-install-recommends`?
> > qemu-system-x86 package _recommends_ qemu-system-gui, but you may
> > install without.
> 
> But Andy Smith wrote recently that "installing without recommends is not a 
> supported use
> case" and I believe him.

I should be clear that when I said, "installing without recommends
is not a supported use case" I was speaking more generally about
packages on Debian, not specifically about the state of KVM on
Debian.

The thing about recommends is that they're maintainer's opinion
about what other software is likely to be needed for broad and
general use of the package. If you are in the habit of not
installing them, occasionally you can find that some feature doesn't
work properly, and it might or might not be obvious why that is. It
can call for some familiarity with the packages involved. What is
considered obvious can vary between maintainer and user.

So again, I do not like to straight off recommend to people that
they go for --no-install-recommends especially when it's for
software that is new to them.

Thanks,
Andy

-- 
https://bitfolk.com/ -- No-nonsense VPS hosting

Re: A hypervisor for a headless server?

[tomas]

On Fri, Jun 02, 2023 at 07:38:15AM +, Victor Sudakov wrote:
> Stanislav Vlasov wrote:
> > пт, 2 июн. 2023 г. в 12:18, Victor Sudakov :
> > 
> > > Running "apt install qemu-kvm" on a Debian 11 AWS EC2 instance which has
> > > never had any X-Window or desktop environment in its entire life,
> > > tries to install qemu-system-gui, adwaita-icon-theme, libgtk-3-common
> > > and a lot of similar stuff.
> > 
> > Even with `--no-install-recommends`?
> > qemu-system-x86 package _recommends_ qemu-system-gui, but you may
> > install without.
> 
> But Andy Smith wrote recently that "installing without recommends is not a 
> supported use
> case" and I believe him.

FWIW, I do install with no-recommends in general:

  tomas@trotzki:~$ cat /etc/apt/apt.conf.d/95no-recommends 
  APT::Install-Recommends no;

"Not supported" seemed a bit strong to me: what does mean "not
supported" in a non-commercial distro, anyway?

I'd prefer to say: not recommended for newbies. The problem with
you is that you want to be a newbie and not a newbie at the same
time.

This doesn't work :)

In the concrete case: some newbie would expect to have the GUI
tools installed when (s)he installs KVM. Some other not. Since
APT doesn't (yet) support mind reading (alas, the proprietary
drivers and that), the GUI tools are recommended.

That's what "recommends" is for. You can switch it off, even
in general (see above), but then you'll have to be prepared to
look into package descriptions and come up yourself with "oh,
perhaps I want to install that, too".

It's not hard. 

Cheers
-- 
t


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Andy Smith]

Hi Victor,

On Fri, Jun 02, 2023 at 07:12:14AM +, Victor Sudakov wrote:
> > On Fri, Jun 02, 2023 at 03:33:09AM +, Victor Sudakov wrote:
> > > So what's the package name for just "kvm" without the GUI tools? Because
> > 
> > For someone that wants to run a hypervisor in a non-newbie manner
> 
> I don't think running a single daemon with a couple of configuration
> files and a simple CLI interface could be considered "running in a
> non-newbie manner".

Well the thing is, it's a complex piece of software and so most of
the guides out there will direct you to install and use helper tools
like virt-manager.

> > I'm afraid you don't seem to be that willing to do any of your own
> > research.
> 
> That's correct, I was not willing to do much of my own research
> of what I think should be a trivial documented best practice.

Okay, well then it seems we disagree on expectations there.

> Now I see that a supported minimal headless configuration probably
> does not exist at all.

I don't think that is correct at all, depending on what you mean by
"supported". You absolutely will find a guide out there to do what
you want, with relative ease I should think.

I just don't know of one that focuses only on the command line
aspect, though I am sure there are some out there.

It definitely is a supported use case however.

> Most articles like this https://ubuntu.com/blog/kvm-hyphervisor or
> this https://help.ubuntu.com/community/KVM/Installation if followed
> bring about lots of GUI packages. Interestingly, the five or six
> articles I've read seem to suggest a slightly different set of packages
> to set up a hypervisor.

