On Monday 28 June 2004 20:33, Stefan Schumacher wrote:
> * Mihai CHELARU (kefren@netbastards.org) wrote:
> > I'm running a postgresql server on a xeon. All great until I upgraded
> > that machine to a dual-xeon. Now a simple update in a table longs very
> > much (even 2 hours) and postgresql is 100% CPU. During this time I see a
> > lot of IRQ/sec. (about 22000) most of them (21600) are TLB shootdowns.
> > Anyone else is seeing this ?
>
> I'm running Postgres on NetBSD 2.0E i386 with 2 PIII CPUs and have no
> problems updating tables. Did you rebuilt Postgres after switching to
> 2 CPUs?

Yes, I just rebuild it several minutes ago. Same thing.
Just for the record, here is the dmesg: http://gw.netbastards.org/dmesg.pna , 
the kernel config file: http://gw.netbastards.org/config.pna and a systat 
vmstat snapshot: http://gw.netbastards.org/vmstat.txt

Thanks,

-- 
Mihai