On 04/19/2014 05:55 PM, James Knott wrote:

Tom Hardy wrote:

...

I've noticed with the 3.14 kernels and a late 3.13 one, with the above usage, that kswapd0 pegs a CPU for extended periods, as reported by KSysGuard.

I've also noticed that, going back quite some time. IIRC, there's already a bugzilla report on it. Search for it and add your 2¢ worth.

Here's one bug I've added my comments to. https://bugzilla.novell.com/show_bug.cgi?id=832537

I mostly experienced it with an old AMD Athlon 4000+ system & 3 GB of memory, but I've also seen it with my new Intel Core i7 & 8 GB. As with you, the CPU pegs, but there's no swapping or even much disk activity going on.

In that case its unlikely to be traditional virtual memory thrashing, but I wonder if you've hit on a syncromic property of the page queue management algorithm where it locks up somehow, inserting, sorting or something, perhaps a dropped link. There are a number of variables that can be set to alter the VM performance. The trouble is that there are so many of them! Of course it depends on the process set, but you might look at the overcommit settings. -- We know nothing about motivation. All we can do is write books about it. --Peter F. Drucker -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org