Johnny Billquist writes:
[...]
> How do you manage to get any information from the system when you have run
> out of memory? For me, the system hangs, period. I cannot execute any
> commands.

You can't start new processes, but the already running processes are
still displaying output.

In my case I'm using two additional xterm/rlogin sessions with `top`
and `vmstat 5` (or `vmstat w 5`) for monitoring the system.

> > I've no clue why this 3100 doesn't swap. Since it behaves the same
> > with swap on sd0b and with swap on nfs, I'm now suspecting a general
> > problem with 1.2D. Is there anybody out there who's 1.2D (or 1.2x)
> > machine is swapping? What could be the reasons for the behaviour I get?
> > Is this problem specific to the VAX port? Or even specific to 3100s?
>
> Think it is a more general problem, but I have no clue to where it is.
> Could be VAX specific though...

BTW: in sys/vm/vm_glue.c there's a piece of code surrounded by
     "#if !defined(vax)" while the comment refers to i386. Is this correct?
     (I've temporarily removed this, but it didn't cure my problem)

> (Btw, how did you get those vmstat lists? vmstat won't run for me. Also, I
> hope Ragge includes my fixes to get the CPU times correct. idle shouldn't
> be 0% :-)

If kernel and vmstat are compiled from the same (-current) source tree
it always worked for me. If vmstat/top are using other/older includes
than kernel during compile they sometimes display garbage. Re-compiling
usually helps.

Ciao,
        bertram