Currently, if a new VM cannot be started (e.g. no host can accommodate the
requested flavor), the ec2 backend does simply recognize the error and
terminate the VM. The next iteration will simply retry to spawn a new instance
with the result that the cloud stack will be constantly stressed with failing
VMs and requested to create new ones.
the EC2 backend should be able to recognize recurring failures when starting
VMs and, for example, reject the next request to spawn a new one.
S.
Original issue reported on code.google.com by sergio.m...@gmail.com on 11 Feb 2014 at 5:41
Original issue reported on code.google.com by
sergio.m...@gmail.com
on 11 Feb 2014 at 5:41