I tried the HA tempates for HEAT. I am using a Fedora 17 image for instance as for creating a heat stack. The fedora instance keeps on crashing, and as a result HEAT recreates another instance whenever the VM crashes. Interesingly, the crashed image is deleted by HEAT before a new instance is created. Is this expected behaviour? Is there any documentation which documents this behaviour?
My concern is, if the instance is automatically deleted, we don't get a chance to peek into the kernel and figure out what is causing the VM's OS to crash. Is there a way HEAT can be configured NOT to delete the instances?