[gentoo-user] Thanks and bye for now

From:
Richard Cox <conardcox DeleteThis @gmail.com>
To:
Roy Wright <roy DeleteThis @wright.org>
Date:
Today 01:08:51
> > Gentoo is difficult to install.

A highly subjective statement to be sure.  Many thousands have successfully
installed it...depends on your definition of 'difficult' I suppose.

>>Also, if it's left un-updated for
>> longer periods of time, it tends to break on the next update.
  So does any dynamic system that is allowed to stagnate.  May I propose a
solution?  Don't leave it un-updated for 'a long period of time'.  It's not
that hard, really...if you are super paranoid, just emerge --sync once a week
and then emerge -up --deep world...I know, that's rocket science, but it can
give you an up-to-date system with little trouble...assuming you have
internet access, of course.

  Again, I suppose that depends on what you mean by 'a long period of time'.  
Let Debian or any other distro remain un-updated for a year or more tell me
about how easy it is to update without breaking.  Actually, that wasn't
fair...because you are basically going to do a re-install (or version
upgrade, as many distros call it these days) when that happens.

On Sunday 28 December 2008 00:50:32 you wrote:
> Nikos Chantziaras wrote:
> > Gentoo is difficult to install.  Also, if it's left un-updated for
> > longer periods of time, it tends to break on the next update.  I guess
> > that's the downside of being versionless.  Debian on the other hand, due
> > to it being versioned, doesn't have that problem.
>
> When I ran an internal gentoo server at my last job, I would try to
> schedule an update about once a month.  That was a lesson learned after
> the box just worked great for 7 months, then we wanted to add a new
> application that needed newer libraries, that turned into a 2 day
> marathon to update the server (it was old, slow hardware).
>
> I will predict that you will most miss portage after having to deal with
> the brain dead apt package manager.
>
> Good Luck

On Sunday 28 December 2008 00:50:32 Roy Wright wrote:
> Nikos Chantziaras wrote:
> > Gentoo is difficult to install. Also, if it's left un-updated for
> > longer periods of time, it tends to break on the next update. I guess
> > that's the downside of being versionless. Debian on the other hand, due
> > to it being versioned, doesn't have that problem.
>
> When I ran an internal gentoo server at my last job, I would try to
> schedule an update about once a month. That was a lesson learned after
> the box just worked great for 7 months, then we wanted to add a new
> application that needed newer libraries, that turned into a 2 day
> marathon to update the server (it was old, slow hardware).
>
> I will predict that you will most miss portage after having to deal with
> the brain dead apt package manager.
>
> Good Luck

You're welcome. Bye.

On Friday 26 December 2008 18:19:11 Mark Kirkwood wrote:
> Due to a new work situation where extensive use is made of Debian, I
> feel the need to have a Debian-based play server. This unfortunately
> means my trusty Gentoo box is to be sacrificed
>
> Thanks for the help I have received over the last few years (think I
> joined in 2005). I have enjoyed being part of the Gentoo community, and
> hope to be back alter sometime.
>
> Best wishes
>
> Mark

On Saturday 27 December 2008 21:13:49 Nikos Chantziaras wrote:
> Dale wrote:
> > Nikos Chantziaras wrote:
> >> On my Gentoo at home, yes. The mouse cursor skips, scrolling gets
> >> skippy/laggy too. I have a dual core E6600 RemoveThis @3.33Ghz with 4GB DDR2 RAM.
> >
> > I have to use version 2.6.23-gentoo-r8 for my kernel or it does the same
> > thing. Someone mentioned that it is a setting in the kernel for one of
> > the new features. At some point I plan to post the info here and try to
> > figure out what setting I should use but just haven't done it yet.
> >
> > The only other thing I can think of is the drives being busy. I think
> > they have ionice now too.
>
> I have this in my make.conf:
>
> PORTAGE_IONICE_COMMAND="ionice -c 3 -p \${PID}"
>
> Helped a bit. But still the GUI (KDE 3.5.10) gets pretty laggy. Just a
> few hours ago I updated to gcc-4.3.2-r1. Even with nice 19 and ionice
> 3, lag is there.
>
> I hope someone finds the magic button in the kernel config to fix that

There isn't one - at least not one that really works.

