I just learned the hard way that you can't run multiple Eucalyptus cloud controllers on the same network.
I'd assumed that as long as each cloud had a different private address pool and physically separate cluster networks that it would work.
After about a day, the primary cloud starts detecting the secondary cloud and things went a bit haywire. The primary cloud controller crashed but left the instances still running.
Rebooting things didn't get much better either - the nodes just didn't work right. Not quite sure what the deal was but nodes would show available, but all instances would just stick at "starting" and never fully reach a running state. Checking "euca_conf --show-nodes" resulted in no instances showing associated with any nodes.
I had to power down the new test cloud and de-register it's cluster controller from the primary cloud (which magically registered itself) and a reboot or two later was able to get my primary cloud back up and running.
To clarify - I wasn't trying to run multiple clusters, but wanted a completely separate cloud with it's own clusters. For future reference: do it on a separate subnet.


I've often stated that Linux is no harder to use than Windows for the average Joe. I think now I've proved my point.



I decided it was time to place a PXE boot install image on my network for installing Ubuntu 10.04 Lucid desktops. For the most part everything went as expected (