On Mon, 20 Nov 2000, Bourdon, Bruce wrote:
> I noticed that the hard drive LED was flickering about once a second while
> in X (witnessed in KDE and GNOME) but goes away if I exit X Windows.
As Peter Cavender pointed out, both GNOME and KDE have components which
monitor removable media to support auto-mounting and auto-run like MS-Windows
does. These will cause drive activity on a regular (every few seconds) basis.
They can also fill up your syslog with "VFS: disk change detected" messages,
which is why I always turn these things off. For GNOME, uninstalling the
"magicdev" package does the job. (You may have to force your package manager
to ignore dependencies (e.g., "rpm --nodeps") to do this.) I dunno how to go
about doing it in KDE.
This brings up another thing which can cause regular disk activity: syslog.
By default, syslogd syncs the log file after every message. If you've got
something (such as magicdev) causing regular log messages to be generated,
that can cause regular disk activity as well. You might try putting a minus
sign (-) in front of some or all file names in your /etc/syslog.conf file (and
restart syslogd). The minus sign will tell syslog not to sync after each log
entry for that file.
> I left the system alone in X for hours Saturday. When I returned the hard
> drive was running continuously and it took about ten seconds for the mouse
> pointer to move in response to physically moving the mouse.
This sounds like something is leaking memory over time. Screen savers are
good at this, as are some window managers (*cough* Enlightenment *cough*).
I've even seen the X server itself grow larger over time. You might try
logging out and back in again; see if that makes a difference.
Some one else pointed out that this might also simply be a scheduled job
kicking in.
> Also, I had Win95 on this box (still got it on a seperate drive) &
> browsing there is MUCH faster than Linux/Netscape.
What browser were you using on Windows? IE? Netscape? IE has a lot of
speed advantages over Netscape on Linux. You might try comparing Netscape on
Windows to Netscape on Linux to get an idea of what is slowing you down.
Another possibility is Mozilla (http://www.mozilla.org). The latest nightly
builds have been doing really well for me. I dunno how well it handles low
memory, though.
On Mon, 20 Nov 2000, Tom Rauschenbach wrote:
> I used KDE with 32 meg for months and while it kinda sucked, it worked.
> And I never had mysterious disk activity or any of these other symptoms.
> Well, that's not completely true. Mouse latency often did become a
> problem when I was seriously abusing my system ...
Something that can greatly help interactive responsiveness on IDE-based
systems is the "-u" switch to hdparm(8). This will allow other interrupts
(e.g., mouse) to occur during IDE I/O. See the man page for "hdparm" for
details, as this switch can interact badly with some hardware combinations.
On Mon, 20 Nov 2000, Derek Martin wrote:
> The only thing I'm running, by the way, is my default gnome session
> (launched via xdm), plus xconsole, 4 xterms, and pine. My window manager
> is Enlightenment.
FYI, if you're running GNOME with low memory, I highly recommend switching
away from Enlightenment to something like Sawfish or even fvwm.
Enlightenment includes a lot of nice features, but "a small memory footprint"
isn't one of them. :-)
On Mon, 20 Nov 2000, Derek Martin wrote:
> XF4 is kinda weird too... my X server claims to have a resident set size
> of 249MB, but I only have 256MB of RAM (the kernel itself is using about
> 4MB according to dmesg). I don't quite understand the mechanics of X
> server's memory usage (and if someone does, please go on at lenght!) plus
> my memory is being reported as only about half in use. So that's
> obviously not quite accurate. The card itself has 32MB on board, so it
> isn't simply playing tricks with that memory... I don't get it.
Most X servers do map the video card memory into their own virtual memory
space, so seeing a very large RSS for an X server is common.
As for why it is 256 MB, I dunno for sure, but it might have something to do
with the virtual memory address the X server is mapping it to. I remember
reading that the kernel doesn't support "discontinuous memory". On the other
hand, it may have nothing to do with that at all. :-)
--
Ben Scott <[EMAIL PROTECTED]>
Net Technologies, Inc. <http://www.ntisys.com>
Voice: (800)905-3049 x18 Fax: (978)499-7839
**********************************************************
To unsubscribe from this list, send mail to
[EMAIL PROTECTED] with the following text in the
*body* (*not* the subject line) of the letter:
unsubscribe gnhlug
**********************************************************