Linux mostly ignores NICE and has done so since day one. The reason according
to Linux himself on some LKML post quite a while back is that Linux has a
semi-decent task scheduler and nice is a 100% manual task scheduler.
Therefore nice is not needed. It did have some uses, such as respecting nice
settings for having X if set to something very negative - makes gui apps more
responsive (X tends to use little cpu and IO time overall but users want it
to be responsive). This has largely gone away with Ingo's last task
scheduler.

On other Unixes, nice has normally been nothing more than a gentle hint to the
kernel how the admin would like the systems to treat a certain process.
That's why Linux could ignore it and get away with it.

You will likely always experience lag compiling something like gcc. It uses
gcc to build a new one, and gcc grabs enormous amounts of memory to do this.
Plus it's rather disk intensive as well. So, running gcc on a large build is
likely to produce lags anyway due to swap and IO no matter how you nice it.
More so if resources are constrained.



--
alan dot mckinnon at gmail dot com

Alan McKinnon wrote:
> On Saturday 27 December 2008 21:13:49 Nikos Chantziaras wrote:
>> I have this in my make.conf:
>>
>> PORTAGE_IONICE_COMMAND="ionice -c 3 -p \${PID}"
>>
>> Helped a bit. But still the GUI (KDE 3.5.10) gets pretty laggy. Just a
>> few hours ago I updated to gcc-4.3.2-r1. Even with nice 19 and ionice
>> 3, lag is there.
>>
>> I hope someone finds the magic button in the kernel config to fix that
>
> There isn't one - at least not one that really works.
>
> Linux mostly ignores NICE and has done so since day one. The reason according
> to Linux himself on some LKML post quite a while back is that Linux has a
> semi-decent task scheduler and nice is a 100% manual task scheduler.

It got worse with kernel 2.6.28, btw, especially mouse cursor movement.
It gets stuck and skips very noticeably. Fortunately, it's not a
point yet where I would describe it as unusable, but if the trend
continues, desktop will be totally unusable by 2.6.31/32 when doing
something that produces load.

On Sun, Dec 28, 2008 at 2:07 AM, Mick <michaelkintzios.DeleteThis@gmail.com> wrote:
> On Saturday 27 December 2008, Nikos Chantziaras wrote:
>> >> Another reason I
>> >> didn't put Gentoo on the server is because everyone would start spamming
>> >> the forums about lag when I emerge -u world while they're getting frags
>> >> in Counter-Strike
>> >
>> > set PORTAGE_NICENESS=19 in /etc/make.conf
>>
>> I'll just hay "ah, ah, ah" at that one OK, I'll also say that it
>> doesn't work. Everything lags even with 19.
>
> Is that measurable?
Niceness effects are most easily noticeable for CPU bottlenecks,
because context switching CPUs is relatively painless. Server loads
are probably more of Memory / IO bottlenecks, and context switching
between that involves disk swaps and disk prereads. If it is anything
more than trivial prereads / swaps, then even a 19-niced app can cause
noticeable bursts of slowdown on a modern system. Niceness isn't
magic, and compiling, which could easily max out all 3 resources (CPU,
memory, disk access, 4 if you add network), probably gets the least
bump from niceness.

Although that isn't to say that PORTAGE_NICENESS doesn't take effect.
Desktop machines with a single user will probably have speed bursts
small enough that it wouldn't matter. But I'd think twice if there was
something intensive supposed to be done on the server.

Nikos Chantziaras wrote:
> Alan McKinnon wrote:
>> On Saturday 27 December 2008 21:13:49 Nikos Chantziaras wrote:
>>> I have this in my make.conf:
>>>
>>> PORTAGE_IONICE_COMMAND="ionice -c 3 -p \${PID}"
>>>
>>> Helped a bit. But still the GUI (KDE 3.5.10) gets pretty laggy.
>>> Just a
>>> few hours ago I updated to gcc-4.3.2-r1. Even with nice 19 and ionice
>>> 3, lag is there.
>>>
>>> I hope someone finds the magic button in the kernel config to fix
>>> that
>>
>> There isn't one - at least not one that really works.
>>
>> Linux mostly ignores NICE and has done so since day one. The reason
>> according to Linux himself on some LKML post quite a while back is
>> that Linux has a semi-decent task scheduler and nice is a 100% manual
>> task scheduler.
>
> It got worse with kernel 2.6.28, btw, especially mouse cursor
> movement. It gets stuck and skips very noticeably. Fortunately, it's
> not a point yet where I would describe it as unusable, but if the
> trend continues, desktop will be totally unusable by 2.6.31/32 when
> doing something that produces load.
>
>
>

So it is not just me that has screwy mouse movement. I been using a old
kernel for a while now and even thought it was just a wrong setting on
my part.

