Re: Issue in glibc in FC3T3 ?

[Ron Yorston]

> Elliot Lee <sopwith(a)redhat.com&gt; wrote:
> On Thu, 28 Oct 2004, Kim Lux (Mr.) wrote:
>
> > I did a freshen of both Qt and Glibc and now I've got prelink and ld
> > running as root !!!!???? Prelink is stealing all my cpu cycles.  What is
> > up with that ?  I finally had to kill it..
>
> prelink speeds up app startup - it does the work once to save startup
> times repeatedly. 
I'm not convinced, for my pattern of usage at least, that the game is
worth the candle.  On my laptop I start my desktop, a few xterms and
maybe Mozilla, then leave them running.  Any time that might be saved
on startup is outweighed by the inconvenience of having the system become
unresponsive when anacron kicks in to perform the prelinking.  Which
always seems to happen at the most awkward moment.  (Yes, I know I can
turn it off.  And I have.)

Prelink seems to do an enormous amount of work when it runs.  Have the
developers checked that all of this effort is really necessary?

What might help would be the ability to limit the resources (CPU, disk
I/O, whatever) that are used by background processes.  Nice is nice, but
not powerful enough.  Something like cpucap

   http://www.rshk.co.uk/projects/cpucap/

would seem to be a better option.

Ron

Issue in glibc in FC3T3 ?

[nodata]

> >Elliot Lee <sopwith(a)redhat.com&gt; wrote:
> >On Thu, 28 Oct 2004, Kim Lux (Mr.) wrote:
> >
> >> I did a freshen of both Qt and Glibc and now I've got prelink and ld
> >> running as root !!!!???? Prelink is stealing all my cpu cycles.  What
> >> is
> >> up with that ?  I finally had to kill it..
> >
> >prelink speeds up app startup - it does the work once to save startup
> >times repeatedly.
>
>  I'm not convinced, for my pattern of usage at least, that the game is
>  worth the candle.  On my laptop I start my desktop, a few xterms and
>  maybe Mozilla, then leave them running.  Any time that might be saved
>  on startup is outweighed by the inconvenience of having the system become
>  unresponsive when anacron kicks in to perform the prelinking.  Which
>  always seems to happen at the most awkward moment.  (Yes, I know I can
>  turn it off.  And I have.)
>
>  Prelink seems to do an enormous amount of work when it runs.  Have the
>  developers checked that all of this effort is really necessary?
>
>  What might help would be the ability to limit the resources (CPU, disk
>  I/O, whatever) that are used by background processes.  Nice is nice, but
>  not powerful enough.  Something like cpucap
>
>     http://www.rshk.co.uk/projects/cpucap/
>
>  would seem to be a better option. 
Is there a potential for lots of race conditions if multiple resources
(i.e. all of cpu, memory and disk) are throttled?

>  Ron
>
>  --
>  fedora-test-list mailing list
>  fedora-test-list(a)redhat.com
>  To unsubscribe:
>  http://www.redhat.com/mailman/listinfo/fedora-test-list
>

Issue in glibc in FC3T3 ?

[Klaasjan Brand]

>  >Elliot Lee <sopwith(a)redhat.com&gt; wrote:
>  >On Thu, 28 Oct 2004, Kim Lux (Mr.) wrote:
>  >
>  >> I did a freshen of both Qt and Glibc and now I've got prelink and ld
>  >> running as root !!!!???? Prelink is stealing all my cpu cycles.  What is
>  >> up with that ?  I finally had to kill it..
>  >
>  >prelink speeds up app startup - it does the work once to save startup
>  >times repeatedly. 
Then isn't it more convenient to run it when installing packages using
rpm (when I expect some thrashing) and let the user decide if and when
to prelink manually installed packages?

Klaasjan

Issue in glibc in FC3T3 ?

[Per Bjornsson]

On Tue, 2004-11-02 at 04:38, Klaasjan Brand wrote:
>  > >prelink speeds up app startup - it does the work once to
> save startup
>  > >times repeatedly.
>  
>  Then isn't it more convenient to run it when installing packages using
>  rpm (when I expect some thrashing) and let the user decide if and when
>  to prelink manually installed packages? 
Not really, since when you install a new library everything that links
to it needs to be prelinked again. Say there's a glibc update; do you
really want the computer to thrash for between 10 minutes and half an
hour, depending on how fast your hardware is?

