Comment 4 for bug 454405

To see the effect of this problem, the simplest setup is a multi-cluster installation (single clc, walrus, and two clusters (cc/sc) each with one node (nc)). Choose a security group (default is fine) and run one instance in one cluster, and one instance in the other cluster (this is controlled at run time using the '-z' option to 'euca-run-instances'). Both instances should appear to have the same private address, since the index value (which starts at 0) should have been used to choose the allocated private IP for both instances.

After the fix, the expected (correct) behavior should be that private IPs are uniquely assigned to VMs; two running VMs should never have the same private IP.