First, the phenomenon
The current network of a host, today received downtime warning, logged in to see the host has not been down, but SSH was rejected. View the/var/log/message log and found a large number of DHCPRequest request logs. The specific contents are as follows:
[Root@datanode2 init.d]# tail-f/var/log/messages
June 1 16:51:09 Datanode2 Dhclient:bound to 169.254.95.120 – renewal in 280 seconds.
June 1 16:55:49 Datanode2 dhclient:dhcprequest on usb0 to 169.254.95.118 Port (xid=0x691bb32a)
June 1 16:55:49 datanode2 dhclient:dhcpack from 169.254.95.118 (xid=0x691bb32a)
June 1 16:55:49 Datanode2 Dhclient:bound to 169.254.95.120 – renewal in 257 seconds.
June 1 17:00:05 Datanode2 dhclient:dhcprequest on usb0 to 169.254.95.118 Port (xid=0x691bb32a)
June 1 17:00:06 datanode2 dhclient:dhcpack from 169.254.95.118 (xid=0x691bb32a)
June 1 17:00:06 Datanode2 Dhclient:bound to 169.254.95.120 – renewal in 239 seconds.
June 1 17:04:04 Datanode2 dhclient:dhcprequest on usb0 to 169.254.95.118 Port (xid=0x691bb32a)
June 1 17:04:05 datanode2 dhclient:dhcpack from 169.254.95.118 (xid=0x691bb32a)
June 1 17:04:05 Datanode2 Dhclient:bound to 169.254.95.120 – renewal in 263 seconds.
June 1 17:08:28 Datanode2 dhclient:dhcprequest on usb0 to 169.254.95.118 Port (xid=0x691bb32a)
June 1 17:08:28 datanode2 dhclient:dhcpack from 169.254.95.118 (xid=0x691bb32a)
June 1 17:08:28 Datanode2 Dhclient:bound to 169.254.95.120 – renewal in 258 seconds.
Second, processing
1, confirmed that the machine did not open DHCP service, and the use of the address is static configuration, nor the use of dhclient;
2, through the query in the Redhat knowledge base to see the relevant documents on the issue. View the environment on this document as described:
Red Hat Enterprise Linux 5
IBM HS22 Blade
IBM x3850 X5 [7145]
And my system environment is REDHAT6, the hardware environment is on the HS22 blade. The solution for this knowledge base is to view the license account. If you do not buy a license, you can also find the IBM hardware manufacturer to confirm the error.
Third, solve
From the above error, the error of the network card is usb0, and the host does not use the network card, but in the/etc/sysconfig/network-scripts/directory did find the IFCFG-USB0 network card, and the default configuration of the network card using a DHCP address. IBM-related engineers confirm that the function of the network card is:
IBM's PC server has a feature that uses USB as a management network (IBM is unique and not found in other brand servers). When you are not connected to the USB network card will continue to request IP, if not found DHCP will be the default allocation of a 169.254.XX.XX IP address, which in the Oracle RAC Environment, there may be a conflict with the virtual IP. Cause the routing information to be lost. Looking at this USB0 on the Internet can also cause Linux ha exceptions.
But fortunately, the host is not used in the RAC environment, since it is not used, turn off the network card from the start, you can solve it.
#ifdown usb0
Vi/etc/sysconfig/network-scripts/ifcfg-usb0
Onboot=no