Re: Can not build kernel on 1GB VM

From: Peter Jeremy <peterj_at_freebsd.org> 
Date: Sat, 16 Apr 2022 10:25:23 +1000:

> On 2022-Apr-15 13:49:53 -0400, Michael Wayne <freebsd07_at_wayne47.com> wrote:
> >I have a VM with 1GB RAM running FreeBSD 12.1-RELEASE-p3
> >
> >I'm trying to upgrade the machine to 12.3 and having swap failures.
> 
> Are you using ZFS or UFS and are you swapping to a raw partition
> or a file/zvol?

Gad, why did I not think to cover that issue in my earlier
note to freebsd07@wayne47.com ? Good catch. In bugzilla
there is:

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=206048

comments #7 and #8 are related to the file/zvol issue.

freebsd07@wayne47.com also indicates a VM context as well.

> >the machine ran out of swap. with a bunch of messages like:
> >   Apr 15 12:11:26 g1 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 240593, size: 4096
> >   Apr 15 12:11:35 g1 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 236224, size: 16384
> >   Apr 15 12:11:37 g1 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 245, size: 12288
> >   Apr 15 12:11:46 g1 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 240593, size: 4096
> >   Apr 15 12:11:55 g1 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 236224, size: 16384
> >   Apr 15 12:11:57 g1 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 245, size: 12288
> 
> That message doesn't indicate "out of swap",

I expect that the "killed due to out of swap space" message that
was originally referenced (though not quoted) was of the structure:

kernel: pid ???? (???), jid ??, uid ??, was killed: out of swap space

That is the message type that I was reporting as being a misnomer
when it is reported: the swap space available does not have to be
low at all for such kills with such messages.

> in indicates that your
> swap device isn't responding within a "reasonable" time. If you are
> swapping to a local ZFS file or zvol, you may be running into a known
> deadlock condition and should switch to using a raw partition for
> swap.  If you can't improve your swap device performance, adjusting
> the vm.pageout_oom_seq and vm.pfault_oom_attempts sysctls should
> resolve the problem.

freebsd07@wayne47.com later reported they did not help --but
I do not know if a swap partition was what was being used or
not.

> A true "out of swap" will appear as a message similar to
> swap_pager_getswapspace(nnn): failed

Yea, there are 2 messages that are not misnomers:
(Quoting from another list exchange that I had once,
that in turn had copy/paste based text from an
actual set of messages.)

QUOTE
Jan  5 21:04:31 r5601g kernel: swap_pager: out of swap space
Jan  5 21:04:31 r5601g kernel: swp_pager_getswapspace(7): failed
END QUOTE

Neither of these messages are announcing kills.

But if neither of these types messages have shown up, then
any "was killed: out of swap space" message is a misnomer
for the "out of swap space" part of the message.

===
Mark Millard
marklmi at yahoo.com