Sun HPC 3.0 SCI Guide

SCI Switch

General Hardware Inspection

Perform the following checks to determine the physical state of various SCI subsystem components. Verify that:

All SCI scrubber jumpers are properly set, depending on the cluster topology.
All SCI cables are properly seated.
All SCI switches have power applied
No SCI status LEDs are red--see Table 6-1 and Table 6-2

SCI Switch Status LED Locations

Clusters with three or four nodes can be connected through one or two SCI switches. The switch status LEDs provide information that can be used to troubleshoot SCI switch failures (Figure 6-1). Guidelines for interpreting these LEDs are provided in "Port Status LEDs"" and "General Switch Status LED".

Figure 6-1 SCI Status LED Locations

Port Status LEDs

The four port status LEDs located on the switch front panel can be used to troubleshoot individual port failures (Table 6-1).

Note -

A switch port sync error can result from a cable being removed.

Table 6-1 SCI Switch Port Status LEDs


Situation	Port LED Status
No power	All four LEDs not lit
Fatal switch errors: fatal hardware error, temperature to high, fan(s) not operative, power supply problem	All four LEDs red
Port errors: SCI cable out, sync error	Associated port LED is red
Port operative, no transactions	Associated port LED is green
Port operative, with transactions	Associated port LED is blinking green

General Switch Status LED

The switch status LED located on the rear panel indicates overall switch failures (Table 6-2).

Table 6-2 SCI Switch Rear Panel LED


Situation	LED Status
Fatal switch errors: fatal hardware error, temperature too high, fan(s) not operative, power supply problem	Red
Switch operational	Green

The `get_ci_status` Command

You can use the results of the get_ci_status command to troubleshoot clusters that have SCI switches. For example, for the configuration in Figure 6-2, if the get_ci_status command is used on interconn1, a typical output would be:

# /opt/SUNWsma/bin/get_ci_status
sma: sci #0: sbus_slot# 1; adapter_id 8 (0x08); ip_address 1; switch_id# 0; port_id# 0; Adapter Status - UP; Link Status - UP
sma: Switch_id# 0
sma: port_id# 1: host_name = interconn2; adapter_id = 72; active | operational
sma: port_id# 2: host_name = interconn3; adapter_id = 136; active | operational
sma: port_id# 3: host_name = interconn4; adapter_id = 200;inactive|inoperational
#

In this example, the line

sma: port_id# 3: host_name = interconn4; adapter_id = 200;inactive|inoperational

indicates that the path between SCI switch 0, port 3 and interconn4 is inactive and not operational.

Figure 6-2 Typical Four-Node Configuration with an SCI Switch

In this instance, if the get_ci_status command were run on all four nodes, and if the same path was inactive and inoperative between SCI switch 0, port 3 and interconn4, it is more than likely that either the SCI switch 0, port 3, the cable, or the interconn4 host adapter is faulty.

However, if the get_ci_status command indicates that the same path is inactive and inoperative for one node only, such as in the instance of interconn1, then it is more than likely that either the interconn 1 host adapter, the cable, or SCI switch 0, port 0 is faulty.

Note that some aspects of the get_ci_status command output, such as host names, will vary according to your configuration.

Client Net Failure

System console messages will identify the specific port that has failed. Otherwise, for information on test commands as well as additional troubleshooting, refer to the documentation that came with your client network interface card.