lkml.org 
[lkml]   [2005]   [Jun]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: 2.6.12-rc6-mm1 & 2K lun testing
Andrew Morton wrote:
> Badari Pulavarty <pbadari@us.ibm.com> wrote:
>
>>>Try this:
>>
...
>>
>> Wow !! Reducing the dirty ratios and the above patch did the trick.
>> Instead of 100% sys CPU, now I have only 50% in sys.
>
>
> It shouldn't be necessary to do both. Either the patch or the tuning
> should fix it. Please confirm.
>
> Also please determine whether the deep CFQ queue depth is a problem when
> the VFS tuning/patching is in place.
>
> IOW: let's work out which of these three areas needs to be addressed.
>

Andrew,

Sorry for not getting back earlier. I am running into weird problems.
When running "dd" write tests to 2048 ext3 filesystems, just with your
patch (no dirty ratio or CFS queue depth tuning), I see "buff"
increasing instead of "cache" and I see "bi" instead of "bo".
Whats going on here ?

But files are getting written to and increasing in size. I am
really confused. Why are we reading stuff to buffers and not
writing to cache or disk ?

Thanks,
Badari


procs -----------memory---------- ---swap-- -----io---- --system--
----cpu----
r b swpd free buff cache si so bi bo in cs us
sy id wa
..
2 0 4 6339920 42712 24884 0 0 0 19 413 1237 46
6 48 0
2 0 4 6233728 42748 25364 0 0 0 13 380 732 50
3 47 0
0 0 4 6168192 42784 25328 0 0 0 13 336 485 44
2 54 0
7 914 4 6156672 71520 24456 0 0 1559 15 336 3498 2
13 61 24
18 1525 4 6081296 145752 24528 0 0 3789 13 813 7354 19
36 0 45
7 1843 4 5995024 220824 24276 0 0 3807 13 883 6637
25 47 0 29
6 2046 4 5898740 299228 23788 0 0 3955 13 876 6372
25 60 0 15
13 2046 4 5790588 385156 24548 0 0 4301 13 860 7171 0
59 0 41
13 2044 4 5676452 475736 24784 0 0 4533 0 848 7169 0
69 0 31
18 2044 4 5557836 569840 24592 0 0 4710 0 841 7227 0
74 0 26
procs -----------memory---------- ---swap-- -----io---- --system--
----cpu----
r b swpd free buff cache si so bi bo in cs us
sy id wa
9 2046 4 5438480 665492 24400 0 0 4785 0 833 7244
0 77 0 23
5 2046 4 5318292 761148 24204 0 0 4787 0 835 7152
0 71 0 29
5 2046 4 5197580 857284 24560 0 0 4806 8 838 7340
0 71 0 29
18 2045 4 5078328 953056 24248 0 0 4789 0 833 7565 0
68 0 32
11 2046 4 4958484 1049240 24556 0 0 4809 0 838 7558
0 68 0 32
20 2044 4 4839300 1144704 24552 0 0 4777 0 840 7479
0 67 0 33
10 2045 4 4720740 1239784 24416 0 0 4757 0 842 7446
0 68 0 32
6 2048 4 4602264 1334736 24408 0 0 4755 0 844 7437
0 66 0 34
10 2044 4 4483712 1429272 24300 0 0 4741 0 873 7433
0 68 0 31
12 2044 4 4366552 1523220 24780 0 0 4712 17 933 7431
0 68 0 31
5 2047 4 4248488 1617952 24476 0 0 4753 0 873 7396
0 69 0 31
10 2047 4 4130764 1712128 24212 0 0 4717 0 840 7437
0 66 0 34
9 2046 4 4011812 1807536 24264 0 0 4776 0 839 7402
0 68 0 32
9 2046 4 3892444 1903080 24180 0 0 4781 0 839 7386
0 70 0 30
15 2046 4 3772796 1998880 24872 0 0 4794 5 835 7358
0 72 0 28
5 2047 4 3656624 2092104 24528 0 0 4701 0 838 7358
0 67 0 33
12 2047 4 3547912 2178752 24568 0 0 4392 0 866 7246
0 57 0 43
10 2046 4 3443524 2261348 25048 0 0 4188 16 880 7038
0 53 0 47
7 2045 4 3340784 2342932 24992 0 0 4149 0 868 7132
0 51 0 49
1 2048 4 3239292 2423576 24328 0 0 4097 0 874 7172
0 50 0 50
12 2044 4 3134688 2507120 24892 0 0 4217 0 873 7294
0 54 0 46
procs -----------memory---------- ---swap-- -----io---- --system--
----cpu----
r b swpd free buff cache si so bi bo in cs us
sy id wa
4 2047 4 3026604 2593700 24484 0 0 4349 0 869 7335
0 56 0 44
16 2046 4 2916504 2681900 24520 0 0 4421 0 861 7317
0 58 0 42
2 2047 4 2806032 2770784 24388 0 0 4451 0 865 7326
0 59 0 41
12 2046 4 2694388 2860208 24748 0 0 4481 0 865 7304
0 61 0 39
8 2044 4 2584656 2948112 24564 0 0 4423 0 857 7245
0 61 0 39


-
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: 2005-06-17 17:14    [W:0.123 / U:0.428 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site