Create a Post
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Participant

Move External Interface to different NIC

Jump to solution

Gaia 3.10 R80.20 fail-over cluster using ClusterXL on Open Servers (Dell R740).  

I need to move the External interface from a 1G copper (eth2) NIC to a 10G SFP+ slot (eth0) utilizing a DAC cable.  Here is my plan so far:

Do during a maintenance window as there will be an outage:

Via CPUSE, Network Interfaces tab;  Start on Standby member delete IP address from eth2 and disable NIC.  Configure eth0 with settings & IP from eth2, disable auto-negotiation, click on link speed and select 10Gbps / Full and enable NIC.   Then do Active member (which will start my outage).  Disconnect copper cables from eth2 and connect DAC cables.

Via SmartConsole, edit FW Cluster object, select Network Management; copy all info from eth2, then delete the eth2 interface.  Then should I :

    a.)  Do a "get interfaces with topology"  OR

    b.)  Add a new interface and input the info from eth2 by hand. 

I'm thinking I should do the "get interfaces with topology" just worried it might change some settings or something with the other interfaces....

Then, save and push policy.  Or do I need to reboot the gateways first?  Since there isn't a discntd.if file anymore I don't think I need to reboot them....

 

PLEASE let me know if I'm missing something or if there is a better way to do this.

 

2 Solutions

Accepted Solutions
Highlighted
Advisor

When you say CPUSE I presume you simply meant the WebUI as not patching software.

Gaia config is Ok

 

What I would do with the Interfaces in the Cluster Object is simply rename the existing eth0 interfaces as eth2 across the two members and cluster section.

As is external interface and same IP then no need to actually delete and attempt to rediscover the new interfaces.

View solution in original post

0 Kudos
Highlighted

A quick and easy way:-)

Change the interface names in the following file and reboot the machine:

/etc/udev/rules.d/00-OS-XX.rules

Example for the contents of this file:

In R80.10 and above, the format for the udev rules has been changed to be based on PCI Bus ID instead of SYSFS{address} used in previous versions.

The following format must be used for the "/etc/udev/rules.d/00-OS-XX.rules" file:

ID=="0000:03:00.0", NAME="eth0"
ID=="0000:03:00.1", NAME="eth1"
ID=="0000:03:00.2", NAME="eth2"
ID=="0000:03:00.3", NAME="eth3"

Edit the file exactly as the example in sk69621

Attention!

If you update your OS, you may have to modify the file again.

View solution in original post

Tags (1)
8 Replies
Highlighted
Advisor

When you say CPUSE I presume you simply meant the WebUI as not patching software.

Gaia config is Ok

 

What I would do with the Interfaces in the Cluster Object is simply rename the existing eth0 interfaces as eth2 across the two members and cluster section.

As is external interface and same IP then no need to actually delete and attempt to rediscover the new interfaces.

View solution in original post

0 Kudos
Highlighted
Participant

Yes, WebUI...does CPUSE just refer to the Updates section of the WebUI?  Seems like I've seen these terms used intermittently for the same thing, so I guess I'm a bit confused on what the difference is...

So, I can just rename the interface as nothing in that entry is actually tied to NIC name?  It will just detect which NIC the IP addresses are configured on?  I didn't think I could just simply rename it....figured somehow the entry must be tied to the NIC name (eth0 or eth2).  I couldn't see how when I selected edit to look at the config, but was afraid there was a hidden connection to the actual NIC name somewhere, and figured I'd better deleted it and re-create it.  Sounds like you are saying that is not the case though and I can just rename it?  If you're sure of that....sweet!

0 Kudos
Highlighted
Participant
So if I go into SmartConsole and edit the Cluster Object via the Nework Management tab, edit the External interface and change the name from Eth2 to Eth0, then when I click on the Advanced tab within the edit interface window, will I have to change the "Interface Names" for both cluster members manually as well?
0 Kudos
Highlighted
Explorer

There is quite a detailed SK for interface changes: sk57100.

I once moved a whole bunch of VLANs from one bond on UTP interfaces to a new one on a 10Gbit interface card, and basically followed this SK.

0 Kudos
Highlighted

A quick and easy way:-)

Change the interface names in the following file and reboot the machine:

/etc/udev/rules.d/00-OS-XX.rules

Example for the contents of this file:

In R80.10 and above, the format for the udev rules has been changed to be based on PCI Bus ID instead of SYSFS{address} used in previous versions.

The following format must be used for the "/etc/udev/rules.d/00-OS-XX.rules" file:

ID=="0000:03:00.0", NAME="eth0"
ID=="0000:03:00.1", NAME="eth1"
ID=="0000:03:00.2", NAME="eth2"
ID=="0000:03:00.3", NAME="eth3"

Edit the file exactly as the example in sk69621

Attention!

If you update your OS, you may have to modify the file again.

View solution in original post

Tags (1)
Highlighted
Participant

nice solution

0 Kudos
Highlighted
Admin
Admin

If by CPUSE you mean WebUI, your plan is actionable. 

0 Kudos
Highlighted
Participant

Don't forget Proxy ARP settings, if any !!

0 Kudos