I would say that documentation from Ubuntu is likely to be more
"enterprisey". The other thing is, if you're coming from a BSD
background (you mentioned Bhyve) you probably are a lot more used to
there being one way of doing things and that way being thoroughly
documented. Whereas on Linux there tends to be multiple ways and
even the same one can be slightly different on different Linux
distributions.

> Of course not. I'm neither qualified nor willing to hack into the KVM
> hypervisor implementation details. In fact, I've come here for advice
> how to install a minimal headless hypervisor hoping for some docs
> documenting the best practice.

Someone may pop up to give you a good KVM and virsh resource. I
think that's your best route. I hesitate to try it myself because I
haven't touched KVM much recently and my advice would be rusty.

I am using Xen more at the moment, but I generally wouldn't
recommend that to newcomers. I tend to recommend KVM just because
there's so many guides for it out there.

> Is this https://help.ubuntu.com/community/KVM/Installation the
> official guide?

No, because that is a document from Ubuntu about software written by
Red Hat (KVM did not start off as a Red Hat project but has been one
for years now). If you're looking for "official" it should probably
come from Red Hat, but this is a Debian mailing list…

> > If the idea of the installation of binary package dependencies that
> > you never use massively offends you I would suggest that Debian is
> > not an ideal match for you, and you may be better off going with
> > Gentoo or Arch or something.
> 
> I must admit it does offend me a bit but not to the degree of using
> Gentoo or Manjaro.

I would probably just install qemu-kvm and accept the bloat of a lot
of packages that I would never use, use virsh to manage the VMs from
command line, and perhaps over time worm out which packages can be
safely removed.

The Debian wiki page for KVM does cover use of "virsh" from the
command line, as well as GUI things like virt-manager.

https://wiki.debian.org/KVM

Thanks,
Andy

-- 
https://bitfolk.com/ -- No-nonsense VPS hosting

Re: A hypervisor for a headless server?

[Victor Sudakov]

Stanislav Vlasov wrote:
> пт, 2 июн. 2023 г. в 12:18, Victor Sudakov :
> 
> > Running "apt install qemu-kvm" on a Debian 11 AWS EC2 instance which has
> > never had any X-Window or desktop environment in its entire life,
> > tries to install qemu-system-gui, adwaita-icon-theme, libgtk-3-common
> > and a lot of similar stuff.
> 
> Even with `--no-install-recommends`?
> qemu-system-x86 package _recommends_ qemu-system-gui, but you may
> install without.

But Andy Smith wrote recently that "installing without recommends is not a 
supported use
case" and I believe him.

On the other hand, https://wiki.debian.org/KVM#Installation advises it:

"When installing on a server, you can add the --no-install-recommends
apt option, to prevent the installation of extraneous graphical
packages"

The truth is definitly out there somewhere.

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Stanislav Vlasov]

пт, 2 июн. 2023 г. в 12:18, Victor Sudakov :

> Running "apt install qemu-kvm" on a Debian 11 AWS EC2 instance which has
> never had any X-Window or desktop environment in its entire life,
> tries to install qemu-system-gui, adwaita-icon-theme, libgtk-3-common
> and a lot of similar stuff.

Even with `--no-install-recommends`?
qemu-system-x86 package _recommends_ qemu-system-gui, but you may
install without.

-- 
Stanislav

Re: A hypervisor for a headless server?

[Victor Sudakov]

Miles Fidelman wrote:
> >
> >
> > On Thu, Jun 1, 2023, 9:58 PM Victor Sudakov  > > wrote:
> >
> > Dear Colleagues,
> >
> > There is a hypervisor called bhyve for FreeBSD. It's completely
> > headless, no graphics, runs as a daemon and provides serial and VNC
> > consoles.
> >
> > Can you please advise a similar headless and minimal hypervisor for
> > Debian or Ubuntu?
> >
> >
> The classic would be Xen (which I've been running for years). There's 
> also Virtual Box, and VMware ESXi.  You might check out the list at 
> https://www.hitechnectar.com/blogs/open-source-hypervisor/

Thanks for reminding about Xen, it looks minimalist enough, I am
going to have a look at it.

