There are two rc.d entries for swap:
		swap1 - all blk devices
		swap2 - any swap on network filesystems

	This causes problems for low memory diskless machines as they
	can run out of VM before swap2 is reached.

	One proposal would be to add a 'fail silently if add swap fails'
	option to swapctl (-q), and to use this in swap1, and add another
	option to not complain if any listed devices were already in use
	as swap (similar to mount -a)

	swap1 could either be:
		swapctl -A -t blk
		swapctl -A -t noblk -q
	or
		swapctl -A -q

	The former is the smallest change from the current behavior - it
	would just attempt to mount any swap on network filesystems.

	The latter would allow for swap to be put on filesystems not in
	$critical_filesystems_beforenet, and picked up by swap2.

	How does this seem?

-- 
		David/absolute		-- www.netbsd.org: No hype required --