IBM RS/6000 SP, Problem Determination Manual

Page: 24 / 316

This soft copy for use by IBM employees only.
Figure 3. The High Performance Switch
As you can see in Figure 3, each node is connected to one chip port, and there
are eight chips per switch board. Each chip has eight ports, four external ports
connecting the nodes, and four internal ports connected to the mirror chips.
This topology gives each node four different routes to communicate with another
node. Therefore, each node has four possible ways to reach another node, and
it could use all of the routes together to have a very high bandwidth.
The software portion of the switch provides the device drivers, the processes and
the
commands needed to use and manage the switch. These components will
be covered later on this book.
The Switch has the ability to partition, which creates separate and non-disrupting
environments. In this way, each partition can be seen and managed as a
separate SP system. This topic will be covered in more detail in the next
chapter.
1.2 Problem Determination
Basically, there are two kinds of problems encountered on the RS/6000 SP:
those related to AIX and those related to the POWERparallel System Support
Programs. For those related to AIX, the approach to solve them is the standard
AIX way, which means you can use
errlog and trace facilities to find the problem.
For those problems specific to the RS/6000 SP, each PSSP component usually
provides log files and tools to solve them. All the log information is located on
/var/adm/SPlogs. The directory structure is shown in Figure 4 on page 5.
4
SP PD Guide