lkml.org 
[lkml]   [2001]   [Jul]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: dead mem walking ;-)
On Thu, 12 Jul 2001, Marcelo Tosatti wrote:


> > a while before the jobs were submitted i did "readprofile | sort -nr | head -10":
> > 296497 total 0.3442
> > 295348 default_idle 5679.7692
> > 300 __rdtsc_delay 10.7143
> > 215 si_swapinfo 1.2500
> > 138 do_softirq 1.0147
> > 107 printk 0.2816
> > 28 do_wp_page 0.0272
> > 17 schedule 0.0117
> > 10 tcp_get_info 0.0077
> > 10 filemap_nopage 0.0073
> >
> > the same after i was able to kill the jobs (see below):
> >
> > 836552 total 0.9710
> > 458757 default_idle 8822.2500
> > 361961 __get_swap_page 665.3695
> > 6629 si_swapinfo 38.5407
> > 1655 do_anonymous_page 5.3734
> > 760 file_read_actor 3.0645
> > 652 statm_pgd_range 1.6633
> > 592 do_softirq 4.3529
> > 498 skb_copy_bits 0.5845
> > 302 __rdtsc_delay 10.7857
>
>
> Ok, I've seen that before. __get_swap_page() is horribly innefficient.

:-(

> The system is _not_ swaping out data, though. Its just aging the
> pte's and allocating swap.

with that jobs it looks to me that swap allocation shouldn't be
neccessary? total of all pages should have been below the physcial mem
size.


> And that is what is eating the system performance.

does it bring up the load up to 30 and make the machine unusable?
(kswapd was also sometimes in the top-list of CPU hogs, but since i
sorted it by memory...)

> <snip>
>
> Can you please show us the output of /proc/meminfo when the system is
> behaving badly ?

hold on, we set s.th. up....

~dirkw

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 12:57    [W:0.082 / U:0.572 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site