VMware ESXi is a bare-metal hypervisor, not exactly what I was looking
for now (though when I had some experience with it, it had some kind
of Linux inside).


-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Victor Sudakov]

Nicholas Geovanis wrote:
> On Thu, Jun 1, 2023, 9:58 PM Victor Sudakov  wrote:
> 
> > Dear Colleagues,
> >
> > There is a hypervisor called bhyve for FreeBSD. It's completely
> > headless, no graphics, runs as a daemon and provides serial and VNC
> > consoles.
> >
> > Can you please advise a similar headless and minimal hypervisor for
> > Debian or Ubuntu?
> >
> 
> Just don't install x-windows or anything that depends on it, like a desktop
> environment. Servers in datacenters run headless more than 95% of cases.
> Debian and its derivatives too.

Running "apt install qemu-kvm" on a Debian 11 AWS EC2 instance which has
never had any X-Window or desktop environment in its entire life,
tries to install qemu-system-gui, adwaita-icon-theme, libgtk-3-common
and a lot of similar stuff.

That's the pain :-)


-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Victor Sudakov]

Andy Smith wrote:
> 
> On Fri, Jun 02, 2023 at 03:33:09AM +, Victor Sudakov wrote:
> > So what's the package name for just "kvm" without the GUI tools? Because
> 
> For someone that wants to run a hypervisor in a non-newbie manner

I don't think running a single daemon with a couple of configuration
files and a simple CLI interface could be considered "running in a
non-newbie manner".

> I'm afraid you don't seem to be that willing to do any of your own
> research.

That's correct, I was not willing to do much of my own research
of what I think should be a trivial documented best practice. 

Now I see that a supported minimal headless configuration probably
does not exist at all.

Most articles like this https://ubuntu.com/blog/kvm-hyphervisor or
this https://help.ubuntu.com/community/KVM/Installation if followed
bring about lots of GUI packages. Interestingly, the five or six
articles I've read seem to suggest a slightly different set of packages
to set up a hypervisor.

> 
> The package is still qemu-kvm (a virtual package that probably
> depends on qemu-system-x86 for you).
> 
> By default that will install a ton of recommended packages that
> include all the GUI tools you probably object to, but that is just
> how Debian works [and necessarily how a general purpose binary Linux
> distribution has to work].
> 
> If you want the minimum amount of packages to be installed you can
> try installing without recommends, i.e.
> 
> # apt --no-install-recommends install qemu-kvm
> 
> which will dramatically reduce the number of packages, BUT:
> 
> - you may miss some package that's really useful in most cases -
>   installing without recommends is not a supported use case.
> 
> - you'll probably still get some packages you will never use.

I tried installing with --no-install-recommends and had the same
impression as you (that I would miss some required packages and still
get some extraneous ones). That is why I really did not want to go
that way.

> 
> This is all very basic Debian systems administration and so if you
> weren't aware of these things I question whether you would have an
> easy time trying to strip a Debian install of KVM down to the bare
> minimum instead of going the easier route of just doing what there
> are thousands of guides out there for.

Of course not. I'm neither qualified nor willing to hack into the KVM
hypervisor implementation details. In fact, I've come here for advice
how to install a minimal headless hypervisor hoping for some docs
documenting the best practice.

Now I see that no such thing as a minimal hypervisor install exists in
Debian/Ubuntu, I may follow some official guide. 

Is this https://help.ubuntu.com/community/KVM/Installation the
official guide?

> 
> If the idea of the installation of binary package dependencies that
> you never use massively offends you I would suggest that Debian is
> not an ideal match for you, and you may be better off going with
> Gentoo or Arch or something.

I must admit it does offend me a bit but not to the degree of using
Gentoo or Manjaro.

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Miles Fidelman]

Nicholas Geovanis wrote:



On Thu, Jun 1, 2023, 9:58 PM Victor Sudakov > wrote:


Dear Colleagues,

There is a hypervisor called bhyve for FreeBSD. It's completely
headless, no graphics, runs as a daemon and provides serial and VNC
consoles.

Can you please advise a similar headless and minimal hypervisor for
Debian or Ubuntu?


