On 19/08/2017 22:00, Cy Schubert wrote:
> An easy way to find out is to run top, type in "w", then "o" and "swap" to
> see which processes are using swap. You'll notice that the numbers won't
> add up. I haven't looked at this but my guess is that there may be swap
> leak. You can verify this by replacing the swapfile (add a new and remove
> the old).
Thanks for the tip. I need to wait (might be a few weeks) to see when it
starts eating swap again, then I'll do what you suggest. I got the
system from 94% swap in use to 39% by restarting some of the VMs, then I
was able to swapoff/swapon to empty the swapfile.
Here's a snapshot of the system in an idle state, sr is mostly above 300
procs memory page disks faults cpu
r b w avm fre flt re pi po fr sr ad0 da0 in sy cs
us sy id
0 0 27 189G 27G 103 0 0 0 77 329 0 0 74 801 1663
0 0 100
0 0 27 189G 27G 38 0 0 0 0 337 0 0 93 745 1997
0 0 100
0 0 27 189G 27G 187 0 0 0 0 329 0 0 90 669 1853
0 0 100
1 0 27 189G 27G 38 0 0 0 0 340 0 0 83 774 1816
0 0 100
0 0 27 189G 27G 37 0 0 0 1 370 0 0 77 767 1839
0 0 100
1 0 27 189G 27G 48 0 0 0 0 294 1 0 125 2239 3382
0 0 100
0 0 27 189G 27G 20 0 0 0 0 329 3 0 88 651 1797
0 0 100
^C
yet mem from top shows 27GB free:
last pid: 71790; load averages: 0.11, 0.09, 0.05
up 121+03:19:29 13:55:49
99 processes: 1 running, 98 sleeping
CPU: 0.0% user, 0.0% nice, 0.1% system, 0.1% interrupt, 99.8% idle
Mem: 769M Active, 11G Inact, 21G Laundry, 64G Wired, 924M Buf, 27G Free
ARC: 60G Total, 8979M MFU, 51G MRU, 16K Anon, 153M Header, 755K Other
60G Compressed, 62G Uncompressed, 1.04:1 Ratio, 41M Overhead
Swap: 4034M Total, 4034M Free
thanks,
--
J.
_______________________________________________
freebsd-current@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"
