Re: Strange load issue on NetBSD

[vanja%vanja.com@localhost]

Thanks George.

I've just noticed this in dmesg output:

file: table is full - increase kern.maxfiles or MAXFILES
<repeated many times>

netbsd: {23} sysctl -a | grep maxfiles
kern.maxfiles = 3404
netbsd: {24} 

I'm not used to NetBSD, so I'll need to figure this one out (read documentation 
:). But I presume it could be 
related to the load issues I am seeing.

Thanks.


On Mon, May 04, 2009 at 01:12:57PM -0700, George Georgalis wrote:
> On Mon 04 May 2009 at 09:37:53 PM +0200, x25%pobox.com@localhost wrote:
> >At the place where I work, I am maintaining mostly Linux servers. However, 
> >turns out there is 1 NetBSD (2.0, 
> >ouch) box that is running Apache.
> >
> >Other admin reported httpd processes there turning into zombies, so I wanted 
> >to take a look.
> >
> >I am puzzled. Average load on the machine is displayed in ranges from 3 to 
> >10, yet top and all other types of 
> >'performance monitoring' tools (vmstat, iostat) show that machine is 99% 
> >idle. No processes using any CPU 
> >time, and no 'major' activity.
> >
> >Example:
> >
> >netbsd: {5} w
> >12:33PM  up 305 days,  1:04, 2 users, load averages: 5.53, 5.50, 5.25
> >USER    TTY FROM                                 LOGIN@  IDLE WHAT
> >username p1 x.x.x.x                             12:26PM     6 w
> >
> >Box hasn't been rebooted for a while, so that is something I'll be looking 
> >at doing soon, but I was wondering 
> >if anyone has ever experienced this behaviour.
> 
> failing disk or other hardware fault (memory) could cause that.
> 
> depending on how important it is, I'd have some replacement plan in place.
> 
> 
> -- 
> George Georgalis <george%galis.org@localhost>, Information Systems <IXOYE><
> SEEKING EMPLOYMENT OPPORTUNITIES, specialties: BSD/Linux, HPC,
> security/VPN, data warehousing, virtualization, validation,
> decision tree analysis, unit testing and change control.