The classic would be Xen (which I've been running for years). There's 
also Virtual Box, and VMware ESXi.  You might check out the list at 
https://www.hitechnectar.com/blogs/open-source-hypervisor/


Miles Fidelman

--
In theory, there is no difference between theory and practice.
In practice, there is.   Yogi Berra

Theory is when you know everything but nothing works.
Practice is when everything works but no one knows why.
In our lab, theory and practice are combined:
nothing works and no one knows why.  ... unknown

Re: A hypervisor for a headless server?

[john doe]

On 6/2/23 04:39, Victor Sudakov wrote:

Dear Colleagues,


We're voulenteers.



There is a hypervisor called bhyve for FreeBSD. It's completely
headless, no graphics, runs as a daemon and provides serial and VNC
consoles.

Can you please advise a similar headless and minimal hypervisor for
Debian or Ubuntu?



Libvirt with or without apt recommend.


Please don't just say "kvm". I've tried installing different
combinations of "qemu-kvm", "virt-manager" etc and they all depend on


Something went rong, if you want headless and you were using virt-manager.


dozens of GUI tools.



Why saying so if you know the answer.


A list of packages for the "apt install" command to install a really
minimal hypervisor would be very much appreciated.


You need to figure that out on your own, the Debian wiki comes to mind
though.


I'm not really
afraid of writing a couple of text or YAML configuration files to
describe VMs if it helps me avoid the GUI configuration.


Look at Puppet or Ansible to provision your infrastructure.

What you want is  definitely possible in Debian (headless host and guest).

--
John Doe

Re: A hypervisor for a headless server?

[Geert Stappers]

On Thu, Jun 01, 2023 at 11:53:26PM -0400, Brian Sammon wrote:
> On Fri, 2 Jun 2023 02:39:43 + Victor Sudakov wrote:
> 
> > Please don't just say "kvm". I've tried installing different
> > combinations of "qemu-kvm", "virt-manager" etc and they all depend on
> > dozens of GUI tools.
> 
> I think the problem that you are running into with qemu-kvm is that
> while the package doesn't "require" GUI tools, they do "recommend" them.
> If you turn off "install-recommends" in your apt config, you may find
> that you can install it without the GUI tools.
 
That might indeed be a solution to the original "problem".

> "virt-manager", on the other hand, appears to be fundamentally a GUI tool.

Original poster might be looking for `virsh`.


|$ man virsh | head
|VIRSH(1)  Virtualization Support  VIRSH(1)
|
|NAME
|   virsh - management user interface
|
|SYNOPSIS
|   virsh [OPTION]... [COMMAND_STRING]
|
|   virsh [OPTION]... COMMAND [ARG]...
|
|$ dpkg -S virsh
|libvirt-clients: /usr/bin/virsh
|libvirt-clients: /usr/share/man/man1/virsh.1.gz
|libvirt-clients: /usr/share/bash-completion/completions/virsh
|$ 
 

Groeten
Geert Stappers
-- 
Silence is hard to parse

Re: A hypervisor for a headless server?

[Andy Smith]

Hello,

On Fri, Jun 02, 2023 at 03:33:09AM +, Victor Sudakov wrote:
> So what's the package name for just "kvm" without the GUI tools? Because

For someone that wants to run a hypervisor in a non-newbie manner
I'm afraid you don't seem to be that willing to do any of your own
research.

The package is still qemu-kvm (a virtual package that probably
depends on qemu-system-x86 for you).

By default that will install a ton of recommended packages that
include all the GUI tools you probably object to, but that is just
how Debian works [and necessarily how a general purpose binary Linux
distribution has to work].

If you want the minimum amount of packages to be installed you can
try installing without recommends, i.e.

# apt --no-install-recommends install qemu-kvm

which will dramatically reduce the number of packages, BUT:

- you may miss some package that's really useful in most cases -
  installing without recommends is not a supported use case.

- you'll probably still get some packages you will never use.

This is all very basic Debian systems administration and so if you
weren't aware of these things I question whether you would have an
easy time trying to strip a Debian install of KVM down to the bare
minimum instead of going the easier route of just doing what there
are thousands of guides out there for.

