Not sure it is related, but I have little time to investigate now
(need to catch a plane). I've seen two freezes on our server, one with
1.6_BETA1 and one with 1.6_BETA2. Similar symptons: ping-able, virtual
screen switching worked, but nothing else. I had a `top' running at
one of the virtual screens and it showed that all (AFAICS) user
processes where in state `pager_ma'. 

I suspected my 2 mfs filesystems (on /tmp and alike) might be involved as
the latest syslog entries before the freeze reported filesystem full
on them. I removed the mfs fs (with UBC not much use anymore anyway)
and hope that fixes it for now.

In our case, the freezes do not corrolate with heavier than average
filesystem activity AFAICS: generating very heavy filesystem I/O does
not reproduce the problem. Also, SOFTDEP is in my kernel but I have
no filesystems mounted using softdep.

Hope this helps the hardcore bug finders a bit,

Feico.


On Sun, Jun 16, 2002 at 02:03:22PM -0400, Jason R. Fink wrote:
> > has anybody else observed a frozen system with NetBSD 1.6_BETA1(*)?
> > For the second time now my system got unresponsive during heavy
> > filesystem activity. It still responded to ICMP echo packets and I
> > could switch the virtual console but it was complete stuck otherwise.
> 
> actually yes, but not me. I did a 1.6A install for a buddy of mine
> because he wants to play with netbsd. the first night it locked up
> during a large build (i think apache and all of its deps). same 
> thing, console went dead. now, here is strange thing, he does not
> seem to have virtual consoles.
> 
> when i catch up with him i will ask him for a dmesg.
> 
> i have 1.6A on a laptop and i have not seen anything similar.