You can determine Ethernet connectivity problems by examining Ethernet
statistics produced by the entstat command. Then, you can
debug the problems using the starttrace and stoptrace commands.
To help debug problems with Ethernet connectivity, follow these steps:
- Verify that the source client partition can ping another client
partition on the same system without going through the Virtual I/O Server. If this fails, the problem is likely in the client partition's
virtual Ethernet setup. If the ping is successful, proceed to the next step.
- Start a ping on the source partition to a destination machine so
that the packets are sent through the Virtual I/O Server. This ping will most likely fail. Proceed to the next step with
the ping test running.
- On the Virtual I/O Server,
type the following command:
entstat –all sea_adapter
where sea_adapter is
the name of your Shared Ethernet Adapter.
- Verify that the VLAN ID to which the partition belongs is associated
with the correct virtual adapter in the VLAN IDs section of the output. Examine
the ETHERNET STATISTICS for the virtual adapter for this
VLAN and verify that the packet counts under the Receive statistics column
are increasing.
This verifies that the packets are being received
by the Virtual I/O Server through
the correct adapter. If the packets are not being received, the problem might
be in the virtual adapter configuration. Verify the VLAN ID information for
the adapters using the Hardware Management Console (HMC).
- Examine the ETHERNET STATISTICS for the physical
adapter for this VLAN and verify that the packet counts under the Transmit
statistics column are increasing. This step verifies that
the packets are being sent out of the Virtual I/O Server.
- If this count is increasing, then the packets are going out of the
physical adapter. Continue to step 6.
- If this count is not increasing, then the packets are not going out
of the physical adapter, and to further debug the problem, you must begin
the system trace utility. Follow the instructions in step 9 to
collect a system trace, statistical information, and the configuration description.
Contact service and support if you need to debug the problem further. See Customer service, support, and troubleshooting
for information about service and support.
- Verify that the target system outside (on physical
side of Virtual I/O Server) is
receiving packets and sending out replies. If this is not
happening, either the wrong physical adapter is associated with the Shared Ethernet Adapter or the Ethernet switch
might not be configured correctly.
- Examine the ETHERNET STATISTICS for the physical
adapter for this VLAN and verify that the packet counts under the Receive
statistics column are increasing. This step verifies that the
ping replies are being received by the Virtual I/O Server. If this count is not increasing, the switch might not be configured
correctly.
- Examine the ETHERNET STATISTICS for the virtual
adapter for this VLAN and verify that the packet counts under the Transmit
statistics column are increasing. This step verifies
that the packet is being transmitted by the Virtual I/O Server through
the correct virtual adapter. If this count is not increasing,
start the system trace utility. Follow the instructions in step 9 to
collect a system trace, statistical information, and the configuration description.
Work with service and support to debug the problem further.
- Use the Virtual I/O Server trace
utility to debug connectivity problems. Start a system trace using
the starttrace command specifying the trace hook ID. The
trace hook ID for Shared Ethernet Adapter is
48F. Use the stoptrace command to stop the trace. Use the cattracerpt command
to read the trace log, format the trace entries, and write a report to standard
output.