IBM RS/6000 SP, Maintenance Information

Page: 25 / 388

MAP 0100: Start of Service Call
– Features may be installed in this product that are not covered in the following MAPs. Should the fol-
lowing diagnostics fail to resolve the field problem, refer to the documentation that came with the
installed features for further diagnostic information.
– If the control workstation hardware is not operating correctly, refer to the appropriate workstation diag-
nostic manual. Otherwise, go to Step 002.
002
– Check for the existence of the high availability control workstation (HACWS) by looking at the frame
supervisor card. Verify the twin tail RS/232 cable hook-up needed for HACWS.
Note: The twin tail frame supervisor card is also needed for frames supporting the SP Switch or for
frames containing 604 or 604e high node(s).
– Start either Perspectives or the System Monitor:
Perspectives
On the command line of the control workstation, type sphardware and press Enter, to start Per-
spectives. Refer to the Perspectives information in
IBM RS/6000 SP: Administration Guide,
GC23-3897.
System Monitor
Ensure that the system monitor software is active on the control workstation by following the proce-
dures below. Refer to the System Monitor information in
IBM RS/6000 SP: Administration Guide,
GC23-3897.
1. Check for a window labeled “System Monitor”.
2. If the window is not found, look for an icon (small box) labeled “disp_in:” or “disp_mgr”; click on
the icon to raise the window.
3. If no window appears, click on an active AIX window and enter:
spmon -g
Open the “3 Digit Display” window by following these procedures:
a. From “Display Manager” or “System Monitor” window, click on “SP”.
b. From the pulldown menu, click on the “All Node Summary Display” option.
c. From the “Available Variables” menu, click on “3-Digit Display”, then click on the “Display All
Node Summary” button.
– Open a TTY window, if possible, to the failing node. Real-time error data can be observed that can help
with problem determination and repair.
– Check the boot Ethernet and switch communication status by issuing the following command from the
control workstation:
SDRGetObjects host_responds (Ethernet)
SDRGetObjects switch_responds (Switch)
A value of "0" for host_responds or switch_responds, for any node, indicates a communication
problem (this may be the result of some other hardware problem).
– Check for system monitor, node, and switch errors by issuing one or more of the following commands
from an AIX window:
Chapter 1. Maintenance Analysis Procedures (MAPs)
1-5