[NTLUG:Discuss] SuSE 9.2 Updates -- LOOKOUT!

[Kevin Brannen]

Stephen Davidson wrote:

> Kevin Brannen wrote:
>
>>
>> Yeh, you don't really have to have ACPI.  You don't need powersaved 
>> either, feel free to kill that process.  I just found out that 
>> gkrellm is for handling themes under gnome; kill that too.  It would 
>> be really interesting to know if one of those was the hog.
>>
>> Too bad you've already reinstalled the system.  But I understand that 
>> real work is far more important than looking at interesting system 
>> configurations. :-)  Afterall, one of them pays the bills...
>>
>> Kevin
>>
> Hi Guys.
>
> More info from misbehaving system.  I wound up having to run some 
> updates, but did not update everything.  Did update powersaved, some 
> KDE stuff that related to power management, and gkrellm.
>
> And that and with a little bit of luck allowed me to chase down where 
> the problem is.  With that update, the 5 sec spike was back.  But it 
> was only to %50 total, and no system lock up.  Increase of time for 
> task processing by about %30.  Not great, but something I can live 
> with, and definately better than the near stalls.  I had not applied 
> the Kernel updates yet, so I think there may be an additional problem 
> in the kernel to drive the spike to 100%.
> Stopping powersaved does not stop a KDE powermonitoring service.  
> Unload the module "battery", and the spike goes away.
>
> Problem: This machine is a laptop.  And I frequently need to run it 
> off of batteries (to the point where I have a spare battery).  I 
> _need_ some kind of battery monitoring service.  Anyways, I am 
> planning on being at the Linux Install Project tommorrow morning, if 
> anyone is interested in having a look.
>
> Regards,
> Steve
>

Very interesting...  So, are you using ACPI?  If so, ditch that and try 
using APM (you may need to recompile your kernel for that, or at least 
you may need an APM kernel module).  I've had problems with ACPI for 
resource monitoring, while APM worked fine; I've heard of problems in 
the reverse direction.  This behavior is *extremely* hardware 
dependent.  Each of the hardware vendors follows the "standards" 
differently. :-(  [or so it seems]

Kevin