> We are rapidly rolling out a wide area network (inter/intra net) with
> over 44 remote linux server boxes (very remote - many hours drive
> away) connected via ISDN and frame relay via cisco routers to our
> central hub and server farm.
>
> We have UPSs on these linux servers, with powerchute running as a
> daemon. It works fine. Well, almost - when power does go out, the
> system goes down as it is told, but left in an unusable state.
>
> ATX motherboards, yeech! (I never did like the concept, but that's
> what we have so that's what I have to live with). They don't power up
> when the power is restored unless the power button is pressed - a big
> problem if the button on the box box is hundreds of km away :) We are
> looking into getting something like "wake on lan" to manage this
> particular problem (although WoL is not the solution in our particular
> case).
Sounds to me like a configuration problem. I have an ATX system hear built
on an ASUS card. Powers down and up perfectly well.
>
> What doesn't work is the halt command, which is issued either on the
> command line or from the pchute daemon - the standard RH6.1 kernel
> OOPS'es badly (and very impressively!) when it actually halts and
> attempts to power doen, locking up the system in an unusable state
> instead of actually shutting off the power. And in this situation
> nothing BUT a push of the big red button will get things working
> again.
My only problem with the halt command is that when I use it the system
always turns itself off. I'd rather choose at that time whether the
computer should wait for me or turn off, not when I build the kernel. To
be fair, I usually halt the system from gdm of kdm these days; I've not
yet checked whether recent halt/shutdown commands do what I want.
I've not been in the habit of using RH kernels - I build from tarballs.
I've never had an oops powering down.
>
> So from our esperiences, "power off on halt" appears to be badly
> broken in these kernels.
>
> Before I go off recompiling and upgrading kernels, I'd like to know if
> anyone else has experienced this, and if they know of a working
> solution. (Eg, will the current rawhide kernel work?)
You won't know if the current rawhide kernel works on your hardware until
you try it. In your position, I'd try it - it can hardly be worse. But I'd
be sure to have one of those troublesome boxes in the office and only go
on a tour of Qld when I got that working.
--
Cheers
John Summerfield
http://os2.ami.com.au/os2/ for OS/2 support.
Configuration, networking, combined IBM ftpsites index.
--
To unsubscribe:
mail -s unsubscribe [EMAIL PROTECTED] < /dev/null