lkml.org 
[lkml]   [2001]   [Jul]   [13]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
Subjectper-zone inactive shortage information

Hi,

I was doing some tests with the VM stats code and now we can see in
practice how bad the lack of a "per-zone inactivation" scheme in the
current kernel is.

Simple test: 4 setiathome's plus 1 big (1GB) fillmem running on a 900MB
total RAM machine.

Here is one vmstat output from the configuration above. (Ok, its just one,
but this is how the machine looks like most of the time during the test)

r b w swpd free buff cache si so bi bo in cs us sy id
5 1 1 702748 2548 360 600188 46 31060 48 31068 473 2759 50 6 44
launder launder_w ref_inact alloc_r kswapd_w krec_w kflush_w
1 1 16 0 0 1187 18
Zone fshort ishort scan clean skipl skipd launder react rescue
DMA 0 255 0 0 0 0 0 0 0
agescan agedown ageup deact deactf_age deactf_ref recfail ptes pteu
319 319 48 0 319 0 1172 735 688
Zone fshort ishort scan clean skipl skipd launder react rescue
Normal 0 0 32639 5167 5434 11882 7858 2318 0
agescan agedown ageup deact deactf_age deactf_ref recfail ptes pteu
33011 32784 8671 11890 20477 644 0 13447 5061


r b w swpd free buff cache si so bi bo in cs us sy id
4 1 1 702748 2796 360 571668 92 23930 92 23950 394 286 49 3 48
launder launder_w ref_inact alloc_r kswapd_w krec_w kflush_w
1 1 0 129 0 29 3
Zone fshort ishort scan clean skipl skipd launder react rescue
DMA 1 257 0 0 0 0 0 0 0
agescan agedown ageup deact deactf_age deactf_ref recfail ptes pteu
0 0 0 0 0 0 28 0 0
Zone fshort ishort scan clean skipl skipd launder react rescue
Normal 0 0 24492 1305 4836 11230 6007 1117 1
agescan agedown ageup deact deactf_age deactf_ref recfail ptes pteu
0 0 0 0 0 0 129 0 0


The DMA inactive/free shortage does not change for quite some time, while
pages from the normal zone are laundered/freed/pte's deactivated, even if
its OK wrt free/inactive pages. That keeps going until the DMA zone has
its inactive/free shortage gone.

Well, everybody knew that the "non-zoned" behaviour was a theorical
problem.

Now everybody know that its a practical live problem, and a big one.



-
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:0.029 / U:0.016 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site