Joe Baker wrote:
Jason Maas wrote:
Hi James,
On Fri, 19 Aug 2005, [EMAIL PROTECTED] wrote:
I can show you a site that will crash ltsp + firefox with 256M ram
on client, reliably, repeatably, and at the same point every time.
Problem description
-------------------
....
The problem didn't happen on all of
http://ww2.lafayette.edu/~collegestore/baby-merchandise.htmlour
terminals though. At first I thought that maybe it was connected to
the video chip in use, which would determine the X driver which might
be buggy. Then I realized that the terminals that could be
consistently crashed all had 64MB of RAM. When I bumped them up to
128MB of RAM then the consistent crashing stopped. Woo Hoo! Now it
seems that the crashing isn't totally gone, but adding RAM *really*
helped.
So Jim, does that sound like the same problem you've seen? Does at
least 256MB of RAM per client seem to make it go away? I'm also
thinking of turning on NFS swap, at least as a stopgap measure.
Anyone else have any experience with this problem and possible
solutions? Any help would be greatly appreciated!
Jason
Hi Jason, (Joe Baker Here)
I loaded up the page with Firefox on a thin client with 64 Meg of ram
and 64 meg of Swap space over NFS. I usually never see the swap space
used. In this case I did see 19 megs of swap space used (the "free"
command will report the memory usage on the thin client). I suspect
it has something to do with the background on the web page. I'm
running 1280x1024 resolution at a 15 bit color depth. My video card
is integrated on the motherboard. I'm running Firefox 1.0.6 on
Mandriva 10.1 with LTSP 4.1.1. Scrolling with Firefox is very slow
(NFS Swapping). Is there any way that we can specify a local hard
disk to be used for swap space on the thin clients? I'm definately
going to increase the swapfile settings to greater than the 64megs
I've been using.
How is the swap over nfs support coming along in the 2.6 kernels? By
the looks of this we are going to need it.
It isn't uncommon for applications to simply crash when enough memory
cannot be allocated. If you have syslog enabled, you might see
something in the loggs about "out of memory" errors. Another instance
where this type of incident may occur is when a user is looking at a
large photo attached to an email message and tries to scroll it.
Jason, in /opt/ltsp/i386/etc/lts.conf can you tell me what the
X_COLOR_DEPTH= setting is for the machine where it crashes?
-Joe Baker
The problem is with the web site design.
Those bozos are putting _HUGE_ images on the page, then adjusting the
size of the image down with the image property tags in the HTML code.
These fxxxxing (I don't use this word loosely on public mailing lists)
images are 1300x1400 in size then scaled down with the web browser. The
webmaster needs to learn how to make smaller image files for publication
on the web. Suggest GIMP or Image Magic.
-Joe Baker
-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.freenode.net
