I for one know what Ben is talking about.(seeing as still test kernel i am cuting some slack) But here what happens with one my machine and still with 2.4.14 I have 768 MB RAM (upgrade from 512 MB) and the box sitting doing nothing much. Out of nowhere it's start swaping. Free tell me that 500 MB free of ram and swaping like crazy. And for record the VM issues as Ben mentionned is reported bug in previous builds. It still happens but as bad at least for me. One thing kernel seem to have solved Shutdown issues i had before.(That make me happy at leat) Regards, Alex :) Lee O'Malley wrote:

On Monday 12 November 2001 07:02 am, Mads Martin Joergensen, went on about:

...

* Ben Rosenberg [Nov 12. 2001 00:53]:

...

Well, this 2.4.14 kernel I got didn't solve the issue. It still shoves data into swap when there is more then enough RAM to be used. I guess I'll just get 3 512M dimms and keep swap off perminantly. If this is the

But that would be like pissing in your pants on a cold winterday. Might get you nice and warm in the beginning :-)

...

new VM that Andrea and Linus came up with..it's shite.

Well, if you think about it for a second, you know that is not the case. On LKML there is only very positive reports about the new VM, and I've yet to see anyone with the same problems as you report. Something is really weird--my machine have been running pure Linux -pre kernels since .4.11-pre and I never see the swap problems you do.

---

Mads, I must agree with you here. I can't say that I have ever experienced the problems with swap that Ben complains of on his machine/setup. I am using the Mantel 2.4.13 kernel now without problems and never seen such things with 2.4.4 or 2.4.12 either and I looked very hard for them after reading Ben's complaints. I suspect he has a problem with something else on his machine. I hope he figures it out sometime, but I don't think he can blame it on the kernel or new VM at this time.

O'Malley -- ---KMail 1.3.1--- SuSE Linux v7.2--- Registered Linux User #225206 /tracerb@sprintmail.com/ *Magic Page Products* *Team Amiga* http://home.sprintmail.com/~tracerb

-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com

I have done my share of testing i reached same result no not DIMM problem. I have machine also that just uprade to test kernel 2.4.14 with no X and still occurs only. That machine only 384 MB RAM and the only if run's Qpopper and get 40-50 MB swap usage from time to time.(Talk about strange) I keep trying to figure it out and figured it out will let know. If figure yourself please let me know. Strange thing my RH with vanilla kernel works great.(RH 7.1 + 2.4.14 +512 MB RAM,GNOME, XF 4.1.0) Regards, Alex Ben Rosenberg wrote:

* SuSe Mailling List subscriber (suse@linuxnews.dyndns.org) [011112 12:44]: ->I for one know what Ben is talking about.(seeing as still test kernel i ->am cuting some slack) ->But here what happens with one my machine and still with 2.4.14 ->I have 768 MB RAM (upgrade from 512 MB) and the box sitting doing ->nothing much. ->Out of nowhere it's start swaping. ->Free tell me that 500 MB free of ram and swaping like crazy. -> ->And for record the VM issues as Ben mentionned is reported bug in ->previous builds. -> ->It still happens but as bad at least for me. -> ->One thing kernel seem to have solved Shutdown issues i had before.(That ->make me happy at leat) ->

I've run memtest86 on this machine..for 8hours in fact. No errors reported. I ran the most recent version of AMIDiags on it. No errors reported. I've run both of these on my OEM Compaq Deskpro at work because it was having weird issues as well. I simply don't know what could be the cause. It could be a mixture of KDE2.2.1 + nVidia + newer kernels...I just don't know. I guess I'll just stop talking about it. It seems like 5% of the SLE is having issue with it. So I'll just deal.

