David.Eckhardt@b.gp.cs.cmu.edu writes:

>One thing I've noticed is what I call "blinky-light mode".  During times of
>heavier-than-normal memory activity, often related to X, the system will
>appear to wedge (no mouse tracking, no keyboard echo) for as much as 30
>seconds or a minute.  If I look down at my disk controller status LED, it
>will be flashing on and off at around .5 or 1 Hz, but I will typically not
>hear a lot of seek activity from the drive.  Eventually the drive will start
>seeking and everything will come back to life.

>I haven't submitted any kind of bug fix for this for a couple of reasons. 
>On the one hand, I haven't observed this on enough NetBSD systems to feel
>I can claim it's independent of my setup (IDE disks, for example); on the
>other hand, I have held off because of a suspicion that this is a
>long-running bug in the Mach VM code (I saw exactly this behavior under Mach
>2.6 on a Sparc 1+) and hence unlikely to be fixed quickly.

>Does this match anybody else's observed behavior?

Yes.  We had a lengthy discussion on this last September.  Basically, when
the pagedaemon ran out of swap buffers it would sleep on lbolt to wait for
swap buffers to become available.  The kernel issues a wakeup on lbolt every
second.  This was fixed, but I don't know if that fix has made it into 1.2.1.
-- 
Christoph Badura

Now available in print: Lion's Commentary on UNIX 6th Edition, with Source Code
			http://www.peer-to-peer.com/