You do not understand me. I understand the client might seem to "freeze" when loading this crappy page - just because it is forced to swap intensively. I also understand the whole pixmap/firefox story - I have read it carefully. Most important is, that eventually my client freezes completely forever (does not even respond to ping) - for me it is a clear indication of a kernel lock. Note I have the X_RAMPERC enabled also to protect X server and to avoid Mr. OOM killer. What I wanted to demonstrate is the vulnerability of NBD swap - it was never intended for this usage (ok it was, and hence the -swap switch to the nbd-client, but it requires kernel patches which are only available to the 2.4 kernel series). Has anyone tried do some stress-testing of NBD-swap? None.
Ondrej Jim Kronebusch wrote: > On Wed, 21 Nov 2007 18:51:27 +0100, Ondrej Valousek wrote > >> Ok, >> to be more exact: >> My client has 512Mb RAM + 1Gb swap. I launch firefox and this page: >> http://www.carteretcountyschools.org/bms/teacherwebs/sdavenport/artgallery6.htm >> makes the machine crash completely. >> Try to replicate. >> Ondrej >> > > It isn't that NBD swap isn't working, it is that the site you reference above > really > sucks. Unfortunately our Junior High art class was searching for art > examples on the > net the first week I deployed our thin client labs. They ran into sites like > this > non-stop and kept freezing all our clients. I used that site as an example > for really > demonstrating the pixmap problem. If you play with xrestop on the client you > can see > that it climbs much higher with the NBD Swap than without. But that site > takes about > 1.5GB to actually fully load, hence the reason for using it as a > demonstration. I > increased my clients to 512MB RAM and increased the NBD Swap from the default > 32MB to > 512MB, this gets us by on most sites. The X_RAMPERC hack kills the browser > before > filling the 1GB available, so sites like the one above still result in a > crashed browser > for me. The RAM usage climbs so fast that monitoring with xrestop will show > the client > freezing before all local RAM and NBD Swap is used, but this is only because > xrestop > doesn't refresh fast enough. If you log into Screen1 on your clients and run > "free" > before loading that site you should see all your RAM and Swap, then switch to > Screen7 > and load the referenced website, then quickly switch back to screen1 and keep > running > "free" over and over again. You should start to see your NBD Swap getting > used before a > hang. > > Again, non-gecko based browsers will handle sites like the one above without > a problem. > I wish I could get the Firefox guys to talk to the Opera guys and learn why > this is. > Hmm, maybe I should join an Opera list and point out their superiority, maybe > they could > shed some light on why this is. > > Hope that helps, > Jim > > The information contained in this e-mail and in any attachments is confidential and is designated solely for the attention of the intended recipient(s). If you are not an intended recipient, you must not use, disclose, copy, distribute or retain this e-mail or any part thereof. If you have received this e-mail in error, please notify the sender by return e-mail and delete all copies of this e-mail from your computer system(s). Please direct any additional queries to: [EMAIL PROTECTED] Thank You. Silicon and Software Systems Limited. Registered in Ireland no. 378073. Registered Office: Whelan House, South County Business Park, Leopardstown, Dublin 18 ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _____________________________________________________________________ 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
