Just wondered if anyone has seen this before -
a bit long winded so the summary is:

'SS20 firewall/proxy locked up but you can still NAT through it
and kernel appears to be running (BLINK_LED).
NFS mounts and proxying (Squid) are broken however. [ disk? ]'

<detail>
 
My SS20 runs 24/7 as a home wireless gateway/firewall/proxy etc,
connected to a 512k cablemodem (no inbound port restrictions).
Ross 150Mhz CPU w/320Mb RAM.

I came back to work Tuesday after a long weekend, logged in remotely
and started reading mail etc. At some point over the morning the session
locked (this happens a fair bit due to an overworked corporate firewall,
screen generally makes it a nonissue. This time I couldn't log in.) .

ISP aren't aware of any faults (and they're usually obvious).

Forgot about it (I'm not hacking at home these days due to RSI).

Next day in work, still can't see the Sparc. GF says Net is slow/unusable.

I thought my dyndns hostname might be invalid so got her to mail to a
shell account. Mail headers confirm the sparc is at the right IP, but nmaps
from several other boxen around the Net can't see it at all.

Girlfriend falls back to dialup.

When I got home, the LED is blinking (options BLINK_LED), PC (NFS client)
has lost its mounts, tip says 'connected' and hangs on serial.

Sod it, pressed the BRS.
Tried investigating, but nothing had been written to any logs between
midday tuesday and the time I rebooted it. Up to that point a seems normal.

</detail>

My guess is overheating or disk, but how can I investigate (and stop
it recurring, obviously)? Cheers.
-- 
Rasputin :: Jack of All Trades - Master of Nuns