Vic,
If the system is really idle, then it would be safe and easy to do a
#cp cpu all trace ext 1004 run
command and see if you're getting a hundred of them a second, or just a
dribble.
Mark Post
-----Original Message-----
From: Vic Cross [mailto:[EMAIL PROTECTED]
Sent: Thursday, March 06, 2003 9:34 PM
To: [EMAIL PROTECTED]
Subject: Timer patch on 2.4.19
G'day all,
Just went through the process of building a new 2.4.19 kernel with timer
patch
applied (linux-2.4.19-timer-1-may2002 from DevWorks). I wonder if anyone
has
seen the /proc/sys/kernel/hz_timer pseudofile, and can describe its
function.
Full story, read on...
When building the kernel, I noticed a new option:
[*] No HZ timer ticks in idle
[ ] Idle HZ timer on by default <-- new one
I read this as "do you want to enable the no-HZ function in your kernel, but
turn it off by default at IPL-time". Based on this, I left it as shown
above
for my kernel build (because I wanted the patched function, not the old
function).
After I IPLed, IND Q EXP told me that my idle guest was still constantly
sitting
in Q3. Then I thought about the new kernel option, and theorised that if
there's an option that allows you to set the default operation at IPL time
there's likely to be a way to change it while running...
A brief search yielded /proc/sys/kernel/hz_timer, which was set to 1. I
echoed
0, and the system still worked, but still seemed to be constantly active in
Q3.
Looking at the patch file it would appear that it shoud be 0 to enable the
'no-HZ' function, but it's not clear (to me, at least).
I must confess that for all of my advocacy of the timer patch, this is the
first
time I've used it on a system under VM. So I admit that I'm not sure what I
should be seeing from VM's perspective.
Does /proc/sys/kernel/hz_timer do what I think it does? Am I confusing
myself
looking at IND Q? Will our (to-be-installed-RSN) z/VM monitoring tool tell
me
the real story?
Thanks in advance,
Vic Cross
