IBM RS/6000 SP, Problem Determination Manual

Page: 131 / 316

This soft copy for use by IBM employees only.
Estart can all create a switch fault. Either the Worm on the nodes or the switch
chips themselves can report a fault to the Worm on the primary node.
The Worm process verifies the switch connections beginning at a node
designated as the primary node. By default, the primary node is node 1 in frame
1. You can override the default and designate another node as the primary
node. With the High Performance Switch, this must be done manually if the
primary is not operational, and the
Estart
command should then be run.
fs_monitor
This daemon monitors the adapter for interrupts that have not been serviced.
To check that both daemons are running on a node run:
# ps -ef | grep css
Note: Add “_SP” to each name to find out the SP Switch daemons.
4.6.1 Switch Responds
The spmon GUI for switch_responds has changed for the SP Switch. Figure 54
shows the differences between them:
High Performance Switch SP Switch
initial state red yellow
on the switch green green
fenced (no autojoin) yellow red
fenced with autojoin yellow yellow
failure red
Figure 54. Switch Responds
For the SP Switch, an extra attribute has been added to the SDR object class
switch_responds called adapter_config_status which gives the results of the
switch adapter
′
s configuration method. This field is left blank for the High
Performance Switch adapters only if you do not have PTF 11 installed. Here are
the possible values for this attribute for the SP Switch at the present time:
•
not_configured (initial value)
•
css_ready (method was successful)
•
odm_fail
• chkslot_fail
•
busresolv_fail
• getslot_fail
•
xilinit_system_fail
•
xilinit_load_fail
•
dd_load_fail
•
genmajor_fail
•
genminor_fail
•
make_special_fail
• build_dds_fail
•
dd_config_fail
• diag_fail
•
fs_load_fail
• fs_init_fail
Chapter 4. The Switch
111