mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Stefan Schmidt <zaphodb@zaphods.net>
To: Marcelo Tosatti <marcelo.tosatti@cyclades.com>
Cc: linux-kernel@vger.kernel.org
Subject: Re: Kernel 2.6.9 Multiple Page Allocation Failures (Part 2)
Date: Thu, 4 Nov 2004 19:18:57 +0100	[thread overview]
Message-ID: <20041104181856.GE28163@zaphods.net> (raw)
In-Reply-To: <20041104121722.GB8537@logos.cnet>

On Thu, Nov 04, 2004 at 10:17:22AM -0200, Marcelo Tosatti wrote:
> What is min_free_kbytes default on your machine?
I think it was 768, definitely around 700-800.
2.6.9 said:
Dentry cache hash table entries: 131072 (order: 7, 524288 bytes)
Inode-cache hash table entries: 65536 (order: 6, 262144 bytes)
Memory: 4058200k/4095936k available (2005k kernel code, 36816k reserved, 995k
data, 196k init, 3178432k highmem)

> > I tried the following kernels: 2.6.9-mm1, 2.6.10-rc1-bk12, 2.6.9-rc3-bk6,
> > 2.6.9-rc3-bk5 all of which froze at some point presenting me only with the
> > above page allocation failure. (no more sysrq) 
> This should be harmless as Andrew said - it would be helpful if you could 
> plug a serial cable to the box - this last oops on the picture doesnt say 
> much.
Well right now the machine is running 2.4.28-rc1 with the 3w-9nnn patch by
Adam Radford from this list and i would like to see it run stable for about a
day before i give 2.6 another try. I think i'll have a terminal server hooked
up by then.

> How intense is the network traffic you're generating?
I was around 60-80 mbit/s each direction at i think 16k interrupts/s.

With 2.4.28-rc1 this is currently at 180mbit/s 27kpps up, 116mbit/s 24kpps down 
still swapping a bit but no kernel messages on this, just around 1.7 rx
errors/s.

> 2.6.7 was stable under the same load?
No, sorry to give you this impression, 2.6.7 is just what some of my collegues
and i consider the more stable 2.6 kernel under heavy i/o load.

> Something is definately screwed, and there are quite an amount of 
> similar reports.
Can i tell people its ok to see nf_hook_slow in the stack trace as it's
vm-related? A collegue was quite bluffed when i showed him. ;)

> XFS also seems to be very memory hungry...
I have 8 XFS-Filesystems in use here with several thousand files from some k to
your 'usual' 4GB DVD-image. XFS built as a module at first and then inline but
that did not change anything off course. 2x200 + 6x250GB that is.

	Stefan

  reply	other threads:[~2004-11-04 18:21 UTC|newest]

Thread overview: 70+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-11-03 22:24 Stefan Schmidt
2004-11-04 12:17 ` Marcelo Tosatti
2004-11-04 18:18   ` Stefan Schmidt [this message]
2004-11-09 16:41     ` Kernel 2.6.9 Multiple Page Allocation Failures Marcelo Tosatti
2004-11-09 22:35       ` Lukas Hejtmanek
2004-11-09 22:46         ` Andrew Morton
2004-11-09 22:44           ` Lukas Hejtmanek
2004-11-09 20:33             ` Marcelo Tosatti
2004-11-10 20:35               ` Lukas Hejtmanek
2004-11-10 21:09                 ` Andrew Morton
2004-11-10 21:24                   ` Lukas Hejtmanek
2004-11-10 21:47                     ` Andrew Morton
2004-11-10 21:28               ` Lukas Hejtmanek
2004-11-10 18:11                 ` Marcelo Tosatti
2004-11-11  1:04                   ` Lukas Hejtmanek
2004-11-11 21:44                   ` Lukas Hejtmanek
2004-11-12 12:09                     ` Nick Piggin
2004-11-13 14:47                       ` Stefan Schmidt
2004-11-16  9:33                         ` Marcelo Tosatti
2004-11-16 17:05                           ` Lukas Hejtmanek
2004-11-21  1:43                             ` Stefan Schmidt
2004-11-21  2:42                               ` Stefan Schmidt
2004-12-02 19:54                                 ` Lukas Hejtmanek
2004-12-02 20:25                                   ` Andrew Morton
2004-12-02 21:03                                     ` Lukas Hejtmanek
2004-12-02 22:31                                       ` Stefan Schmidt
2004-12-02 22:48                                         ` Lukas Hejtmanek
2004-12-02 22:56                                         ` Andrew Morton
2004-12-02 23:18                                           ` Lukas Hejtmanek
2004-12-03  0:18                                             ` Andrew Morton
2004-12-03 12:11                                               ` Lukas Hejtmanek
2004-12-03 12:17                                                 ` Lukas Hejtmanek
2004-12-07 22:52                                                 ` Nick Piggin
2004-12-07 22:59                                                   ` Lukas Hejtmanek
2004-12-07 23:05                                                     ` Nick Piggin
2004-12-08 11:18                                                       ` Lukas Hejtmanek
2004-12-08 11:23                                                         ` Nick Piggin
2004-12-08 11:46                                                           ` Lukas Hejtmanek
2004-12-08 13:14                                                           ` Lukas Hejtmanek
2004-12-09  8:52                                                             ` Nick Piggin
2004-12-09  9:02                                                               ` Lukas Hejtmanek
2004-12-09 10:29                                                                 ` Nick Piggin
2004-12-09 10:37                                                                   ` Lukas Hejtmanek
2004-12-03  6:18                                           ` Nathan Scott
2004-12-03  7:06                                             ` Andrew Morton
2004-12-07 11:17                                             ` Lukas Hejtmanek
2004-12-08  0:15                                               ` Nathan Scott
2004-12-08  0:36                                                 ` Lukas Hejtmanek
2004-12-03 10:35                                           ` Christoph Hellwig
2004-12-03 10:58                                     ` P
2004-12-03 17:11                                       ` Andrew Morton
2004-11-09 23:52           ` Stefan Schmidt
2004-11-10  1:27             ` 2.6.10-rc1-mm4 -1 EAGAIN after allocation failure was: " Stefan Schmidt
2004-11-10  1:39               ` Andrew Morton
2004-11-10  2:03                 ` Stefan Schmidt
2004-11-10  2:21                   ` Andrew Morton
2004-11-10  4:24                   ` Nick Piggin
2004-11-10 10:28                     ` Stefan Schmidt
2004-11-10 12:06                       ` Stefan Schmidt
2004-11-10  8:58                         ` Marcelo Tosatti
2004-11-10 12:48                           ` Stefan Schmidt
2004-11-10 10:56                             ` Marcelo Tosatti
2004-11-11  1:23                             ` Nick Piggin
2004-11-11 18:31                               ` jhigdon
  -- strict thread matches above, loose matches on Subject: below --
2004-10-25 10:48 Kernel 2.6.9 Page Allocation Failures w/TSO+rollup.patch Justin Piszcz
2004-10-25 11:33 ` Nick Piggin
2004-10-27 21:40   ` Kernel 2.6.9 Multiple Page Allocation Failures (Part 2) Justin Piszcz
2004-10-27 21:58     ` Andrew Morton
2004-10-27 21:55       ` Justin Piszcz
2004-10-27 22:19         ` Francois Romieu
2004-10-27 22:23           ` Justin Piszcz
2004-10-28  0:33     ` Lee Revell

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=20041104181856.GE28163@zaphods.net \
    --to=zaphodb@zaphods.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=marcelo.tosatti@cyclades.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®