 | | | I/O Utilization Statistic Strangeness in U5 | I/O Utilization Statistic Strangeness in U5 2005-06-14 - By Tom Sightler
Back
On Tue, 2005-06-14 at 01:18 +0200, Dag Wieers wrote:
> On Mon, 13 Jun 2005, Tom Sightler wrote:
>
> I can confirm the behaviour of iostat, but there has been no change in
> the /proc/partitions format, I verified this with older copies.
Verified with older copies of the kernel, or older copies of iostat?
> The 2 last values in /proc/partitions seems to be resp. increasing and
> decreasing at a constant rate while the disks are not being used. So
> there's a bug in the kernel updating these counters.
Yes, I noticed that too.
> Tom, can you put your info together with this in a bugzilla report (first
> check if there's not already one). That would be appreciated :)
I may not be the best candidate for this as, with a little more
research, I noticed that several of the systems that were showing this
problem were using EMC PowerPath software. Since this involves binary
modules I would think I would get a quick "go away".
What type of systems are you seeing the error on, SCSI, FC, or ATA? I
do have two systems that do not use PowerPath that are also showing the
problem, but they are ATA systems rather than SCSI and are only showing
the issue with the main disk device, not the partitions.
What's very strange is that I've never noticed this on previous kernels.
I guess maybe I just don't remember, or maybe it's a strange interaction
with PowerPath. PowerPath is not certified for the U5 kernel and those
system probably need to be reverted back to the U4 kernel.
I'll research it a little more. I may at least see if I can reproduce
it on the IDE systems with the old kernel, if not, then it is a new bug.
Also, I'll look at some of my other systems and see if they are showing
the problem.
Later,
Tom
--
Taroon-list mailing list
Taroon-list@(protected)
http://www.redhat.com/mailman/listinfo/taroon-list
Earn $52 per hosting referral at Lunarpages.
|
|
|