Run as root on the EMS.
This test uses xdsh and uses "lpar" for the node group. Use a different node group if desired.
Note xdsh can leave behind processes if you do a <CNTRL-C> in the middle of a command.
To gaurd against results which are messed up by old copies of the ping test, kill
any leftover ping processes.
1) Create the ping_address_file. This is only required if changes have been made since the last
run; e.g., octants deconfigured, ARP entries changed, etc.
"./create_ping_address_file [interface]"
If the "interface" argument is not provided, it defaults to "ml0" as the interface to test.
Remove any nodes you don't want to run the ping test.
2) Copy "ping_address_file" to /tmp on each node:
xdcp lpar -v ping_address_file /tmp
3) Copy "ping_all" to /tmp on each node:
xdcp lpar -v ping_all /tmp
4) Remove any output files from previous runs:
xdsh lpar -v "rm /tmp/ping.all.out 2>/dev/null"
5) Run the ping test:
"./start_ping_all"
Each node in the test will write any ping failures to "/tmp/ping.all.out" (local /tmp on each node).
(No "ping.all.out" file indicates no ping failures.)
6) To determine if we have ping failures:
xdsh lpar -v "ls -l /tmp/ping.all.out 2>/dev/null" | sort -n +5
If a given set of "ping.all.out" files are the same length, it's likely they will have
duplicate ping failures.
