[Linux-HA] Death of my servers!

[Frank R Callaghan]

On Tuesday 09 March 2004 01:20 pm, Alan Robertson wrote:
> Alan Robertson wrote:
> > Frank R Callaghan wrote:
> >> Thanks Alan,
> >>
> >>>> It seems that the loading caused HB to miss beats and conclude
> >>>> that I'm dead so take my services - and then missed lserver3's
> >>>> heartbeat
> >>>> and killed him too !
> >>>>
> >>>> Is this what suposed to happen under heavy load ?
> >>>
> >>> If heartbeat cannot hear it's own heart beat for 'deadtime', then it
> >>> will
> >>> restart itself - because that means *something* bad has happened.
> >>> (and it
> >>> doesn't, of course, know what).
> >>
> >> But should it conclude that the other server is dead also ? that seems
> >> to defete the whole point !
> >
> > If the scheduler fails to schedule heartbeat for a long time, it will
> > unfortunately think everyone is dead.
> >
> > Looking forward to hearing back from you on 1.2.0!
>
1.2.0 is the business, I have loaded my servers to 100% with everything
I can throw at them - an not a beat was missed :) what a great piece of s/w
even DRBD seems solid now i've moved from reiser back to ext3 !

> One other little tip:
> 	avoid logging into files directly.  Use syslog (logfacility)
> 	instead.

Not sure what you mean by this ?

Cheers,
	Frank.