Dans l'épisode précédent (Thu, Jun 23 2005 - 20:12), Michael nous apprenait que :
> Helo,
> 
> Hello,
> 
> >  Although those machines ended in my room because they "looked" dead
> >  (after savage electric shutdown and such), they worked pretty well
> >  with Solaris 9 and OpenBSD 3.6. In the same time, using NetBSD (2.0.2
> >  or 3.0_BETA), the machines freeze rather quickly on pkgsrc build
> >  process.
> 
> Hmm, I didn't have any freeze for quite some time ( U10/333, Creator3D )

 yeah me too with my home machine.
 only named&clamav dies very often but I have an automatic restart in
 cron so that's ok.

> and I did a LOT of compiling. Most freezes ( maybe all ) I've ever had
> were kernel panics with X running ( I should really add something to
> force the console back into 'text mode' at least for cg6 and ffb
> graphics boards where we never change resolution ). Applications crash
> occasionally but Martin found a likely cause for most of these crashes.
> ( ok, I'm running -current. It would be interesting to know it you still
> get freezes with that )
> 

 why not... I'll try 200506190000Z as there is a binary sets ready.

> >  I thought, Sun released quite "standard" machine (compared to DELL,
> >  for eg, where a single server ref might contain various ethernet card
> >  and such). I mean, an U5/333 is 100% identical to another U5/333.
> 
> Yes, I'd assume the mainboard that can take the faster CPU modules would
> be the one with Rage Pro and the slower ones would have a Rage II, but
> I'm not too sure about that. The rest should be more or less the same
> though.
> 
> >  All that to say, I'm wondering if it's possible that NetBSD/sparc64
> >  deal worse with some "small" hardware issues than Sol|OpenBSD do.
> 
> It's certainly possible but that should better b answered by someone who
> knows more about the lowlevel gruntwork in the kernel than I do.
> 
> > NetBSD locks (no gdb, nothing on console, nothing on screen)
> >  but the other OS don't say anything special.
> 
> As I said - I didn't see anything like that on my U10. On the other hand
> it would be pretty unlikely that all your boxes have exactly the same
> hardware problem.
> 

 the only thing that they have in common is the RAM (because those are
 spare machines where the ram have been taken to put in the production
 servers). maybe the ram is faulty and makes netbsd freeze - but I don't
 have spare module to test.

-- 
,-- This mail runs ---------.
`------------ NetBSD/i386 --'