-----=====-----=====-----=====-----=====----- Ben Rosenberg mailto:ben@whack.org -----=====-----=====-----=====-----=====----- "Technological progress is like an axe in the hands of a pathological criminal" -AE

-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com

To the original poster: For shits and giggles mostly, try the Robert Love's pre-emptive linux kernel patches from http://www.tech9.net/rml/linux/ . It is supposedly faster for desktop machines when throughput is not much of an issue. If it is faster and still swaps, we can all assume John Scott's wisdom is truth. BTW, AA's VM was chosen (in part, of course) because of its speed over Rik's. If a little swap is used for that speed increase, then hell, so be it. Also, try disabling swap altogether. Do you have a speed incrase or decrease? That will probably answer your question (in part, at least). On 13 Nov 2001, John Scott wrote:

If I read the original complaint correctly, it wasn't about bad performance, it was about the fact that "some" swapping was taking place. The logic here is that if one has enough RAM, one has no need of a swap file/partition. That is incorrect. The OS (and not just linux) is designed to swap less frequently used data to the vm. That is a proven performance enhancer. In fact, it reduces the need for adding more RAM. You can put in 4GB of RAM and the system will still do some swapping. It's a waste of money adding RAM just to see the HD stop doing what it was meant to do. In addition, the vm keeps a table of the contents of every single storage location in vm whether empty or not and it is constantly updated. Sometimes that may be all that's being written (although most likely not).

It is not uncommon to see 138MB of my 256MB of RAM being filled, with the rest free, at the same time my swap file has 40MB of 256 filled. If that other 40MB is placed back in physical memory and then I start a memory intensive program, it has to be swapped back out. Linux (and Windows too for that matter) takes that data that is not in demand and swaps it out so that when your program needs the other memory, it's free already and its instructions can be loaded directly. Again, this is a normal performance enhancing feature of modern OS's. You can't defeat it and what's more, you shouldn't want to.

-- Karol Pietrzak PGP KeyID: 3A1446A0

--- Karol Pietrzak wrote:

To the original poster:

Ben?

For shits and giggles mostly, try the Robert Love's pre-emptive linux kernel patches from http://www.tech9.net/rml/linux/ .

I love this patch. Starting up large-ish applications like Mozilla does see a time decrease, though not exactly night-and-day. I'd compare this performance increase to something along the lines of going from 2.2.16 to 2.4.1 or so; you can feel it, if not actually measure it. It's supposed to decrease latency for things like real-time audio processing, which I hope translates into an increase for real-time video processing, too (make sense, doesn't it?). Applying this patch for a file, mail, or web server may not be the best idea one ever had, but for a workstation, it's *definitely* worth the few-kay download and the time to run patch. ===== -- -=|JP|=- Hit me! - http://www.xanga.com/cowboydren/ Jon Pennington | Debian 2.3 -o) cowboydren @ yahoo . com | Auto Enthusiast /\\ Kansas City, MO, USA | ICQ UIN 69 67 29 31 _\_V __________________________________________________ Do You Yahoo!? Find a job, post your resume. http://careers.yahoo.com

--- Mads Martin Joergensen wrote: [Robert Love's Preemptable Kernel Patch]

It's probably going in early 2.5 too

This is interesting, to say the least. I'm not going to say "nay" here, but ... well, okay. :) ===== -- -=|JP|=- Hit me! - http://www.xanga.com/cowboydren/ Jon Pennington | Debian 2.3 -o) cowboydren @ yahoo . com | Auto Enthusiast /\\ Kansas City, MO, USA | ICQ UIN 69 67 29 31 _\_V __________________________________________________ Do You Yahoo!? Find the one for you at Yahoo! Personals http://personals.yahoo.com

At 04:13 AM 11/14/01, you wrote:

* Jon Pennington [Nov 13. 2001 22:55]:

...

This is interesting, to say the least. I'm not going to say "nay" here, but ... well, okay. :)

Please elabortate.

I'm just thinking that you'd have to keep it an option and mark it clearly with BIG RED FLAGS so that people trying to build a kernel for a server would not "accidentally" enable it. It's a good idea for desktops, but not a good idea for servers. _________________________________________________________ Do You Yahoo!? Get your free @yahoo.com address at http://mail.yahoo.com

You do know that the .5 in 2.5.x is an uneven number which means it's the development kernel tree, not intended for any serious use except development. It will also be capable of eating your breakfast, seduce your wife, put sand in your coffee and as a final touch you can expect it to shave off one of your eyebrowes while sleeping.

I hope they'll set "EAT_BREAKFAST=n" by default. That would really ruin my day...