[Sunhelp] Help! Ultra 1 170E 3D hangs without warning

[Gary Rafe]

We have an Ultra 1 170E Creator3D with 256 MB RAM
running Solaris 7 (kernel 106541-12)
that's recently started to hang completely without warning.

After running 3 months without the need of a reboot,
we came in to work on Tuesday last week to find it crashed with
a kernel panic (BAD TRAP, type = 0x31).
The system came right back up,
but apparently restarted itself later that evening
(after another BAD TRAP, type = 0x31 kernel panic).

Then, sometime on Friday afternoon,
the system "locked up" while running normally;
no kernel messages were reported at the time.

When we came in to investigate Saturday evening,
we found the system powered on,
with no unusual messages on the console,
and otherwise completely unresponsive
(from the network, the keyboard or the mouse).

The system came up again after recycling power,
and we took the opportunity to install the 31 October
Sun Recommended Patches.
Then, sometime Sunday morning, the system "locked up",
again with no kernel messages reported.

When we came back in to restart the system
(again, by simply recycling the power),
we ran the "maximum" power-on self-test (POST),
which appeared to pass without incident.

Now, this afternoon we came in to find the system frozen, once again.
This time, we tried to reset the keyboard by disconnecting its cable
at the chassis, and managed to get the boot prompt (ok>) on the console.
Typing "go" at the prompt resulted in another unresponsive keyboard.
POST ran again on the next power cycle,
and all tests passed without complaint.

We will appreciate greatly any and all suggestions
that help us diagnose (and correct!)
these (seemingly random) system crashes (lock-ups, hangs, etc.).
--
Gary E. Rafe
gerst4+ at pitt.edu