RPI3 swap experiments ["was killed: out of swap space" with: "v_free_count: 5439, v_inactive_count: 1"]

[bob prohaska]

On Sun, Aug 12, 2018 at 08:36:01PM -0700, Mark Millard wrote:
[snip]
> 
> I'd keep multiplying by 10 until it works (or fails some
> other way), then back off by smaller factors if you want
> a narrower range to be known between failing and working
> (or failing differently).
> 
[snip]
> 
> The factor of 10 rule makes the number of tests
> logarithmic to find an sufficient upper bound (if
> there is an upper bound). After that with high/low
> bounds binary searching is a possibility.
> 
Updated to r337688 with 2 GB of swap divided between USB and microSD.
Using vm.pageout_oom_seq=1024, a -j4 buildworld failed with da0 errors 
reported, panic'd and rebooted. A -j3 buildworld ran to completion with 
several dozen "indefinite wait..." messages but no other complaints.

IIRC, this swap configuration wouldn't even run -j2 to completion with
vm.pageout_oom_seq=12, so the increase to 1024 clearly helps and the
da0 errors suggest something else gives up if taxed to -j4.

During the -j3 buildworld, at about the 26 MB point in the log file 
top reported ld.lld having "size" of 1037 MB, but only 60 MB of swap
were in use. With only 1 GB of main memory, is that to be believed?

The backtrace emitted by the -j4 panic is somewhat longer than
usual, it's in the file named console at 
http://www.zefox.net/~fbsd/rpi3/swaptests/r337688/1gbsd_1gbusb/
in case it's of interest.

Thanks for reading!

bob prohaska