2.1 SPARCcluster Architecture

A SPARCcluster HA Server is comprised of redundant, on-line components, which can continue system operation through failure, repair, and relocation of one assembly or device. To maintain a high level of availability, failed components should be replaced as soon as possible. Also, service precautions must be taken to maintain cluster operation while maintenance is being accomplished. See the section on "Maintenance Authorization."

2.2 Maintenance Authorization

The site system administrator must be contacted to prepare a node for maintenance and, after maintenance, to return the node to cluster membership. The procedures in this manual note points where the system administrator must be contacted. However, the equipment owner's administrative requirements supercede the procedures contained herein.

2.3 Troubleshooting a Remote Site

Use telnet to communicate with either node in a cluster.

2.4 Troubleshooting Flow

2.4.1 Takeover

The Solstice HA software allows one node to takeover when a critical hardware or software failure is detected. When a failure is detected, an error message is generated to the system console and, if required, the service provider is notified (depending upon the system maintenance contract). When a takeover occurs, the node assuming control becomes the I/O master for the disksets on the failed node and redirects the clients of the failed node to itself. The troubleshooting flow for a takeover is further depicted in Figure 2-1.

2.4.2 Switchover

Administrators can manually direct one system to takeover the data services for the other node. This is referred to as a switchover (refer to the SPARCcluster High Availability Server Software Administration Guide).

2.4.3 Failures Where There is no Takeover

For noncritical failures, there is no software takeover. However to continue to provide HA data services, troubleshooting should be accomplished in the following order:

---

Warning - DO NOT plug a keyboard directly to a node system board. If a keyboard is plugged into a system board, it then becomes the default for console input, thus preventing input from the system administration workstation/terminal concentrator serial port. In addition, plugging a keyboard directly into a node system board while power is applied to the node sends a break signal to the Solaris operating system, just as if you had typed L1-A on the console.

---

1. You will be contacted by the system administrator to replace a defective part, or to further isolate a system defect to a failed part.

2. Request that the system administrator prepare the applicable assembly containing the defective part for service.

3. Isolate fault to the smallest replaceable part.

4. Shut down specific assembly containing defective part.

5. Replace the defective part.

6. Contact system administrator to return the repaired assembly to the cluster.

Figure 2-1

2.5 Fault Classes/Principal Assemblies

SPARCcluster HA troubleshooting is dependent on several different principal assemblies and classes of faults. The fault classes and their associated assemblies are:

• SPARCstorage Array faults

  · Data disk

  · Controller

  · Fibre Channel Optical Module

  · Fibre Channel SBus card

  · Fiber optic cables/interfaces

• Node (SPARCcenter 2000 or SPARCserver 1000) faults

  · Boot disk

  · System board

  · Control board

  · Fibre Channel Optical Module

  · Fibre Channel SBus card

  · Fiber optic cables/interfaces

  · Client net Sbus card

  · Client net/connections

  · SunFastEthernet SBus card/interfaces (SunFastEthernet)

• Terminal concentrator/serial connections faults
• Software faults

  · Application program died

  · System crash (panic)

  · Hung system (lock up)

  · Cluster-wide failures

All troubleshooting begins at the system console. The console should be checked regularly, as should any other source of operator information. For example, the output of the hastat command should be checked regularly. For more information on the hastat command, refer to the SPARCcluster High Availability Software Administration Guide.

Error Messages/Symptoms

Table 2-1 lists error messages or symptoms together with the probable cause and troubleshooting reference.

Table 2-1 Error Message/Symptom

Troubleshooting

Probable Cause Cluster Service Reference

Reference

Nodes

eboots;.........SPARCcenter

Section 3.3, "Node Failures..SPARCcenter

e;..........2000 or

2000/SPARCserver 1000

esponse..SPARCserver

System Service Manual

1000

Private Net

eboots;.........SPARCcenter

Section 3.3, "Node Failures..SPARCcenter

e;..........2000 or

2000/SPARCserver 1000

esponse..SPARCserver

System Service Manual

1000

SunFastEthernet Section 3.4.1, "Private Net

SunFastEthernet Adapter

Failure (SunFastEthernet)"..User Guide

Client Network

Client net...Refer to your client network

As applicable

documentation

Public Network

Cable....(See Chapters 9 and 10 of the

Not applicable

SPARCcluster HA Hardware

Planning and Installatin Manual for cable detail.)

Table 2-1 (Continued)

Error Message/Symptom	Probable Cause	Cluster Service Reference SPARCstorage Array	Troubleshooting Reference
soc.link.5010 soc#: port: # Fibre Channel is OFFLINE; c2t4d8s2 failed (see Appendix D for additional messages)	Disk	Section 3.2, "SPARCstorage Array/Optical Connections Faults" Terminal Concentrator	SPARCstorage Array Model 100 Series Service Manual
No messages for one of the nodes on the system console; no messages from either node on the system console	Terminal concentrator	Section 3.5, "Terminal Concentrator/Serial Connection Faults"

2.7 Device to Troubleshooting Cross Reference

Table 2-2 references devices to the appropriate troubleshooting manual.

Table 2-2

Device/Trouble Area	Reference	Part Number
Array Controller/Fibre Optic Connector/ Fibre Channel Optical Module	SPARCstorage Array Model 100 Series Service Manual (Chapter 2 "Troubleshooting")	801-2206
Disk drive	SPARCstorage Array Model 100 Series Service Manual	801-2206
Terminal concentrator	SPARCcluster HA Server Service Manual (Section 3.5, "Terminal Concentrator/Serial Connection Faults")	802-3512
SPARCcenter 2000	SPARCcenter 2000 Service Manual (Chapter 2, "Troubleshooting Overview")	801-2007
SPARCserver 1000	SPARCserver 1000 System Service Manual (Chapter 2, "Troubleshooting Overview")	801-2895
SunFastEthernet adapter	SunFastEthernet Adapter User Guide (Appendix C, "Running Diagnostics")	801-6109

2.8 Device Replacement Cross Reference

Table 2-3 refers to devices and replacement procedures.

Table 2-3

Device	Cross Reference	Part Number
		SPARCserver 1000	SPARCcenter 2000
Disk drive	Disk Drive Installation Manual for the SPARCstorage Array	801-2207	801-2207
Optical Module	Fibre Channel Optical Module Installation Manual	801-6326	801-6326
SunFastEthernet	SunFastEthernet Adapter User Guide	801-6109	801-6109
System board, control board, power supply, SPARC module, boot disk	SPARCcenter 2000 or SPARCserver 1000 System Service Manual	801-2007	801-2895