IBM RS/6000 SP, Maintenance Information

Page: 66 / 388

MAP 0130: Processor Node
Figure 1-13. POWER3 SMP Thin and Wide Node High Level Component Diagram
001
Error reported in system monitor log or visual indication of a problem in processor node.
Read these warnings and then follow steps to ensure continuity of customer’s jobs in the queue:
Attention: The processor node(s) must be removed from active configuration before continuing. If
processor node(s) is/are off, continue; otherwise, ask customers to initiate shutdown procedure and
power-off processor node(s) from the control workstation, or defer maintenance until all jobs are com-
pleted. Powering off a processor node(s) in a parallel environment will cause all jobs to flush from the
queue and switch initialization to rerun.
Attention: Servicing a processor node with the High Performance Switch (HiPS) feature or Scalable
POWERparallel Switch (SPS) feature installed, will impact the entire switch network, unless the processor
node has already been powered off (or fenced) and/or the switch data cable has been disconnected.
– Refer to "Service Position Procedures" in
IBM RS/6000 SP: Maintenance Information, Volume 3,
Locations and Service Procedures for placing or removing processor node(s) in/from service position.
– Refer to "Viewing Switch Partitions" in IBM RS/6000 SP: Maintenance Information, Volume 3, Locations
and Service Procedures for locating and fencing/unfencing nodes within a switch partition.
– From the control workstation, check the hardware monitor log for indication of problem.
– The following table contains a prioritized list of conditions. Find the first condition that describes your
problem, then perform the associated action:
1-46
RS/6000 SP MAPs