Re: [vserver] Rant/Moved

From: Herbert Poetzl <herbert_at_13thfloor.at>
Date: Thu 21 Oct 2010 - 14:02:24 BST
Message-ID: <20101021130224.GB2708@MAIL.13thfloor.at>

On Thu, Oct 21, 2010 at 02:27:15PM +0200, Rik Bobbaers wrote:
> Just a small question, bertl:

first, please don't hijack threads, especially closed ones
(I can only consider this trolling :)

> People that use vserver in production environment want a
> stable environment.

then they should pay some developers, preferably the
existing ones, to do the testing and code cleanups

> If you ever want linux-vserver to become something widely
> used and accepted, there should be a stable patch, there
> should be warnings if something is wrong and so on...

there should also be contributions, this isn't a one
man show. period.

> You should inform people as much as possible and cooperate
> with people doing packaging for distributions.

I always cooperate with any distro maintainer and even
volunteered a bunch of times to do the initial port for
a given distro (to ease adoption)

> As i get it, you hate every distribution in the world and
> don't want to have anything to do with their patches.

no, I do not want to maintain any distro patches over
a longer period, and that is mainly because I am only
one person, and after all, that's what distro maintainers
are for (maintaining distro specific packages)

> If you just say: that's a crappy kernel, just upgrade...
> it's fun when you're playing around.

well, what should I say about a crappy kernel then?

> not if you 've been doing tests etc for weeks
> before putting stuff in production.

if you are referring to the debian kernel, I strongly
advised against using the 2.6.26 patches, but despite
my warnings, it was adopted, and of course, it turned
out to be a fiasco ...

> Right now, what we have is a very clear dev version of
> util-vserver, a 2.3 patch series that's relatively stable
> and testing version of the utilities

I do not consider the experimental releases 'dev' or
production ready, despite the fact that they work fine
in production for many people out there

> All the rest is is very old (kernel 2.6.22 and 0.30.215
> of util-vserver).
> Whenever someone asks a question about a kernel that old
> or version of util-vserver that's that old, you just say:
> upgrade to the latest.

no, the latest stable release and stable util-vserver as
well as older stable releases will work fine, as they are
thoroughly tested, and given that somebody reports a bug
to that kernel/patch version, it will of course be fixed

folks come to me with 'obscure' kernel versions, because
their distro or hardware or simply because they randomly
picked a kernel version and complain about issues which
are usually known and already fixed in more recent patches

there are a few maintained branches (I do that exactly
because those are long term kernels to be used by distros)
like 2.6.27 or now 2.6.31/32 and of course recent kernel
branches to give some choices ...

> so essentially, you're saying "put the latest
> unstable/testing kernel in production, but don't
> complain if it dies, because i call it dev, so
> you're on your own."

what I am saying (usually) is, if you want a recent
kernel/patch, you have to live with experimental, because
we do not have the resources to do the required testing
for a stable release ...

> All that i do for the grsecurity/linux-vserver patch,
> i just do because people want it.
> There are many people that really use it and find it
> useful...

good, that's why we have the patches linked on the
main page, no?
but I doubt that you do the extensive testing required
to label those patches 'stable' and that doesn't even
relate to the Linux-VServer part :)

> so here's the question: What is your view on this
> project? do you really hate all the users that want
> this to be production-ready?

definitely not, all users that want a stable release,
please start testing/reviewing the code/contributing
to the project, so that we can get there eventually

> do you really hate the grsec/vserver patch?

I am fairly indifferent about the grsec patch ...
I hate the fact that folks believe that adding the
grsec patch to the mix will increase security out
of the box (without doing proper setups) and I hate
the fact that there is no properly adapted grsec
which handles the special aspects of Linux-VServer

> if so, just say so, and i will stop doing anything
> all together.

It seems that the community likes those patches, so
the only folks you would hurt are those who use the
patches ...

> If you keep saying: you're using a distro kernel, so
> screw you!, i don't want to be involved.

I'm saying: if you want to use a distro kernel, how
broken it may be, go ahead, be my guest, but if you
encounter known or unknown issues with that kernel,
you have to contact the distribution maintainer and
not me ...

once again, I cannot fix every distribution kernel
out there, mostly because I do not have the time for
that and in certain cases (debian for example) because
they refuse to fix issues at all ...

> It's normal that people want something that WORKS.
> If you say: the only thing worth running now is 2.3
> with the latest util-vserver, then make it stable/
> support it/whatever! but don't tell people "there

all recent kernels/patches are 'supported' and well
maintained ...

> is no reason not to upgrade to the latest version of
> everything", when that version is unstable and not
> working properly

let me put it this way, why not install linux-2.6.36-rc1
with the coresponding Linux-VServer pre-release patch?

probably because -rc2 to -rc7 fixed issues found during
mainline stabilization, and similar happened for Linux-
VServer patches, so AFAICT, there is no reason whatsoever
to use linux-2.6.36-rc1 in production ...

> Sorry for my rant, but some people here (including me)
> would like something stable that's not almost 3 years
> old.

then start contributing to the stabilization, it's a
community project after all ....

> If not, this project can be considered dead and people
> will abandon this project.

as with Linux-VServer guests, this project will be dead
when the last 'user' dies ...

> so for me it's time to choose: or, you start making
> it work and start supporting stuff or, this project
> will die because of frustration of all the users that
> just get a : you don't have the latest utils that just
> got out yesterday and you don't use the patch i
> uploaded 2 days ago.

well, there is no way to travel back in time and fix
issues before they get into a patch, so if somebody
comes to me with a known (and thus fixed) issue, then
I'll advise him/her to use the fixed patch/tools because
that is the easiest way to get the problem solved ...

I don't think it would be better to advise them to
backport changes to their kernel version, because that
is the job a distro maintainer is supposed to do, given
that they want to stick with an older kernel.

> just my 2 cents,
>
> KR,
>
> Rik Bobbaers
Received on Thu Oct 21 14:02:32 2010

[Next/Previous Months] [Main vserver Project Homepage] [Howto Subscribe/Unsubscribe] [Paul Sladen's vserver stuff]
Generated on Thu 21 Oct 2010 - 14:02:34 BST by hypermail 2.1.8