/Per

FC3T3rc3: "Release Notes" output error

[ron flory]

hi-

  This is mostly a cosmetic issue, but if one opens the
FC3T3rc3 "Rlease Notes" window of the installer, the last char
of section headings, bullet points, and some bold text is wrapped
around to the next line, for example:

  "make oldconfig" is displayed as:

  "make oldconfi
   g"

ron

FC3T3rc3: 'Release Notes' output error

[nodata]

>  hi-
>
>    This is mostly a cosmetic issue, but if one opens the
>  FC3T3rc3 "Rlease Notes" window of the installer, the last char
>  of section headings, bullet points, and some bold text is wrapped
>  around to the next line, for example:
>
>    "make oldconfig" is displayed as:
>
>    "make oldconfi
>     g"
>
>  ron
>
>  --
>  fedora-test-list mailing list
>  fedora-test-list(a)redhat.com
>  To unsubscribe:
>  http://www.redhat.com/mailman/listinfo/fedora-test-list
>  
I think that's a known problem.
Something to do with a Python/anaconda html library and nested lists
methinks. Might be wrong..

FC3T3rc3: 'Release Notes' output error

[Kyrre Ness Sjobak]

Is the releasenotes anywhere to be found?

Just wanna take a look at it :)

tir, 02.11.2004 kl. 20.13 skrev nodata:
>  > hi-
>  >
>  >   This is mostly a cosmetic issue, but if one opens the
>  > FC3T3rc3 "Rlease Notes" window of the installer, the last char
>  > of section headings, bullet points, and some bold text is wrapped
>  > around to the next line, for example:
>  >
>  >   "make oldconfig" is displayed as:
>  >
>  >   "make oldconfi
>  >    g"
>  >
>  > ron
>  >
>  > --
>  > fedora-test-list mailing list
>  > fedora-test-list(a)redhat.com
>  > To unsubscribe:
>  > http://www.redhat.com/mailman/listinfo/fedora-test-list
>  >
>  
>  I think that's a known problem.
>  Something to do with a Python/anaconda html library and nested lists
>  methinks. Might be wrong..

Issue in glibc in FC3T3 ?

[Kyrre Ness Sjobak]

tir, 02.11.2004 kl. 18.12 skrev Per Bjornsson:
>  On Tue, 2004-11-02 at 04:38, Klaasjan Brand wrote:
>  > > >prelink speeds up app startup - it does the work once to save startup
>  > > >times repeatedly.
> >   
>  > Then isn't it more convenient to run it when
> installing packages using
>  > rpm (when I expect some thrashing) and let the user decide if and when
>  > to prelink manually installed packages?
>  
>  Not really, since when you install a new library everything that links
>  to it needs to be prelinked again. Say there's a glibc update; do you
>  really want the computer to thrash for between 10 minutes and half an
>  hour, depending on how fast your hardware is?
>   
It wouldn't be a big problem - if there was an option not to do it, rpm
was freed (so you could bacground it/open another console and install
another package), and it was stoppable (control+c).
>

Issue in glibc in FC3T3 ?

[Kyrre Ness Sjobak]

tir, 02.11.2004 kl. 10.50 skrev Ron Yorston:
>  >Elliot Lee <sopwith(a)redhat.com&gt; wrote:
>  >On Thu, 28 Oct 2004, Kim Lux (Mr.) wrote:
>  >
>  >> I did a freshen of both Qt and Glibc and now I've got prelink and ld
>  >> running as root !!!!???? Prelink is stealing all my cpu cycles.  What is
>  >> up with that ?  I finally had to kill it..
>  >
>  >prelink speeds up app startup - it does the work once to save startup
>  >times repeatedly.
>  
>  I'm not convinced, for my pattern of usage at least, that the game is
>  worth the candle.  On my laptop I start my desktop, a few xterms and
>  maybe Mozilla, then leave them running.  Any time that might be saved
>  on startup is outweighed by the inconvenience of having the system become
>  unresponsive when anacron kicks in to perform the prelinking.  Which
>  always seems to happen at the most awkward moment.  (Yes, I know I can
>  turn it off.  And I have.)
>  
>  Prelink seems to do an enormous amount of work when it runs.  Have the
>  developers checked that all of this effort is really necessary?
>  
>  What might help would be the ability to limit the resources (CPU, disk
>  I/O, whatever) that are used by background processes.  Nice is nice, but
>  not powerful enough.  Something like cpucap
>  
>     http://www.rshk.co.uk/projects/cpucap/
>  
>  would seem to be a better option.
>   
agreed. But as far as i know, the CPU is seldom the limiting factor when
running a multi-tasking OS such as Linux. And i am using a 650 Mhz. What
i mean, is that i hardly notice when the CPU is chrunching away at full
speed, but when something hits disk (updatedb and sometimes memory
(OpenOffice dictionary load) hard - it is noticed.