Since swapping over NFS doesn't seem to work too well, I figured I'd try
a little harder to ressurect the three dead RD53s I have.

The first one refuses to spin up at all, no matter how much I play with
the servo connector.  I've almost written it off, but it shows up as an
RD51, which makes me suspect the electronics...  I have yet to try.

The second and third drives I can get to spin up and hold a ready light,
but as soon as anything accesses it, I get a "blip blip <pause> blip blip"
of the ready light.  I opened one to make sure that the heads were not
stuck, and actually watched it to see what was going on.  The heads
weren't stuck, but on first access to the disk, the heads started
seeking across the platters repeatedly.  I put the cover back on, fired
up diagnostics, and it's now formatting.  :)  Hopefully the third drive
(unopened) will respond kindly to low level formatting as well.

Anyway... so where does one get RD53 / Micropolis 1325s or any
DEC-compatible drives for these things nowadays?  Are there any places
that sell refurbished drives?

It finished formatting... so now I at least have local swap.  But I'm
still seeing the same problem -- swapping doesn't seem to work too well:
the machine deadlocks.  I'm running 19990101 generic kernel.  vmstat
looks like this:

sam# vmstat 5
 procs   memory     page                       disks      faults   cpu
 r b w   avm   fre  flt  re  pi   po   fr   sr r0 r0 r1   in   sy  cs us sy id
 0 0 0  1864   224   58   0   0    0    2    2  0  0  0    0   67  39 23 57 20
 1 1 0  2032   172   20   0   0    5    4    6  1  0  0    0   24  13  4 23 72
 0 3 0  2168    20   15   0   0    0    0    0  0  0  0    0   20   8 17 12 71
 0 3 0  2168    20    1   0   0    0    0    0  0  0  0    0    2   1  0 1 99

the last line repeats and the machine responds to pings, but is
otherwise locked.  Currently running processes (like vmstat) continue to
run, but nothing else can start, since no memory ever gets freed.

I can force messages from the scheduler (scheduler: no room for pid
89(portmap), free 1) by poking at things like the portmapper and trying
to telnet in, but the machine never frees itself up.  :(

I seem to remember other people (Brian Chase?) having similiar problems
with newer VAX kernels on their uVAX-IIs...

NetBSD 1.3I (GENERIC) #22: Fri Jan  1 22:52:38 CET 1999
    ragge@subzero:/multi/src/sys/arch/vax/compile/GENERIC

MicroVAX II
realmem = 6283264
avail mem = 3719168
Using 76 buffers containing 311296 bytes of memory.

I'm currently downloading the packages for 1.3.2 to see if that makes
any difference.  (Although shared libraries on a 6MB box is really the
way to go, I'll take whatever's reliable.)

----
  Aaron J. Grier  | "Not your ordinary poofy goof." | agrier@poofy.goof.com
      "A few compilers designed on and for the original IBM 370 series
       machines have a more effective register allocater than anything
       GCC has ever posessed." -- David S. Miller on linux-kernel