From: Aaron Porter <atporter@primate.net>
To: Chris Snook <csnook@redhat.com>, linux-kernel@vger.kernel.org
Subject: Re: miserable performance of 2.6.21 under network load
Date: Tue, 24 Jul 2007 11:53:51 -0700 [thread overview]
Message-ID: <20070724185351.GE24543@primate.net> (raw)
In-Reply-To: <46A649CB.7080401@redhat.com>
On Tue, Jul 24, 2007 at 02:49:47PM -0400, Chris Snook wrote:
> Is it always CPU 0, or does it move? Are you running irqbalance? If
> you're running irqbalance, you can run a script that alternates between
> 'cat /proc/interrupts' and 'mpstat -P ALL 5 10' and watch the offending
> interrupt jump around between processors. It's not as informative as
> oprofile, as Andi suggested, but it's really easy to set up.
Always cpu0
# cat /proc/interrupts
CPU0 CPU1 CPU2 CPU3
0: 166 0 0 0 XT-PIC-XT timer
1: 8 0 0 0 XT-PIC-XT i8042
2: 0 0 0 0 XT-PIC-XT cascade
3: 11 0 0 0 XT-PIC-XT serial
4: 830 0 0 0 XT-PIC-XT serial
5: 370787109 0 0 0 XT-PIC-XT eth0
6: 3 0 0 0 XT-PIC-XT floppy
8: 4 0 0 0 XT-PIC-XT rtc
9: 0 0 0 0 XT-PIC-XT ohci_hcd:usb1, ohci_hcd:usb2
10: 80937699 0 0 0 XT-PIC-XT 3w-9xxx, 3w-9xxx
12: 5 0 0 0 XT-PIC-XT i8042
14: 50931 0 0 0 XT-PIC-XT ide0
NMI: 1110304 275156 276732 233082
LOC: 278410000 278410022 278410013 278410014
ERR: 55003177
MIS: 0
# mpstat -P ALL 5 1
Linux 2.6.21.5 (*.com) 07/24/07
11:52:30 CPU %user %nice %system %iowait %irq %soft %idle intr/s
11:52:35 all 0.40 0.00 16.12 13.56 0.25 25.13 44.54 299.80
11:52:35 0 0.00 0.00 0.00 0.00 1.00 99.00 0.00 299.80
11:52:35 1 0.80 0.00 23.65 23.05 0.00 0.40 52.10 0.00
11:52:35 2 0.40 0.00 21.64 14.23 0.00 0.40 63.53 0.00
11:52:35 3 0.60 0.00 19.44 17.03 0.00 0.60 62.73 0.00
Average: CPU %user %nice %system %iowait %irq %soft %idle intr/s
Average: all 0.40 0.00 16.12 13.56 0.25 25.13 44.54 299.80
Average: 0 0.00 0.00 0.00 0.00 1.00 99.00 0.00 299.80
Average: 1 0.80 0.00 23.65 23.05 0.00 0.40 52.10 0.00
Average: 2 0.40 0.00 21.64 14.23 0.00 0.40 63.53 0.00
Average: 3 0.60 0.00 19.44 17.03 0.00 0.60 62.73 0.00
#
prev parent reply other threads:[~2007-07-24 18:54 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-07-24 17:25 miserable performance of 2.6.21 under network load Aaron Porter
2007-07-24 18:48 ` Andi Kleen
2007-07-24 18:49 ` Aaron Porter
2007-07-24 19:13 ` Jeff Garzik
2007-07-24 19:13 ` Andi Kleen
2007-07-24 20:41 ` Aaron Porter
2007-07-24 19:22 ` Chris Snook
2007-07-24 19:45 ` David Miller
2007-07-24 18:49 ` Chris Snook
2007-07-24 18:53 ` Aaron Porter [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20070724185351.GE24543@primate.net \
--to=atporter@primate.net \
--cc=csnook@redhat.com \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.