Solaris Troubleshooting Jumpstart – “Timeout waiting for ARP/RARP packet”
first identify the MAC address of the interface used to boot the client for the network installation. You can also use below procedures to identify the MAC address of the client for the server configuration files.
1. Identification Client MAC address from the Server’s Configration files:
# getent hosts
# getent ether
2. Identification of Server’s Interface that is receiving rarp reuests from the client
lo0: flags=2001000849 mtu 8232 index 1
inet 127.0.0.1 netmask ff000000
ce0: flags=201000843 mtu 1500 index 2
inet xx.yy.1.239 netmask ffffff00 broadcast xx.yy.1.255
ce1: flags=201000843 mtu 1500 index 3
inet xx.yy.2.239 netmask ffffff00 broadcast xx.yy.2.255
ce2: flags=201000843 mtu 1500 index 4
inet xx.yy.3.239 netmask ffffff00 broadcast xx.yy.3.255
in this example, ce0 is the interface that should receive rarp requests from the clients, because the ip address and the servers interface both are configured for same IP segement i.e. xx.yy.1.0
3. run the snoop command to check the traffic related to client installation
# snoop -d ce0 0.bh.2c.15.3c
output should show something like below:
Using device /dev/ce (promiscuous mode)
OLD-BROADCAST -> (broadcast) RARP C Who is 0.bh.2c.15.3c
server-1 -> client-1 RARP R 0:3:ba:15:2c:3c is xx.yy.1.101, CLIENT1
If no output observed , it is possible that there the is a connectivity problem between the server and client.
if the output observed but the MAC address is different than what we have configured in “ethers” file, then it is possible that the client is configured to use a different interface, or ethernet address to what is expected.
follow the link “Identify MAC of boot interface” to identify which network interface used to boot on a sparc system, and go forward with below troubleshooting:
4. Ensure that Jumpstart / Boot Server can be contacted using the same cable and port the client is using.
Many times the problem comes with VLAN setup done by network team, if the client connection from the network switch side configured to different VLAN ( virtual IP segment) comapared to boot server, that will block the client requests to reach the boot server.
This should be simple to check at the switch itself. If a VLAN configuration is in place, please verify the VLAN configuration on the network switch to confirm both the client and server are on the same subnet
5. Ensure that the switch port that the client is attached to is set to AUTO NEGOTIATE, and that the port is enabled.
This is most import and single most common cause of network issues in big datacenters. Normally with hardcoded switch port speed, we will have problems to establish connection from client network interface, so during the installation just keep the setting to”auto negotiate” and after the installation just hardcode the speed as per the requirement