On Wed, Dec 07, 2005 at 10:34:11PM +1100, Nick Piggin wrote:
> Wu Fengguang wrote:
> >A question about the current one:
> >
> >For a NUMA system with N nodes, the way kswapd calculates lru_pages - only
> >sum
> >up local zones - may cause N times more shrinking than a 1-CPU system.
> >
>
> But it is equal pressure for all pools involved in being scaned the
> simplifying assumption is that slab is equally distributed among
> nodes. And yeah, scanning would load up when more than 1 kswapd is
> running.
>
> I had patches to do per-zone inode and dentry slab shrinking ages
> ago, but nobody was interested... so I'm guessing it is a feature :)
I rechecked shrink_dcache_memory()/prune_dcache(), it seems to be operating in
a global manner, which means (conceptually) if 10 nodes each scans 10%, the
global dcache is scanned 100%. Isn't it crazy? ;)
Thanks,
Wu
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
