PM/Ethernet Test Procedure

1. etherpmctl command
   Be sure the etherpmctl command is already issued during boot via the pm_ethernet rc script on all hosts. If it has not been issued yet, invoke it on all hosts as follows:

   $ su
   Password:
   # /etc/rc.d/init.d/pm_ethernet start
   

2. Point-to-Point test (Message)
   Make sure that the PM kernel driver is installed on all hosts. To test the PM communication function, use the rpmtest command. For example, if you want to test communication from comp0 (node 0) to comp1 (node 1), please follow the instruction below: P> Issue the following commands in a window on the server host:

   $ rpmtest comp1 ethernet -reply
   

   Issue the following commands in another window on the server host:

   $ rpmtest comp0 ethernet -dest 1 -ping
   	Success: Two numbers similar to "8     0.000113725" are printed
                    The first number is the data size (in bytes)
                    The second number is the latency (in seconds)
   	Failure: Error and dump messages are printed
   

   When the test finishes, do not forget to kill the pmtest process on the server host because the process is in an infinite loop.
   If this test fails, read the Troubleshooting section.

3. Total test
   This is the stress test. All hosts randomly send burst messages to other hosts. Issue the following commands on the server host:

   $ scout -g pcc
   SCOUT: Spawn done.   
   SCOUT: session started
   $ scstest -network ethernet
   CSTEST: BURST on ethernet
   50 K messages.
   100 K messages.
   150 K messages.
   200 K messages.
   250 K messages.
   300 K messages.
           Success: number of messages to send are printed.
           Failure: Error messages are printed.
   

   To stop this test, press Ctrl-C or another interrupt command.
   If this test fails, read the Troubleshooting section.

4. Network Trunking
   If you want to install Network Trunking, refer Network Trunking for PM/Ethernet Administrator's Guide

Troubleshooting

etherpmctl command fails.

• The network interface of the host is not configured correctly.
• PM/Ethernet driver of the host is not configured correctly. Please check the kernel boot log using the dmesg command.

Point-to-point test fails.

• etherpmctl command failed or was not executed.
• If the message ``pmOpenDevice: No such file or directory'' is displayed, the configuration file (e.g. pm-ethernet.conf) does not exist.
• If the message ``pmOpenDevice: Device or resource busy'' is displayed, some other processes using PM/Ethernet exist on the host.
• If the message ``pmOpenDevice: No route to host'' is displayed, please check hostname of the host. The hostname in pm-ethernet.conf must be same the node`s hostname, which is the results of hostname command or results gethostname() libc function.
• If the message ``pmOpenDevice: Invalid argument'' is displayed, please set environment variable ``PM_DEBUG'' to 1 (i.e. if you are using the Bourne shell, please execute ``PM_DEBUG=1; export PM_DEBUG'', or, if you are using csh, please execute ``setenv PM_DEBUG 1''), and run pmtest again.
  
• The node number specified using the -dest argument is different from the number used in the configuration file.

Communication Performance is too slow.

• Check whether auto-negotiation status between a switch and a NIC is good or not. If a switch status 100 Base/T half duplex and a NIC status 100 Base/T full duplex, the performance is very slow is spite of good TCP/IP communication.