Sar showing large disk activity for 1 second periods

Hi,

I've been looking at some sar logs on a couple of machines (RH8) and I
keep getting some huge peaks - it looks to me like some sort of bug in
sar, but I don't know if anyone has seen this before or not.

I did sar -p -d at intervals of 1 second and caught it happening when
my await went up to 60M when it is normally about 20. These happen ever
hour or so (but not exactly on the hour, and not tallying up with any
cron process.)

08:31:41 DEV tps rd_sec/s wr_sec/s avgrq-sz avgqu-sz await
svctm %util
08:31:42 cciss/c0d0 3.00 0.00 96.00 32.00 0.03
8.33 8.33 2.50
08:31:43 cciss/c0d0 7.00 8.00 168.00 25.14 0.07
10.14 7.86 5.50
08:31:44 cciss/c0d0 5.00 0.00 168.00 33.60 0.06
12.80 6.60 3.30
08:31:45 cciss/c0d0 4.00 0.00 40.00 10.00 0.03
7.25 5.25 2.10
08:31:46 cciss/c0d0 9.00 0.00 664.00 73.78 0.21
23.67 7.89 7.10
08:31:47 cciss/c0d0 11.00 64.00 96.00 14.55 0.10
9.09 5.27 5.80
08:31:48 cciss/c0d0 36.00 16.00 688.00 19.56 1.23
34.08 5.06 18.20
08:31:49 cciss/c0d0 10.00 0.00 208.00 20.80 0.24
24.40 6.00 6.00
08:31:50 cciss/c0d0 6.00 0.00 168.00 28.00 0.12
20.50 8.83 5.30
08:31:51 cciss/c0d0 5.00 0.00 104.00 20.80 0.07
14.20 7.00 3.50
08:31:52 cciss/c0d0 4.00 0.00 120.00 30.00 0.05
11.75 9.25 3.70
08:31:53 cciss/c0d0 71.00 0.00 912.00 12.85 5.77
60374428.97 2.59 18.40
08:31:54 cciss/c0d0 4.00 0.00 80.00 20.00 0.05
12.25 8.25 3.30
08:31:55 cciss/c0d0 5.00 0.00 136.00 27.20 0.07
14.40 7.40 3.70
08:31:56 cciss/c0d0 8.00 0.00 136.00 17.00 0.19
23.75 6.38 5.10
08:31:57 cciss/c0d0 4.00 0.00 40.00 10.00 0.04
9.25 6.75 2.70
08:31:58 cciss/c0d0 8.26 0.00 300.92 36.44 0.10
12.11 8.56 7.06
08:31:59 cciss/c0d0 132.00 1688.00 160.00 14.00 1.57
5.23 7.01 92.50

Ever seen anything like this before? Any ideas of where I should look?

Thanks,

Keith

.

Relevant Pages

  • Sar showing large disk activity for 1 second periods
    ... I've been looking at some sar logs on a couple of machines (RH8) and I ... keep getting some huge peaks - it looks to me like some sort of bug in ... await svctm %util ...
    (linux.redhat)
  • Re: How to monitor the cpu usage?
    ... >On our HP machines we can do so via top, but it seems its not part of ... But sar is much more informative. ... Get paid for writing about tech: http://aplawrence.com/publish.html ...
    (comp.unix.sco.misc)
  • Re: Red hat 4 SAR multiple cpus
    ... Result from one of our machines (sar -P ALL): ... Is it a Solaris or Linux? ...
    (comp.unix.solaris)
  • Re: Red hat 4 SAR multiple cpus
    ... Result from one of our machines (sar -P ALL): ... Is it a Solaris or Linux? ...
    (comp.unix.solaris)