On Saturday, February 12, 2011 12:27:55 pm Rick Sewill wrote:
On Saturday, February 12, 2011 12:09:34 pm M. Fioretti wrote:
On Sat, Feb 12, 2011 19:03:56 PM +0100, Marco Fioretti
(mfioretti@nexaima.net) wrote:
On Sat, Feb 12, 2011 12:55:16 PM -0500, Lamar Owen (lowen@pari.edu)
wrote:
On Saturday, February 12, 2011 12:19:33 pm M. Fioretti wrote:
besides hard drive and DVD burner there are only Logitech webcam, wheelmouse and earphone microphone, but everything is plugged in the back which is not really accessible without moving furniture. I'll do that if needed, but isn't a way to check for those interrupts from the prompt?
Let's see if iowaits are you issue. Install the sysstat package (yum install sysstat) and run: iostat -x 1
here it is, thanks for the tip. When it isn't zero, the await column gives anything from 27.36 to 35.78 (last line) to 5 (I have already posted top output in a comment to the web page):
[root@polaris ~]# iostat -x 1 | egrep -i 'device|sda'
Sorry, of course that's only the part of the story about sda. here is one complete run of iostat:
Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util sda 0.00 5.00 0.00 5.00 0.00 64.00 12.80 0.03 6.00 6.00 3.00 dm-0
0.00 0.00 0.00 8.00 0.00 64.00 8.00 0.044.38 3.75 3.00 dm-1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
other runs show all null values for dm-0 / dm-1, or values similar to these
Marco
Could you show the output of iostat -x 1, not iostat -x 1 | egrep -i 'device|sda' please?
On my system, when I do iostat -x 1 I get "avg-cpu" besides drive information. avg-cpu: %user %nice %system %iowait %steal %idle 5.05 0.00 4.04 0.00 0.00 90.91
Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util sda 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
It might help to see the "avg-cpu". If we are lucky, either the %user or %system or ... will show high cpu usage.
Another question please...if it's spurious interrupts, I found the device file, /proc/interrupts, which has a row for Spurious interrupts.
We haven't demonstrated the problem is interrupt related. Can we try to isolate or rule out this as a problem please?
Could you show us the output of twice, the second time a few seconds after the first time so we can see if any interrupt number changes fast. more /proc/interrupts ... more /proc/interrupts
Can people suggest any information/files in /proc which might help us?
I assume there is a periodic hardware clock interrupt for your CPU. Can we find out this clock interrupt rate somewhere?