Re: 2.6.14-rc1 load average calculation broken?

[Jan Dittmer <jdittmer@ppp0.net>]

Jan Knutar wrote:
> On Thursday 15 September 2005 15:40, Jan Dittmer wrote:
> 
>>Jan Dittmer wrote:
>>
>>>Get a steady 2.00 there. I stopped unnecessary processes etc.
>>>load average seems to be invariant
>>>
>>>top - 13:41:32 up  4:44,  2 users,  load average: 2.00, 2.00, 2.00
>>>Tasks: 108 total,   2 running, 105 sleeping,   0 stopped,   1 zombie
>>>Cpu(s):  0.0% us,  0.0% sy,  0.0% ni, 99.7% id,  0.3% wa,  0.0% hi,  0.0% si
>>
>>Hmm, reboot to 2.6.14-rc1-git1 cured it. Will see if it happens again.
>>(btw. it was not invariant but the lower limit was 2 even after stopping
>>everything but some essential processes (ssh, init, getty))
> 
> 
> Did you check with ps aux, if there were processes stuck in D state? Those
> count towards the load average...

No, I didn't explicitly. But I killed nearly every process but the kernel
threads and it didn't change. Shortly before rebooting the load even rised
again without me doing anything - very creepy (I thought about a rootkit but
the machine just has a NATed ssh port open to the internet on a non-standard
port).
uptime is now 6:32 and it has not happened again.

Jan