cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted

5220 Appliance - Re-configuring eth0 port to change address

When configuring a 5220 (2.0.2 code level) today I came accross a couple of issues when configuring the network.

By default, I plugged the laptop into eth0, attached via browser to 192.168.1.1, and started the initial configuration.  My aim was to achieve the following:

eth0 - 192.168.71.4

eth2 & eth3 - bonded in failover mode - 192.168.72.1

When I tried to configure the bonded set I could only select all of the remaining nic's (eth1 thru eth5) from the dropdown.  I wanted to select just eth2 & eth3.  I went ahead and selected all of the remaining nic's and selected balance-alb for the bonding mode.

I then wanted to change the IP address on eth0.  I attached the laptop to eth1 (now on the 192.168.72.1 address) and went into an SSH session.  I could not see how to change the address on eth0.  I managed to unconfigure eth0, and it now has no IP address associated with it.  However, I canot see what command to use to enable this again with a specific IP address.......

At present the situation is that eth0 has no address (I want it configured on it's own with an IP address) and eth1 thru eth5 are bonded in balance-alb mode with a single IP address (and I want just eth2 &eth3 bonded in a failover mode with a single IP address).

I have a second appliance to configure in the morning, as well as this one to correct, so any help appreciated.

Thanks,

AJ

1 Solution

Accepted Solutions
Highlighted
Accepted Solution!

Resolved

OK - in order to achieve the desired result I did the following:

Attached to the Appliance on another interface (eth1) via the bonded IP address

LINKAGGREGATION DISABLE (needed to do this before I could unconfigure eth0 even though it was not bonded to anything)

IPV4 command to set the IP address on eth0

LINKAGGREGATION ENABLE eth2,eth3 balance-alb

GATEWAY SHOW (to check the default gateway is still correct)

GATEWAY ADD to add new gateway for eth0

Whilst running these commands the network was restarted several times, however........

I went on and installed the Java Admin Client on a PC and connected to one of the appliances, and then tried to connect.  It did connect but was very slow and was intermittent as to which functions I could select.  I noticed in the Java window that it was trying to resolve to the old IP address that was on eth0 - this no longer exists !!  As a long shot (as I thought that the network being restarted should have sorted this) I rebooted the Appliances (as I was having exactly the same problem on both).  This resolved the issue immediately - so I guess the old IP address was cached somewhere in the appliance.

Bottom line - problem solved and appliances seem to be operating OK at present as we start testing.

Thanks,

AJ. 

 

View solution in original post

4 Replies
Highlighted

Just out of curiosity, why

Just out of curiosity, why the big need to configure eth0 instead of eth1 for that network?

Either way, as long as you configure a different port to connect to, use PUTTY and login as admin/P@ssw0rd to get to the CLI shell (CLISH).  You can then use the network configuration to unconfigure the 192.168.1.1 and re-configure with the desired IP.  One thing I would suggest, if you're appliance is going to be living on multiple networks I would strongly suggest not using the web gui or the "Configure" command in the CLISH as both require the input of a default gateway.  If you're going to be configuring multiple different networks, I always suggest using the IPv4 command to configure an IP on the NIC and then use the "Gateway" command to configure the routing for the particular NICs, thereby avoiding having to configure a system wide default gateway that has gotten in my way when using multiple networks in the past.  

Once you have your networking set up, then go through the web gui to configure the rest of the appliance, just skip past the networking piece.

Highlighted

Hey Mate,   Even simpler

Hey Mate,

 

Even simpler method (and the efect will be the same as described by Chad). 

Log to the appliance using ssh (admin/P@ssw0rd). 

Do Support>Maintenance>elevate. Your privileges will be elevated to root. 

do:

 

cd /etc/sysconfig/network

do:

vi ifcfg-eth (you need to check what interfaces are already configured on your appliance). 

Modified the IP address. 

Restart the network:

 

/etc/init.d/network restart

 

In regards to other interfaces (creating additional bonds):

 

Verify the bond config:

 

vi ifcfg-bond0

 

Check which interfaces are configured as slaves.

 

Do:

 

lspci | grep -i intel

 

find the pci id. 

 

Use that information to configure additional bond interface ... like ifcfg-bond1.

 

Hope it helps

 

S

 

Highlighted

I generally encourage folks

I generally encourage folks to do what they can to avoid the root prompt unless you're working with support. One of the key reasons is that for many things the CLISH is referencing script files that in some cases do more than what the associated command in the OS would do.  A prime example is the "hostname" function in the CLISH.  In an unconfigured 52xx appliance it actually seeds the specified hostname for the system into several different locations for use in other parts of the installation process, whereas the hostname command from the root prompt would just set the hostname for the system without changing a lot of the security information that is built into the appliance.

I'm not saying that this is necessarily the case for the networking steps Sebastian outlined above, and I'm sure they would work perfectly well to achieve the desired result, but I always encourage all of my customers, even the most Linux-savvy ones, to do what they can from the CLISH and go to the root prompt only when they really have to and preferably with support folks on the line.

Obviously there are some limitations with the CLISH today, particularly as it relates to the running of NBU commands on the box and this has been recognized by the product management folks.  In a coming release expect to see the ability to run any NBU command via the CLISH through the use of a section of the CLISH dedicated to taking NBU commands and piping them to the OS.

Highlighted
Accepted Solution!

Resolved

OK - in order to achieve the desired result I did the following:

Attached to the Appliance on another interface (eth1) via the bonded IP address

LINKAGGREGATION DISABLE (needed to do this before I could unconfigure eth0 even though it was not bonded to anything)

IPV4 command to set the IP address on eth0

LINKAGGREGATION ENABLE eth2,eth3 balance-alb

GATEWAY SHOW (to check the default gateway is still correct)

GATEWAY ADD to add new gateway for eth0

Whilst running these commands the network was restarted several times, however........

I went on and installed the Java Admin Client on a PC and connected to one of the appliances, and then tried to connect.  It did connect but was very slow and was intermittent as to which functions I could select.  I noticed in the Java window that it was trying to resolve to the old IP address that was on eth0 - this no longer exists !!  As a long shot (as I thought that the network being restarted should have sorted this) I rebooted the Appliances (as I was having exactly the same problem on both).  This resolved the issue immediately - so I guess the old IP address was cached somewhere in the appliance.

Bottom line - problem solved and appliances seem to be operating OK at present as we start testing.

Thanks,

AJ. 

 

View solution in original post