Has anyone been telling the kernel folks about this problem so they can
fix it or roll something back?

Dale

On Sunday 28 December 2008 22:52:08 Dale wrote:
> Nikos Chantziaras wrote:

> > It got worse with kernel 2.6.28, btw, especially mouse cursor
> > movement. It gets stuck and skips very noticeably. Fortunately, it's
> > not a point yet where I would describe it as unusable, but if the
> > trend continues, desktop will be totally unusable by 2.6.31/32 when
> > doing something that produces load.
>
> So it is not just me that has screwy mouse movement. I been using a old
> kernel for a while now and even thought it was just a wrong setting on
> my part.
>
> Has anyone been telling the kernel folks about this problem so they can
> fix it or roll something back?

The cynic in me wants to say that Colin Kolivas tried telling the kernel devs
for years about it and got stone-walled and ignored for years, despite
maintaining a set of desktop patches that worked really well. Eventually he
gave up and walked away in disgust when Ingo Molnar submitted scheduler
patches that looked awfully like Colin's, and his were accepted....

I'm not that much of a cynic though. Instead I'll recommend you find a set of
desktop patches that work well and roll a kernel from those. The kernel devs
are mostly paid by organizations that have a vested interest in having Linux
work fabulously on big iron, so that's where the focus will tend to go.

This is opensource, where you get to bash the code into any shape you need to
get it to suit your needs I myself don't need desktop patches (yet), but
if I did, I would probably first look at the patches the Ubuntu kernel devs
apply, followed by fedora


--
alan dot mckinnon at gmail dot com

Alan McKinnon wrote:
> On Sunday 28 December 2008 22:52:08 Dale wrote:
>
>> Nikos Chantziaras wrote:
>>
>
>
>>> It got worse with kernel 2.6.28, btw, especially mouse cursor
>>> movement. It gets stuck and skips very noticeably. Fortunately, it's
>>> not a point yet where I would describe it as unusable, but if the
>>> trend continues, desktop will be totally unusable by 2.6.31/32 when
>>> doing something that produces load.
>>>
>> So it is not just me that has screwy mouse movement. I been using a old
>> kernel for a while now and even thought it was just a wrong setting on
>> my part.
>>
>> Has anyone been telling the kernel folks about this problem so they can
>> fix it or roll something back?
>>
>
> The cynic in me wants to say that Colin Kolivas tried telling the kernel devs
> for years about it and got stone-walled and ignored for years, despite
> maintaining a set of desktop patches that worked really well. Eventually he
> gave up and walked away in disgust when Ingo Molnar submitted scheduler
> patches that looked awfully like Colin's, and his were accepted....
>
> I'm not that much of a cynic though. Instead I'll recommend you find a set of
> desktop patches that work well and roll a kernel from those. The kernel devs
> are mostly paid by organizations that have a vested interest in having Linux
> work fabulously on big iron, so that's where the focus will tend to go.
>
> This is opensource, where you get to bash the code into any shape you need to
> get it to suit your needs I myself don't need desktop patches (yet), but
> if I did, I would probably first look at the patches the Ubuntu kernel devs
> apply, followed by fedora
>
>
>

I use the Gentoo sources so why don't they patch them? I've tried
patching kernels before and it didn't work to well.

Dale

On Sun, Dec 28, 2008 at 2:57 PM, Alan McKinnon <alan.mckinnon DeleteThis @gmail.com> wrote:
> On Sunday 28 December 2008 22:52:08 Dale wrote:
>> Nikos Chantziaras wrote:
>
>> > It got worse with kernel 2.6.28, btw, especially mouse cursor
>> > movement. It gets stuck and skips very noticeably. Fortunately, it's
>> > not a point yet where I would describe it as unusable, but if the
>> > trend continues, desktop will be totally unusable by 2.6.31/32 when
>> > doing something that produces load.
>>
>> So it is not just me that has screwy mouse movement. I been using a old
>> kernel for a while now and even thought it was just a wrong setting on
>> my part.
>>
>> Has anyone been telling the kernel folks about this problem so they can
>> fix it or roll something back?

