test

Sun Cluster 2.2 System Administration Guide

13.2 Servicing

From the perspective of the Sun Cluster software, the multiple disks of the Sun StorEdge A3x00 or the Sun StorEdge A1000 appear to be a single, highly reliable disk. Most service procedures are done using the Sun StorEdge RAID Manager software. However, if you ever need to disconnect the interface card between the node and the SCSI bus (the UDWIS card), use the procedure in this section. For example, use these procedures to change UDWIS cards or to replace the node system board.

13.2.1 How to Disconnect the UDWIS Card

  1. Use the haswitch(1M) command to migrate any logical hosts from the Sun Cluster node to be serviced.

    phys-hahost1# haswitch phys-hahost2 hahost1 hahost2

  2. Quiesce activity on the SCSI buses involved with the hardware service.

    If the configuration has both SCSI buses to a Sun StorEdge A3x00 or a Sun StorEdge A1000 on a common board, select one bus to perform this step. If there are multiple expansion units connected to this board, one bus on each expansion unit can be prepared in parallel. If only the UDWIS card needs replacement, other expansion unit connections can remain connected.

    Perform this step using the Maintenance/Tuning application of the RAID Manager software.

  3. Disconnect each quiesced SCSI bus cable at the disk expansion unit end and install the appropriate terminator.

    If the remaining SCSI bus of any expansion unit is also connected to this board and needs to be removed (for example, for system or I/O board replacement), route all I/O to the first SCSI bus (the one just terminated). Disconnect the SCSI cable at the expansion unit end and terminate it.

  4. The system or I/O board is now ready for servicing.

    Disconnect the SCSI cables from the prepared UDWIS cards and perform the required hardware service procedures. When complete, re-establish the original connections by continuing with the following steps. Continue with this node halted or powered off.

  5. If necessary, quiesce one SCSI bus to each expansion unit that needs reconnection.

  6. Remove the terminator on these connections.

    This is the terminator that was installed in Step 3.

  7. Reconnect the SCSI cable between the expansion unit and the UDWIS card.

  8. If another SCSI connection to an expansion unit was affected by the service procedure, route all I/O to the newly reconnected SCSI bus, remove the terminator, and recable the expansion unit to the UDWIS card.

  9. Boot this node.

  10. Restore the logical unit (LUN) distribution back to the standard configuration using the RAID Manager software.

  11. Use the haswitch(1M) command to migrate any logical hosts back to the Sun Cluster node that has been serviced.

    phys-hahost1# haswitch phys-hahost1 hahost1