Problems and fixes for the online SAPCs available via mtip on ulab.
Keywords: SAPC, mtip, reset, download, reset server.
Courses: cs341, cs444, cs644
Reported problem: an online SAPC cannot be reset by ~r in mtip.
PCs in hardware lab: Start with step 3 below.
Background: How to do ~r (the mtip reset command) in mtip for a
certain system (on ulab only), say system 3, or "board 3"
u13% mtip -b 3 --or 1 or 2 or ..., the board no.
~resetting board 3! --you type ~r, reset service outputs rest
--wait 12 seconds for PC reboot
Please confirm ... --first output from the board: Type CR
If mtip is not in your path, do "module load ulab" and try again.
Fixing most cases:
1. Try ~r yourself, on any board.
If you see "No ack of SOH in 5 tries--reset server hung!!"
First (if possible) power-cycle the reset server in S/3/142, high
on the metal shelves at the back of the room. If still in trouble,
use mkill to kill all the processes using mtip and its helper,
rserver_client. Just run "mkill" and answer "y" to all questions
about killing processes. This can be done without privileges. It only
kills mtip and related processes, not any other user processes.
2. If ~r gives no abnormal output, the reset server is probably ok.
(Normal output: "~resetting board 3!", for board 3)
Try ~r on the reported board, to make sure the user provided the
right board number. If the board gives no output, it has probably
lost its saved BIOS configuration information or has some other
hardware problem. If *no* system gives any output, even after ~r,
or mtip won't even connect completely, then suspect the Portmaster.
Go to step 4.
3. Confirmation of individual system problem: Find the physical system in
the hardware lab (number on upper rim). Turn it off and on again.
If it still doesn't work, go to 5.
4. Fixing the Portmaster. It is in S/3/142. It can be power cycled.
5. If the fixes here don't work, report by email to jlentz and eoneil.
6. In any case, log problem as usual as a system problem.
Fixing lost BIOS configs on a PC.
If the PC beeps *twice* when powered up, it has lost its configs
(or worse). If it beeps once but doesn't work for users, something
else is wrong.
Connect keyboard and monitor to SAPC. Power on while typing DELETEs
until you hear a buzz. Now you see the full BIOS menu for configs,
unless something else is wrong. Find the menu item to reestablish the
factory defaults and confirm it, and then ESC to get you back.
In some cases, more CRs may be needed to confirm special questions
the BIOS feels it needs to ask you. Just take defaults, but don't
set passwords or any other security feature! When in doubt, just
power-down and report the problem.
Fixing a hung terminal controller.
The SAPCs and the reset server are all connected to ulab by
the Portmaster terminal concentrator. This is a separate box on the
Ethernet from ulab, so we should not have to reboot ulab to fix it.
It can be separately power-cycled. It is in S/3/142 on the same
set of metal shelves as the SAPCs.
Previous: The mtip Command: reset