[Bug 234559] memory reporting inconsistency 12-STABLE vs. 11-STABLE

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234559

            Bug ID: 234559
           Summary: memory reporting inconsistency 12-STABLE vs. 11-STABLE
           Product: Base System
           Version: 12.0-STABLE
          Hardware: Any
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: kern
          Assignee: bugs at FreeBSD.org
          Reporter: freebsd at oldach.net

Created attachment 200695
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=200695&action=edit
memory graph

I have recently upgrade from 11-STABLE to 12-STABLE. While the total memory
(active+inactive+laundry+wired+cache+free) in 11-STABLE was fixed, this sum
seems to be somewhat floating in 12-STABLE. Occasionally it even appears to be
above physical memory!

Please see attached graph (similar to what is produced by munin): The picture
is covering the last 14 days with the oldest data to the left; reboot to
12-STABLE was last Sunday afternoon (where the gap is visible). Before the
upgrade (running 11-STABLE) there is almost a flat total, after the upgrade
(running 12-STABLE) the total is jumpy.

Also the top(1) output is weird on this machine with 4G DRAM:

# top -n | fgrep Mem
Mem: 269M Active, 2083M Inact, 15M Laundry, 700M Wired, 336M Buf, 1073M Free
# echo $((269+2083+15+700+1073))
4140
#

Clearly this cannot be just a rounding issue - it appears the stats counters
are not updated appropriately in 12-STABLE:

# sysctl -a vm.stats.vm | fgrep _count
vm.stats.vm.v_cache_count: 0
vm.stats.vm.v_laundry_count: 3926
vm.stats.vm.v_inactive_count: 533266
vm.stats.vm.v_active_count: 76185
vm.stats.vm.v_wire_count: 179345
vm.stats.vm.v_free_count: 267048
vm.stats.vm.v_page_count: 980845
# echo $((0+3926+533266+76185+179345+267048))
1059770
# echo $(($(sysctl -n vm.stats.vm.v_page_size)/1024))
4
# echo $((1059770*4))
4239080
#

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