<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
On 20/11/08 13:21, Alain Williams wrote:
<blockquote cite="mid:20081120132125.GK18695@mint.phcomp.co.uk"
type="cite">
<pre wrap="">On Thu, Nov 20, 2008 at 12:53:07PM +0000, Luke Dudney wrote:
</pre>
<blockquote type="cite">
<pre wrap="">All of that output refers to stats since boot. For real-time info try
"iostat -x 5" and wait 5 seconds for the second lot of output
</pre>
</blockquote>
<pre wrap=""><!---->
Errrm: I said that I had removed the first block of iostat output. No matter.
</pre>
</blockquote>
My bad, I misunderstood.<br>
<blockquote cite="mid:20081120132125.GK18695@mint.phcomp.co.uk"
type="cite">
<pre wrap=""></pre>
<blockquote type="cite">
<pre wrap="">First thing to check would be to see if any processes are blocking on IO:
while true; do ps auwx |awk '$8 ~ /D/'; sleep 1; done
</pre>
</blockquote>
<pre wrap=""><!---->
That was interesting. I ran that & walked away for 5 minutes. sda & sdb are SATA disks
that are mirrored (raid 1) providing md1. mdadm shows no problem with it and
nagios would have told me if there was a problem.
I must confess to not understanding how the kernel raid works; is it perhaps writing
status blocks back to disk ?
# while true; do ps auwx |awk '$8 ~ /D/'; sleep 1; done
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
ups 3426 0.0 0.0 37820 684 ? Ds Nov17 0:02 //bin/usbhid-ups -a mintups
root 1908 0.0 0.0 0 0 ? D< Nov17 0:51 [kjournald]
root 4008 0.0 0.0 10188 660 ? D Nov17 0:25 hald-addon-storage: polling /dev/hda
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 477 0.0 0.0 0 0 ? D< Nov17 0:04 [kjournald]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 1889 0.0 0.0 0 0 ? D< Nov17 0:00 [kjournald]
exim 6432 0.3 1.3 168364 55080 ? D 09:46 0:38 spamd child
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 4008 0.0 0.0 10188 660 ? D Nov17 0:25 hald-addon-storage: polling /dev/hda
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 1884 0.0 0.0 0 0 ? D< Nov17 0:00 [kjournald]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 470 0.0 0.0 0 0 ? D< Nov17 2:12 [md1_raid1]
root 4008 0.0 0.0 10188 660 ? D Nov17 0:25 hald-addon-storage: polling /dev/hda
</pre>
</blockquote>
Having not used it I'm not sure how the kernel raid works either; it
may be expected behaviour for those raid1 kernel threads to sit there
blocking the whole time without actually doing much.<br>
<br>
Could your exim + spamassin be the culprit? Is there much happening in
the mail logs? <br>
<br>
Cheers<br>
Luke<br>
</body>
</html>