From: Mark Hahn <hahn@physics.mcmaster.ca>
To: Mickael Bailly <mickael.bailly@telintrans.fr>
Cc: linux-mm@kvack.org
Subject: Re: looking for explanations on linux memory management
Date: Wed, 26 Nov 2003 14:01:22 -0500 (EST) [thread overview]
Message-ID: <Pine.LNX.4.44.0311261351220.18209-100000@coffee.psychology.mcmaster.ca> (raw)
In-Reply-To: <200311261830.02711.mickael.bailly@telintrans.fr>
> 1/ Hardware
> We are working on SMP 2.8GHz Pentium Xeon Hyperthread
only relevant in the sense that there are some artifacts of 32b-ness
(highmem, etc).
> We have between 2 and 2.5 gigas of RAM.
what does /proc/meminfo look like?
> 2/ Kernel
> We are using RedHat kernel 2.4.20-20.7 (last kernel upgrade for RedHat 7.3)
you should seriously consider running a modern kernel.org kernel.
RH doesn't have all that much in the way of special magic that they put
in their kernels, and you can make one yourself that's at least as good.
> Memory usage:
> In the attached graph you can see last month memory usage for this host.
it's remarkable that you have so much wasted memory!
> 1/ can you explain me what happened in week 47 so cached memory don't get down
> anymore ? Nothing really changed in this week on the server.
I'm guessing someone did a "find /" or similar, which caused lots of
dcache/icache entries to be created. of course, it could also be normal
cached file pages, stale SHM segments (run ipcs -a), or maybe even
a big-VM proces that's gotten into some limbo state...
> 2/ how can I know when my server needs more RAM/SWAP, if free memory is always
> about 0
free memory is WASTED memory - you might as well have not bought it.
you know you need more memory when you see swapin traffic (NOT swapouts,
which are normal and in fact good). swapins are a sign that the kernel
has either chosen the wrong pages to swap out, or is needing to swap out
so much that hot pages are getting swapped, or that you simply have a
working set that's larger than physical memory.
> 3/ can you tell me where to find PER PROCESS memory usage (/proc/[process
> id]/stat ? /proc/[process id]/statm ? )
why not just run top or ps? they both reformat info from /proc/<pid>
to make it easier to read.
regards, mark hahn.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>
next prev parent reply other threads:[~2003-11-26 19:01 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-11-26 17:30 Mickael Bailly
2003-11-26 18:13 ` Rob Love
2003-11-26 19:01 ` Mark Hahn [this message]
2003-11-27 14:53 ` Mickael Bailly
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=Pine.LNX.4.44.0311261351220.18209-100000@coffee.psychology.mcmaster.ca \
--to=hahn@physics.mcmaster.ca \
--cc=linux-mm@kvack.org \
--cc=mickael.bailly@telintrans.fr \
/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