Subject: RE: NetBSD 1.6 & i386 SMP, ACPI, mly
To: 'Brett Lymn' <email@example.com>
From: Conrad T. Pino <Conrad@Pino.com>
Date: 01/16/2003 20:38:46
> -----Original Message-----
> From: Brett Lymn [mailto:firstname.lastname@example.org]
> Sent: Thursday, January 16, 2003 20:00
> To: Conrad T. Pino
> Cc: 'Brett Lymn'; 'Manuel Bouyer'; 'i386 Port'; 'SMP'
> Subject: Re: NetBSD 1.6 & i386 SMP, ACPI, mly
> What it tells you is that NetBSD SMP works well for some people.
> This, at least, gives you hope that you can get something working.
> Whether or not it will work well for you is something that you will
> probably need to find out yourself by trying it.
Thank you. I appreciate the suggestion.
> Just as another datapoint, I am running SMP on my main machine at
> home. I have had a couple of panics (one while doing a large copy
> over the network to another machine, another when I was trying to feed
> part /usr/src/sys into Coda), the last panic seemed to be fixed by
> updating my kernel sources and booting a new kernel. I _am_ happy
> with the SMP work, it rocks. But there are lurking bugs there which
> would give me pause recommending it for production. Over time these
> bugs are being found and fixed, people running SMP and feeding back
> the crashes is the best way to accelerate the process of making NetBSD
> SMP production ready. So, run SMP where it won't hurt you but play
> conservative when your arse is on the line :-)
OK. I want to help. Can you steer me to documentation that explains
how "feeding back the crashes" takes place?
Can you steer me to documentation on how to setup a system to auto
reboot after a panic?
> I agree with Thor on this, NetBSD is probably not going to be a good
> fit for that application much as it pains me to say so. Perhaps you
> should investigate hardening your base Linux platform so that it is
> not so easily hacked.
I'm using Solaris 8 Intel for now.
I would like to see these perceived short comings overcome. I prefer
to spend my extra time on NetBSD rather than Linux. I assume Tomcat 4
is not the high risk area. Oracle is the likely problem area. What
can I do to help advance the cause and ease your pain?
> Brett Lymn