If the idea of the installation of binary package dependencies that
you never use massively offends you I would suggest that Debian is
not an ideal match for you, and you may be better off going with
Gentoo or Arch or something.

Cheers,
Andy

-- 
https://bitfolk.com/ -- No-nonsense VPS hosting

Re: A hypervisor for a headless server?

[Brian Sammon]

On Fri, 2 Jun 2023 02:39:43 +
Victor Sudakov  wrote:

> Please don't just say "kvm". I've tried installing different
> combinations of "qemu-kvm", "virt-manager" etc and they all depend on
> dozens of GUI tools.

I think the problem that you are running into with qemu-kvm is that while the 
package doesn't "require" GUI tools, they do "recommend" them.  If you turn off 
"install-recommends" in your apt config, you may find that you can install it 
without the GUI tools.

"virt-manager", on the other hand, appears to be fundamentally a GUI tool.

Re: A hypervisor for a headless server?

[Victor Sudakov]

Andy Smith wrote:
> Hello,
> 
> On Fri, Jun 02, 2023 at 02:39:43AM +, Victor Sudakov wrote:
> > Please don't just say "kvm". I've tried installing different
> > combinations of "qemu-kvm", "virt-manager" etc and they all depend on
> > dozens of GUI tools.
> 
> "kvm" is the generally accepted answer. None of the GUI tools are
> necessary, they are optional extras. There are thousands upon

So what's the package name for just "kvm" without the GUI tools? Because

# apt install kvm
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Package kvm is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

E: Package 'kvm' has no installation candidate


-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature

Re: A hypervisor for a headless server?

[Andy Smith]

Hello,

On Fri, Jun 02, 2023 at 02:39:43AM +, Victor Sudakov wrote:
> Please don't just say "kvm". I've tried installing different
> combinations of "qemu-kvm", "virt-manager" etc and they all depend on
> dozens of GUI tools.

"kvm" is the generally accepted answer. None of the GUI tools are
necessary, they are optional extras. There are thousands upon
thousands of VM hosting companies out there running kvm on Linux and
providing only SSH and VNC access to guests.

Xen is another option, but you'll find far more guides for kvm.

I've used both kvm and Xen in the way you describe.

Cheers,
Andy

-- 
https://bitfolk.com/ -- No-nonsense VPS hosting

Re: A hypervisor for a headless server?

[Nicholas Geovanis]

On Thu, Jun 1, 2023, 9:58 PM Victor Sudakov  wrote:

> Dear Colleagues,
>
> There is a hypervisor called bhyve for FreeBSD. It's completely
> headless, no graphics, runs as a daemon and provides serial and VNC
> consoles.
>
> Can you please advise a similar headless and minimal hypervisor for
> Debian or Ubuntu?
>

Just don't install x-windows or anything that depends on it, like a desktop
environment. Servers in datacenters run headless more than 95% of cases.
Debian and its derivatives too.


Please don't just say "kvm". I've tried installing different
> combinations of "qemu-kvm", "virt-manager" etc and they all depend on
> dozens of GUI tools.
>
> A list of packages for the "apt install" command to install a really
> minimal hypervisor would be very much appreciated. I'm not really
> afraid of writing a couple of text or YAML configuration files to
> describe VMs if it helps me avoid the GUI configuration.
>
> --
> Victor Sudakov VAS4-RIPE
> http://vas.tomsk.ru/
> 2:5005/49@fidonet
>

A hypervisor for a headless server?

[Victor Sudakov]

Dear Colleagues,

There is a hypervisor called bhyve for FreeBSD. It's completely
headless, no graphics, runs as a daemon and provides serial and VNC
consoles.

Can you please advise a similar headless and minimal hypervisor for
Debian or Ubuntu?

Please don't just say "kvm". I've tried installing different
combinations of "qemu-kvm", "virt-manager" etc and they all depend on
dozens of GUI tools.

A list of packages for the "apt install" command to install a really
minimal hypervisor would be very much appreciated. I'm not really
afraid of writing a couple of text or YAML configuration files to
describe VMs if it helps me avoid the GUI configuration.

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet


signature.asc
Description: PGP signature