Re: 12.3: "swapon -a" crashes the system

On Thu, Dec 09, 2021 at 02:35:55PM +0700, Eugene Grosbein wrote:
! 09.12.2021 13:11, Peter wrote:
! 
! >  if you run out of swapspace and you think you might just create
! > some extra devices and add them to /etc/fstab and then run "swapon -a"
! > to enable them, don't do that.
! > 
! > The result might look like this:
! > kernel: pid 12296 (daemon), jid 5, uid 5100: exited on signal 10 (core dumped)
! > kernel: pid 17717 (ruby27), jid 5, uid 5100: exited on signal 6 (core dumped)
! > kernel: pid 14938 (daemon), jid 10, uid 5100: exited on signal 10 (core dumped)
! > kernel: pid 19184 (ruby27), jid 10, uid 5100: exited on signal 11 (core dumped)
! > kernel: pid 19182 (ruby27), jid 10, uid 5100: exited on signal 10 (core dumped)
! > ...
! > with a subsequent kernel crash.
! > 
! > In my case I did run out of swapspace despite proper sizing, because
! > of unbalanced numa-domains. (Each of them may hit free_target on their
! > own behalf, and then start paging.)
! > 
! > So I did as described above, but I added them with the "late" option,
! > and then did run "swapon -a", which did not even add them - but
! > nevertheless produced the crash. The processes that do coredump might
! > be those that are fully swapped out?
! 
! Did you try using ZFS-backed files or devices for swap? If so, this is somewhat expected.


Hi Eugene,
  that's a negative:
  
# cat /etc/fstab
/dev/ada3p3             /                       ufs     rw 1 1
/dev/ada3p4             /usr                    ufs     rw 1 2
/dev/ada3p5             /var                    ufs     rw 1 2

proc                    /proc                   procfs  rw 0 0

/dev/ada1p2.eli         none                    swap    sw,trimonce 0 0
/dev/ada3p2.eli         none                    swap    sw,trimonce,late 0 0
/dev/ada1p10.eli        none                    swap    sw,trimonce,late 0 0
/dev/ada3p10.eli        none                    swap    sw,trimonce,late 0 0
...


And it happened two times: first when I ran into exhaustion and found
some space on a spinning to quickly add. And second after I had
figured out the numa matter, reassessed my calculations, and added
proper SSD space.


cheerio,
PMc