Paralell startup

[James Harrison]

The origional point I tried to make was that if you dont give some indication
that something failed to start on boot then unless you start diging around in
system files you wont know something failed.

> How would falling back to serial boot help?
Jumping back into serial boot could be an indication that the boot process is
not running correctly and that the operator should do some invesitgation to
fix the problem. Perhaps where serial boot starts could be the indication to
the problem.

> Your definition of 'boot properly' needs explanation. What does it mean to 
> 'boot properly'? 
The machine boots into multiuser mode without any errors in the startup
process.  Sorry if this wasnt clear.

Its pointless having a very fast booting system if the time you saved is spent
finding out why the machine didnt "boot properly".

Error checking and user notification is important.

We already have rhgb as a tool for this.... 

James


--- Roberto Peon <grmoc at yahoo.com> wrote:

> 
> 
> 
> On Wednesday 20 October 2004 01:13 pm, James Harrison wrote:
> > > recache its fonts, or maybe ntpdate is syncing with the time server.
> >
> > Sometimes daemons fail, so there has to be a mechanism to show a failure.
> 
> How would falling back to serial boot help?
> 
> Not launching the services whose dependancies are not satisfied seems like a
> 
> good idea to me. It is better than the current, try it anyway, and see it 
> fail (sometimes spectacularly).
> 
> Your definition of 'boot properly' needs explanation. What does it mean to 
> 'boot properly'? 
> 
> My personal opinion is that you get any system that runs well enough that
> you 
> can get in and fix the problem (hopefully, as quickly as possible...) has 
> booted successfully, if not satisfactorily.
> 
> 
> -Roberto JP
> 
> > What happens if the daemon/service fails that other daemons depend on?
> Does
> > the machine fail to boot properly or does it fail back to a serial boot?
> >
> > --- Jeff Pitman <symbiont at berlios.de> wrote:
> > > On Wednesday 20 October 2004 22:53, Arjan van de Ven wrote:
> > > > problem with parallel startup is that it *ALSO* increases how much
> > > > the disk has to seek, which slows things down. For me it's not clear
> > > > if it's actually a real gain or just a placebo one.
> > >
> > > It's a gain for those really long startup daemons.  Maybe xfs has to
> > > recache its fonts, or maybe ntpdate is syncing with the time server.
> > >
> > > Then again, if Apache were written in a Bash script with all of its
> > > functions in separate script files and configuration spread out in
> > > different files and sourced in, parallelism would be an issue wouldn't
> > > it?
> > >
> > > I actually retooled this: http://www.fefe.de/minit/ for Redhat once and
> > > the results were just night and day.  Of course, I didn't run kudzu or
> > > anything complicated like that.  Just brought up the stuff I needed.
> > > You practically don't even need hibernate or sleep when you get it this
> > > good. (Doesn't help with X/GNOME/KDE startup, though) ;-)
> > >
> > > have fun,
> > > --
> > > -jeff
> > >
> > > --
> > > fedora-devel-list mailing list
> > > fedora-devel-list at redhat.com
> > > http://www.redhat.com/mailman/listinfo/fedora-devel-list
> >
> > __________________________________________________
> > Do You Yahoo!?
> > Tired of spam?  Yahoo! Mail has the best spam protection around
> > http://mail.yahoo.com
> 
> -- 
> fedora-devel-list mailing list
> fedora-devel-list at redhat.com
> http://www.redhat.com/mailman/listinfo/fedora-devel-list
> 




		
_______________________________
Do you Yahoo!?
Declare Yourself - Register online to vote today!
http://vote.yahoo.com