Oracle Fabric Interconnect F1-15
 
Oracle Fabric Interconnect F1-4
 
Oracle Virtual Compute Appliance X4-2 Hardware
 
Private Cloud Appliance X5-2 Hardware
 

PLA-Support>Sun Systems>SAND>Network>SN-SND: Oracle Virtual Networking
 

In this Document

Symptoms

Changes

Cause

Solution

Created from <SR 3-11513443891>

Applies to:

Symptoms

After XgOS install or upgrade or installing a new Gen2 Front Panel, the 'ethmgmt' network interface is not found.  Without the 'ethmgmt' network interface present, networking cannot be configured to connect the Fabric Interconnect to a network.  When looking at the Fabric Interconnect over serial console the errors show that eth1 is being renamed to eth4.  The errors are also found in /var/log/kern.log.  Here is the errors where the network devices are being renamed:

(snip below is also found also in /var/log/kern.log)

The issue here is eth1 gets renamed to eth4.  When XgOS expects to rename eth1 to ethmgmt, this is naturally going to fail.

In /etc/udev/rules.d/70-persistent-net.rules, can see the result of "udev: renamed network interface eth1 to eth4" where 00:13:97:00:a5:9a is associated to eth4 instead of eth1.

In /etc/udev/rules.d/70-persistent-net.rules, normally would expect to see only 3 lines for eth0 ~ eth2.  However, for some reason, there are 2 ADDITIONAL interfaces listed in /etc/udev/rules.d/70-persistent-net.rules, taking up eth1 and eth2, so this pushed the management interface to eth4.

The Gen2 FP comes with the COMX motherboard.

Based on the logic in /etc/init.d/xgifrename.sh, would expect the renaming to be:

eth0 -> ethiop
eth1 -> ethmgmt
eth2 -> ethha
eth3 -> ethfabric

XgOS is expecting the management interface to be eth1 before renaming it to ethmgmt, but, In this case, the management interface is eth4, so eth4 is left as it is.

eth0 -> ethiop           Link encap:Ethernet  HWaddr ec:9e:cd:13:42:58
eth1 -> eth4             Link encap:Ethernet  HWaddr 00:13:97:00:a5:9a
eth2 -> eth3 ->         ethfabric Link encap:Ethernet  HWaddr 00:13:97:00:a5:9b

Manually editing /etc/init.d/xgifrename.sh and changing all instances of eth4 back to eth1 works until the Fabric Interconnect is rebooted.

Also the manual changes made using 'ifconfig' to force to use eth4 revert to original whenever the Fabric Interconnect is rebooted.

For example to force ethmgmt to use eth4 and configure IP reverts when the Fabric Interconnect is rebooted - using ifconfig to manually plumb an IP address to eth4 is a workaround, but will be lost every time the Fabric Interconnect is rebooted.

 Reinstalling the XgOS from scratch also doesn't resolve the problem.

Changes

 XgOS install or upgrade/downgrade or brand new Gen2 Front Panel installation.

Cause

The 'forcebaseos' option forces a new fresh install of the Linux kernel in addition to the XgOS downgrade/upgrade.   It is possible if this option is needed, that there was a bad XgOS install at the factory, and the 'forcebaseos' option forces a new install of the Linux kernel.

Solution

Reinstall the XgOS using 'forcebaseos' option.  After the XgOS .xpf file is placed on the Fabric Interconnect as user root under /var/fsroot/admin, as admin user run this command:

This solution has no attachment