[PlanetCCRMA] I had time for a quick test

Fernando Lopez-Lezcano nando@ccrma.Stanford.EDU
Tue Jan 2 13:16:02 2007

On Tue, 2007-01-02 at 04:19 -0800, Tracey Hytry wrote:
> Err, make that a quick test and a long clean up.
> I thought that it would be a good idea to try let the machine run doing a 
> latency/xrun test while I was working outside.  Not a good idea, didn't even 
> make it outside until I cleaned up the mess.
> I tried Ingo's newest(29th) kernels on both fc5 and fc6(i686 for fc5 and 
> x86_64 for fc6).  I modified rtirq and adjusted rtlimits to work with the 
> the new kernels and jack on fc6.
> My first try was on fc5.  It booted fine, rtirq did it's job, and went 
> into X fine.  I got jack running, conservative settings, and began by 
> loading up hydrogen.  I was going to load up aeolus too, so I was playing 
> with the setup to load load hydrogen... AND... total freeze up... 
> motherboard speaker putting out a continuous beeeep.
> So I hit the reset button on the computer and decide to try fc6.
> I remember on an Ingo kernel from about week or two ago that I didn't get 
> lockups like this on the machine.  It's hard to say what was happening, 
> but I was just happy to get the machine working again after that.  I then 
> decided maybe dinner would be better, and then work outside.  If I feel 
> brave in the next few days I may try again, but it was a bit scary there 
> for a while, as the fc5 is what I'm using right now and is my main 
> machine/distro.
> Tracey,  pushing computers to the edge, saying whoops as I nudged them 
> too far toward the edge, and putting the pieces back together again.

Ouch... sorry... could you look at your logs (/var/log/messages*) to see
if there's anything there we could forward to Ingo? Kernel panics,
BUG's, etc, etc...

And I was getting happy because it seemed to work fine at last :-(

I did see a couple of crashes over the long weekend but only after quite
a bit of uptime (and all apps seemed to be happy). In one of them
apparently the system run out of memory and the oom killer triggered and
left the machine in an unusable state (I was logging in remotely), the
other left no clues behind - these were on a fc4 system with the new
kernel so there could be other userland issues other than the kernel. 

-- Fernando