IBM eserver Cluster 1350, Installation And Service Manual

Page: 55 / 122

If following the steps in Table 12 on page 38 did not correct the problem, continue
with the steps shown in “Clustering with one network.”
Clustering with one network
Use this section for more troubleshooting information about clustering with one
network.
Ping failure on one or some nodes
If one or more nodes experience a ping failure, this indicates a problem with the
node hardware or software. Complete the following steps to resolve the problem:
1. Telnet or use (shell script) SSH to connect to the node through the serial
console or KVM and make sure the node is operational.
a. If telnet or SSH succeeds, check the syslog for errors.
1) If there are errors, go to “Isolating software problems” on page 48 and
complete the steps in that table to resolve the problem.
2) If there are no errors, this indicates a network problem. See Table 13
and complete the steps in that table to resolve the problem.
b. If telnet or SSH fails, this indicates a node hardware problem. See “Isolating
hardware problems” on page 42 for problem resolution.
Ping failure on all nodes
If all nodes experience a ping failure, it indicates a problem on one of the following:
v
Network. Go to Table 13 and complete the listed actions to resolve the problem.
v
Network adapter on the management node
v
DHCP configuration
v
Network configuration
v
Cisco blade failure
Table 13. Network troubleshooting for a cluster with one network
Symptom Action
Cannot ping a node or nodes
on the cluster network from
the management node, yet the
rconsole command and
access from the KVM work
correctly.
1. At the console prompt, type the ifconfig command to
Make sure that the IP settings are correct.
2. Make sure that the cables are fully plugged into the
switch and node, and that everything else is plugged
into the correct port. See the cabling information printed
on each cable label and “VLAN options” on page 14 if
you are unsure where a cable belongs. Make sure that
the link LEDs are lit.
3. Swap ports on the Ethernet switch with a known
working cluster node port.
4. Make sure that the Ethernet switch port is configured
for the Management VLAN.
Clustering with two networks
Use the following troubleshooting information about clustering with two networks.
Ping failure on one or more nodes
If one or more nodes experience a ping failure, it indicates a problem with the node
hardware or software. Complete the following steps to resolve the problem:
v Telnet or use (shell script) SSH to connect to the suspect node via the serial
console or KVM and make sure that the node is operational. If telnet or SSH
succeeds, check the syslog for errors.
Chapter 8. Troubleshooting hardware and software problems
39