[Bug 290207] [ZFS] lowering "vfs.zfs.arc.max" to a low value causes kernel threads of "arc_evict" to use 91% CPU and disks to wait. System gets unresponsive...

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=290207

--- Comment #22 from Mark Millard <marklmi26-fbsd@yahoo.com> ---
(In reply to Mark Millard from comment #21)

Looks like my vmstat -o use idea does not apply to the context.
Mark Johnston writes:

On Oct 31, 2025, at 07:52, Mark Johnston <markj@freebsd.org> wrote:

. . .

No, I would not expect so.  Many wired pages do not belong to a VM
object and so won't be counted in vmstat -o output.  For instance, page
table pages (though on amd64 we have the vm.pmap.user_pt_page_count and
vm.pmap.kernel_pt_page_count sysctls to count them), and single-page UMA
slabs (which back many kernel memory allocations, including ZFS ARC
data).

-- 
You are receiving this mail because:
You are the assignee for the bug.