Overview

There are numerous IP addresses listed against Lattus Controller Nodes, with the management Controller Nodes having the most IP addresses. This is because almost all of the IP addresses have the actual node IP address as well as a Virtual IP (VIP) that will hit whichever IP address is being used by the Management Controller Node. To better understand Lattus Controller Node IP addresses, we'll review the Controller Nodes IP address information for an actual Lattus customer.

Customer Example: QVC

This CMC's Controller Node page seemingly displays a mass of IP addresses: 
 

 
The key is on the Network Configuration page, which tells us what each subnet is doing:
 
 
 
There are two 10 GbE ingest subnets: 10.15.21.0 and 10.15.20.0. Note their "Public" status:
 
 
 
Then we have the customer management subnet (10.16.16.0) used for GUI access and Phone Home (Event Reports), and controller failover (though this doesn't work, so stay clear). Note its "Public" status:
 

 
Then we have the private 1 and 2 subnets (10.15.22.0 and 10.15.23.0) used for controllers and nodes to talk to each other. Note that Private1 is management AND storage, where Private2 is just storage. On a node Private1 translates to Eth0 and Private2 to Eth1. All node status and management is only handled over Eth0, which is why a node can show "halted" (i.e. a ping to Eth0 fails), but data can still be flowing over Eth1.
 
 
 
Note we also have a DHCP LAN Private1 subnet (10.15.22.0). This, as far as I know is used for Controller Nodes, and as a temporary IP assignment when new nodes are added, before they are initialized.
 
 

This leaves only IP. ///TS: Can you clarify the previous sentence?/// This is the IPMI subnet range (10.15.24.0) used for handing out IPMI DHCP IP addresses to single site (non Multi-Geo) Storage Nodes. It looks like there is an entry for the subnet, then an additional one for DHCP over the subnet--both of which have the same subnet. The name displays here as "S10 IPMI" because initially there were only S10s. Now there are S10s, S20s and S30s:
 
 
 
We can now start to translate the list of IP addresses. It's easier to start with the non-Management Controller Nodes. We know the subnets for Private 1 and Private 2 are 10.15.22.0 and 10.15.23.0:
 
 
 
You will notice that the Management Controller Node has two 10.15.22.0 and 10.15.23.0 subnet IP addresses, the higher one in each case is the VIP. Also the other IP addresses will follow in order across the Controllers Nodes: C01 10.15.22.1, through C05 10.15.22.5, so these are the fixed IP addresses for the Controller Nodes. The other two 10.15.22.214 and 10.15.23.254 are the two VIPs. If we put this altogether, we now have:
 
 
 
From the Networks table and the node listings, we know the IPMI IPs addresses are on the 10.15.24.0 subnet:
 
 
 
(Typically on the Controller Nodes, the lowest subnet number is Private 1, then Private 2, then IPMI: Private 1 10.15.22.0, Private 2 10.15.23.0, IPMI 10.15.24.0). 

So we can now start identifying the IPMI info:
  
 

Now to fill in the CMC network used for sending out the 'Phone Home' 10.16.16.0:
 
 
 

 

Now to fill out the two 10 GbE ingest ports on 10.15.20.0 and 10.15.21.0:
 

 
 
 
Finally as the GUI service is running only on the Management Controller Node 01, we can reach it either via the CMC VIP or the actually CMC IP address. Also if the customer can see the 10Gb paths that go to StorNext form his workstation, then he could use those IP addresses as well: