Re: Reducing I/O load of logging

From: Michael Fortson <mfortson#gmail.com>
Date: Fri, 13 Feb 2009 14:52:39 -0800


Great suggestions, guys -- thank you.
swapoff -a and flushing the caches didn't seem to help, so I set up a grep to trigger it and ran vmstat while it happened. Here's what it looked like:

http://pastie.org/388711

(same below)

# vmstat 1 20
procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------
 r b swpd free buff cache si so bi bo in cs us sy id wa st
 3 0 0 5812304 30264 1778336 0 0 54 335 0 1 9 3 87  1 0
 2 0 0 5811920 30468 1778936 0 0 4 1796 6978 8867 15 8 77  0 0
 3 0 0 5812408 30508 1778748 0 0 8 420 5927 7629 18 10 72  0 0
 1 1 0 5681840 30668 1908620 0 0 130028 452 7604 10496 13 6 72 9 0
 1 1 0 5550332 30800 2040232 0 0 131720 1172 7764 9461 13 6 71 10 0
 0 2 0 5430436 30928 2159248 0 0 118900 840 8050 10660 11 6 70 13 0
 1 2 0 5315092 31068 2273484 0 0 114616 460 6937 8869 11 5 66 18 0
 3 2 0 5208540 31192 2380296 0 0 106432 564 7668 10066 12 6 65 17 0
 1 2 0 5118892 31288 2469988 0 0 89500 544 7689 9516 12 5 64 19 0
 1 2 0 5028684 31400 2558524 0 0 88844 808 7326 9238 12 5 63 20 0
 1 2 0 4924492 31544 2662840 0 0 103988 628 7456 9810 12 5 63 19 0
 1 2 0 4831736 31648 2756024 0 0 93144 608 7689 9795 12 6 65 18 0
 1 2 0 4719720 31772 2866888 0 0 109676 744 7711 10941 12 6 63 19 0
 1 3 0 4616352 31880 2969904 0 0 102896 292 6922 8493 12 5 61 21 0
 1 2 0 4494184 32020 3090440 0 0 120708 796 6869 8188 12 5 60 22 0
 1 2 0 4369972 32160 3215716 0 0 125060 476 8177 10652 13 6 62 19 0
 1 2 0 4235760 32296 3348232 0 0 133160 416 6612 9239 12 5 63 20 0
 1 2 0 4105792 32436 3478512 0 0 129872 740 6275 8310 12 5 63 19 0
 0 2 0 3988792 32576 3595920 0 0 117792 680 7893 10583 6 5 74 15 0
 1 0 0 3984824 33252 3598744 0 0 1812 4064 7432 10556 11 5 83 1 0
[root#www log]# Received on 2009/02/13 23:52

This archive was generated by hypermail 2.2.0 : 2009/02/14 01:00 CET