-----Original Message----- From: Linux on 390 Port [mailto:linux-...@vm.marist.edu] On Behalf Of Rob van der Heij Sent: Wednesday, April 01, 2009 5:30 PM To: LINUX-390@VM.MARIST.EDU Subject: Re: swappiness & drop_caches ?
On Wed, Apr 1, 2009 at 11:16 PM, r.stricklin <b...@typewritten.org> wrote: > I'm pretty unhappy with the way Linux has been managing memory, > especially w/rt the block caches being allowed to page out process > data. I was hopeful that we could affect some semblance of sane > behavior by twiddling vm.swappiness. Yes, you should be able to, like Marcy experienced. If you can determine the setting. As I understand the motivation for swappiness was the other way around. Desktop users would want to protect some of the cache and swap out some unwanted stuff. > My experience doing so, however, was that it opened us up to > situations where I would start to see processes get pranged by the out- > of-memory desperation kill "feature", even though there was quite a > bit of memory still sitting unused. You mean "unused" as "free" tells you, or what you think would be available when it would give up page cache and/or buffers? If it's in page cache then it could be that it really needed it, or that pages were dirty or not written out fast enough? And you were not playing with CMM? I do recall an issue with CMM where the pages in the balloon were incorrectly believed unused and thus no replenishment was triggered. Rob -- Rob van der Heij Velocity Software http://www.velocitysoftware.com/ ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