Getting in a little late but my AMD64 machine's mouse died so I
grabbed a VERY cheap $6 Microsoft mouse I had received for free some
years ago and never used. It works but often jumps to one corner or
another at odd times. It sits atop one of my Gentoo Linux mouse pads
and I've noticed that if I turn the mouse pad 45 degrees it happens
far more often.

I figured it was just a cheap mouse. It hadn't occurred to me that it
might be a driver issue...

- Mark

On Monday 29 December 2008 01:06:59 Dale wrote:
> > I'm not that much of a cynic though. Instead I'll recommend you find a
> > set of desktop patches that work well and roll a kernel from those. The
> > kernel devs are mostly paid by organizations that have a vested interest
> > in having Linux work fabulously on big iron, so that's where the focus
> > will tend to go.
> >
> > This is opensource, where you get to bash the code into any shape you
> > need to get it to suit your needs I myself don't need desktop patches
> > (yet), but if I did, I would probably first look at the patches the
> > Ubuntu kernel devs apply, followed by fedora
> >
> >
> >  
>
> I use the Gentoo sources so why don't they patch them?  I've tried
> patching kernels before and it didn't work to well.

You'll have to ask the gentoo devs why they apply the patches they do. After
all, not every patch set out there will suit their goals and purposes.

--
alan dot mckinnon at gmail dot com

Alan McKinnon wrote:
> On Monday 29 December 2008 01:06:59 Dale wrote:
>
>>> I'm not that much of a cynic though. Instead I'll recommend you find a
>>> set of desktop patches that work well and roll a kernel from those. The
>>> kernel devs are mostly paid by organizations that have a vested interest
>>> in having Linux work fabulously on big iron, so that's where the focus
>>> will tend to go.
>>>
>>> This is opensource, where you get to bash the code into any shape you
>>> need to get it to suit your needs I myself don't need desktop patches
>>> (yet), but if I did, I would probably first look at the patches the
>>> Ubuntu kernel devs apply, followed by fedora
>>>
>>>
>>>
>>>
>> I use the Gentoo sources so why don't they patch them? I've tried
>> patching kernels before and it didn't work to well.
>>
>
> You'll have to ask the gentoo devs why they apply the patches they do. After
> all, not every patch set out there will suit their goals and purposes.
>
>

I was curious if they know some desktop puters are even having this
problem or not. After all, if they don't know, they won't even consider
patching it then. I'm just sticking with 2.6.23 I guess. It works well.

Dale

On Mon, 29 Dec 2008 00:57:13 +0200
Alan McKinnon wrote:

....[snip]...

> This is opensource, where you get to bash the code into any shape you
> need to get it to suit your needs I myself don't need desktop
> patches (yet), but if I did, I would probably first look at the
> patches the Ubuntu kernel devs apply, followed by fedora

Alan,

Curiosity prompts me to ask: have you a link to Ubuntu's kernel
patches?

Regards,

David

I'm having the same exact problem with my mouse. I'm on AMD64 and my
cursor often jumps to a corner.

Is there a fix?

-Chris

On Sun, Dec 28, 2008 at 6:08 PM, Mark Knecht <markknecht.RemoveThis@gmail.com> wrote:
> On Sun, Dec 28, 2008 at 2:57 PM, Alan McKinnon <alan.mckinnon.RemoveThis@gmail.com> wrote:
>> On Sunday 28 December 2008 22:52:08 Dale wrote:
>>> Nikos Chantziaras wrote:
>>
>>> > It got worse with kernel 2.6.28, btw, especially mouse cursor
>>> > movement. It gets stuck and skips very noticeably. Fortunately, it's
>>> > not a point yet where I would describe it as unusable, but if the
>>> > trend continues, desktop will be totally unusable by 2.6.31/32 when
>>> > doing something that produces load.
>>>
>>> So it is not just me that has screwy mouse movement. I been using a old
>>> kernel for a while now and even thought it was just a wrong setting on
>>> my part.
>>>
>>> Has anyone been telling the kernel folks about this problem so they can
>>> fix it or roll something back?
>
> Getting in a little late but my AMD64 machine's mouse died so I
> grabbed a VERY cheap $6 Microsoft mouse I had received for free some
> years ago and never used. It works but often jumps to one corner or
> another at odd times. It sits atop one of my Gentoo Linux mouse pads
> and I've noticed that if I turn the mouse pad 45 degrees it happens
> far more often.
>
> I figured it was just a cheap mouse. It hadn't occurred to me that it
> might be a driver issue...
>
> - Mark
>
>