Re: More testing reveals

[James v. Z. <ja...@dv...>]

HT P4 3G: 2G RAM, 1x AGP NV, 1x PCI NV

Open source drivers on FC4 xorg 6.8.2-37 are no good for multi-console?
on P4, only a single console would work.

Using text mode console, no console switching issues.

<ctrl><alt><bkspc> stress test :

On secondary (AGP) console _only_: secondary X failed after prolonged
test, then both while attempting to recover - interestingly, it appears
that at the point of failure the X server has accidentally come up on
vt8 instead of (faketty) vt51.

Trying to recover, I eventually took it to runlevel 1 and then
discovered the X session running on vt8 - very slowly - drawing the
login screen and taking mouse and keyboard input. <ctrl><alt><bkspc>
seemed to terminate it, but X failed again on init 5. uptime=0

J


On Sat, 2006-01-07 at 17:24 +1000, James van Zeeland wrote: 
> A bit more stress testing has revealed something interesting.
> 
> On my two console server with 2.6.15-vz1 and nV binaries, I tried
> performing a <ctrl><alt><bkspc> stress test on the secondary display
> _only_.
> 
> No crashes. At all. GDM complains about the X server restarting more
> than 6 times in 90 seconds, pauses for two minutes then continues
> loading X and we can do it all over again.
> 
> Doing the same test with the primary console, after quite a few kills
> will eventually cause a lockup and/or X crash. I inserted a 5 second
> pause using an X wrapper, but results are the same.
> 
> I am going to test again on another machine, this time trying open
> source drivers as well as nV 8178 binaries.
> 
> J
-- 
James van Zeeland <ja...@dv...>